PASEO Group / People / André Jalobeanu

FILTER: Display only PASEO group-related publications

Show link boxes

Journal papers and Book Chapters – peer-reviewed

D.D. Fitzenz, A. Jalobeanu, M. Ferry: “A Bayesian Framework to Rank and Combine Candidate Recurrence Models for Specific Faults” - Bulletin of the Seismological Society of America (BSSA), in press, Nov 2011
We propose a probabilistic framework in which different types of information pertaining to the recurrence of large earthquakes on a fault can be combined, in order to constrain the parameter space of candidate recurrence models and provide the best combination of models knowing the chosen data set and priors.
We use Bayesian inference for parameter and error estimation, graphical models (Bayesian networks) for modeling, and stochastic modeling to link cumulative offsets to co-seismic slip. The COBBRA method (Fitzenz et al., 2010) was initially developed to use cumulative offset data to further constrain and discriminate between recurrence models built from historical and archaeological catalogs of large earthquakes. We discuss this method and present an extension of it that incorporates trench data. For our case study (Jordan Valley Fault) the relative evidence of each model slightly favor the Brownian Passage Time and the lognormal models.
We emphasize that 1) the time-variability of fault slip rate is critical to constrain recurrence models, 2) the shape of the probability density functions of paleoseismic events is very important, in most cases not Gaussian, and should be reported in its complexity; 3) renewal models are in terms of intervals between consecutive earthquakes, not dates, and the algorithms should account for that fact; 4) maximum likelihood methods are inadequate for parameter uncertainty evaluation and model combination or ranking. Finally, more work is needed to define proper priors, and to model the relationship between cumulative slip and co-seismic slip, in particular when the fault behavior is more complex.
@article{ref109,
}
M. Petremand, A. Jalobeanu, C. Collet: “Optimal Bayesian Fusion of Large Hyperspectral Astronomical Observations” - Statistical Methodology (STAMET), Special issue on Astrostatistics, 9(1-2), Apr 2011
New generation integral-field spectrographs (IFS) such as MUSE will soon start observing distant astronomical objects with much higher spectral and spatial resolutions than today’s instruments. The new hyperspectral observations will represent huge amount of scientific data (up to 1.2 GB per each MUSE raw acquisition) whose analysis requires the development of dedicated processing methods. In addition, in the framework of long acquisition sessions (typically 80 hours for deep-field observations acquired with the MUSE instrument) spread over several nights and split into one- hour exposures, the same field of view is observed under varying atmospherical and physical conditions (e.g. seeing, noise, spectral and spatial shifts. . . ) thus producing multiple hyperspectral cubes to analyze. In this paper, we propose a new data fusion method that aims at reconstructing a single data cube from this large CCD (raw) data set (up to 100 GB for a MUSE acquisition session) taking into account acquisition parameter variations and whose analysis becomes easier for astronomers. This last point clearly means that we need to produce a fused hyperspectral image together with associated uncertainties on each fused pixel. On the one hand, a forward model precisely describes the complex MUSE acquisition process and, on the other hand, the inverse problem is solved to yield the fused cube from the whole observation set. The reconstruction and the fusion of large raw observations is performed in a sequential way so as to minimize computing time and memory usage. This challenging task is performed in the rigorous Bayesian framework, which provides a fusion scheme that is more optimal (in the statistical sense) than existing 3D reconstruction methods used in astronomy while giving access to uncertainties (in the form of a precision matrix) associated to the fused image: results can then be reused by astronomers for further analysis. The global fusion scheme is validated on small-size simulated observations with varying acquisition conditions while real MUSE raw observations will not be available before 2013.
@article{ref101,
}
D.D. Fitzenz, M. Ferry, A. Jalobeanu: “Long-term slip history discriminates among occurrence models for seismic hazard assessment” - Geophysical Research Letters (GRL), 37(L20307), Oct 2010
Today, the probabilistic seismic hazard assessment (PSHA) community relies on stochastic models to compute occurrence probabilities for large earthquakes. Considerable efforts have been devoted to extracting information from long catalogs of large earthquakes based on instrumental, historical, archeological and paleoseismological data. However, the models remain only and insufficiently constrained by these rare single-slip event data. Therefore, the selection of the models and their respective weights necessarily involves ruling by a panel of experts. Since cumulative slip data with high temporal and spatial resolution are now available, we propose a new approach to incorporate these pieces of evidence of mid- to long-term fault behavior into PSHA: the Cumulative Offset-Based Bayesian Recurrence Analysis (COBBRA). For the Dead Sea Fault, our method provides weights to the competing recurrence and rupture models, allows time-independent models to be ruled out, and provides a means to compute the cumulative probability of occurrence for the next full-segment event reflecting all available data.
@article{ref96,
}
L. Blanc-Féraud, L. Mugnier, A. Jalobeanu: “Blind Image Deconvolution” - in Inverse Problems in Vision and 3D Tomography (ISTE/Wiley), A. Mohammad-Djafari ed., John Wiley and Sons, Dec 2009
@inbook{ref85,
}
C. Collet, F. Flitti, S. Bricq, A. Jalobeanu: “Fusion and Multi-Modality” - in Inverse Problems in Vision and 3D Tomography (ISTE/Wiley), A. Mohammad-Djafari ed., John Wiley and Sons, Dec 2009
@inbook{ref86,
}
M.V. Joshi, A. Jalobeanu: “MAP estimation for Multiresolution Fusion in Remotely Sensed Images using an IGMRF Prior Model” - IEEE Trans. on Geoscience and Remote Sensing (TGRS), 48(3), Jul 2009
In this paper we propose a model based approach for the multiresolution fusion of satellite images. Given the high spatial resolution panchromatic (Pan) image and a low spatial and high spectral resolution multispectral (MS) image acquired over the same geographical area the problem is to generate a high spatial and high spectral resolution multispectral image. This is clearly an ill-posed problem and hence we need a proper regularization. We model each of the low spatial resolution MS images as the aliased and noisy versions of their corresponding high spatial resolution i.e., fused (to be estimated) MS images. A proper aliasing matrix is assumed to take care of the undersampling process. The high spatial resolution MS images to be estimated are then modeled as separate Inhomogeneous Gaussian Markov Random Fields (IGMRF) and a Maximum A Posteriori (MAP) estimation is used to obtain the fused image for each of the MS bands. The IGMRF parameters are estimated from the available high resolution Pan image and are used in the prior model for regularization purposes. Since the method does not directly operate on the Pan pixel values as most of the other methods do, the spectral distortion is minimum and the spatial properties are better preserved in the fused image as the IGMRF parameters are learned at every pixel. We demonstrate the effectiveness of our approach over some existing methods by conducting the experiments on synthetic data as well as on the images captured by the Quickbird satellite.
@article{ref75,
}
A. Jalobeanu, J.A. Gutiérrez, E. Slezak: “Multisource data fusion and super-resolution from astronomical images” - Statistical Methodology (STAMET), Special issue on Astrostatistics, 5(4), Jul 2008
Virtual Observatories give us access to huge amounts of image data that are often redundant. Our goal is to take advantage of this redundancy by combining images of the same field of view into a single object. To achieve this goal, we propose to develop a multi-source data fusion method that relies on probability and band-limited signal theory. The target object is an image to be inferred from a number of blurred and noisy sources, possibly from different sensors under various conditions (i.e. resolution, shift, orientation, blur, noise...). We aim at the recovery of a compound model "image+uncertainties" that best relates to the observations and contains a maximum of useful information from the initial data set. Thus, in some cases, spatial super-resolution may be required in order to preserve the information. We propose to use a Bayesian inference scheme to invert a forward model, which describes the image formation process for each observation, and takes into account some a priori knowledge (e.g. stars as point sources). This involves both automatic registration and resampling, which are ill-posed inverse problems that are addressed within a rigorous Bayesian framework. The originality of the work is in devising a new technique of multi-image data fusion that provides us with super-resolution, self-calibration and possibly model selection capabilities. This approach should outperform existing methods such as resample-and-add or drizzling since it can handle different instrument characteristics for each input image and compute uncertainty estimates as well. Moreover, it is designed to also work in a recursive way, so that the model can be updated when new data becomes available.
@article{ref69,
}
D.D. Fitzenz, A. Jalobeanu, S.H. Hickman: “Integrating Laboratory Creep Compaction Data With Numerical Fault Models: a Bayesian Framework” - Journal of Geophysical Research (JGR), AGU, 112(B08410), Aug 2007
We developed a robust Bayesian inversion scheme to plan and analyze laboratory creep compaction experiments. We chose a simple creep law that features the main parameters of interest when trying to identify rate-controlling mechanisms from experimental data. By integrating the chosen creep law or an approximation thereof, one can use all the data, either simultaneously or in overlapping subsets, thus making more complete use of the experiment data and propagating statistical variations in the data through to the final rate constants. Despite the non-linearity of the problem, with this technique one can retrieve accurate estimates of both the stress exponent and the activation energy, even when the porosity time series data are noisy. Whereas adding observation points and/or experiments reduces the uncertainty on all parameters, enlarging the range of temperature or effective stress significantly reduces the covariance between stress exponent and activation energy. We apply this methodology to hydrothermal creep compaction data on quartz to obtain a quantitative, semiempirical law for fault zone compaction in the interseismic period. Incorporating this law into a simple direct rupture model, we find marginal distributions of the time to failure that are robust with respect to errors in the initial fault zone porosity.
@article{ref68,
}
A. Jalobeanu, J. Zerubia, L. Blanc-Féraud: “Bayesian estimation of blur and noise in remote sensing imaging” - in Blind image deconvolution: theory and applications (CRC), P. Campisi and K. Egiazarian ed., Taylor & Francis / CRC Press, May 2007
We propose a Bayesian approach to infer the parameters of both blur and noise in remote sensing images. The modulation transfer function (MTF) of the imaging system, including atmosphere, optics and pixel-level sampling, is modeled by a parametric function with a small number of parameters. The noise is assumed to be white, additive and Gaussian. Both blur and noise processes are supposed to be stationary. To constrain this ill-posed inverse problem, the unknown scene is modeled by a scale-invariant stochastic process governed by a fractal exponent and a global energy term. The main novelty consists of treating all parameters as random variables whose mean is estimated within a fully Bayesian framework. The chosen approach can be summarized as the computation of the mean posterior marginal related to useful parameters only. This requires integrating the joint probability density function (PDF) with respect to all the nuisance parameters, which is achieved through Laplace approximations.
In this chapter we present two approaches; the former is straightforward, and the latter leads to a more efficient, simplified and optimized estimation algorithm. In addition, we investigate methods of uncertainty estimation and model assessment, in order to validate our approach on real images and to propose further improvements.
@inbook{ref48,
}
A.R. Hajian, S.M. Movit, D. Trofimov, B. Balick, Y. Terzian, K.H. Knuth, D. Granquist-Fraser, K. Huyser, A. Jalobeanu, D. McIntosh, A.E. Jaskot, S. Palen, N. Panagia: “An Atlas of [N II] and [O III] Images and Spectra of Planetary Nebulae” - Astrophysical Journal Supplement (APJSup), University of Chicago Press, 169, Nov 2006
As part of a multi-epoch observing program designed to acquire precise expansion distances to planetary nebulae, we have completed a longslit, spectroscopic survey of selected ob jects. Our overall strategy is to deduce the distance to each nebula by measuring the angular expansion rate of the nebular shell from multiepoch narrowband imagery, and the nebular expansion velocity from spatially resolved narrowband spectra. However, the nebular geometry must be properly considered in order to infer a distance from an observed tangential motion and radial velocity. For spherical nebulae, vexp is easy to determine from the observed spectra. For even slightly more complex ellipsoidal nebulae, uncertainties in the nebular morphology and kinematics can dominate the uncertainties in the resulting distance. In this paper, we present an atlas of Hubble Space Telescope images and groundbased, longslit, narrowband spectra centered on the 6584A line of [N II] and the 5007A line of [O III]. Almost all of the images were obtained by us for this pro ject as part of GO 8390 and 8773 (duplicated in GO 7501). The spectra were obtained for a variety of slit positions across each target (as shown on the images) in an effort to account for non-spherical nebular geometries in a robust manner. We have extended the Prolate Ellipsoidal Shell model of Aaquist and Kwok (1996) and Zhang and Kwok (1998) to generate synthetic images as wel l as longslit spectra. Using this model, we have derived basic parameters for the subsample of PNe which present ellipsoidal appearances and regular kinematic patterns. We find differences between our parameters for the target PNe as compared to those of Zhang and Kwok (1998), which we attribute to increased spatial resolution for our image data and the inclusion of kinematic data in the model fits. The data and analysis presented in this paper can be combined with detections of nebular angular expansion rates to determine precise distances to the PN targets.
@article{ref52,
}
A. Jalobeanu, L. Blanc-Féraud, J. Zerubia: “An adaptive Gaussian model for satellite image deblurring” - IEEE Trans. on Image Processing (TIP), IEEE SPS, 13(1), Jan 2004
The deconvolution of blurred and noisy satellite images is an ill-posed inverse problem, which can be regularized within a Bayesian context by using an a priori model of the reconstructed solution. Since real satellite data show spatially variant characteristics, we propose here to use an inhomogeneous model. We use the Maximum Likelihood Estimator (MLE) to estimate its parameters and we show that the MLE computed on the corrupted image is not suitable for image deconvolution, because it is not robust to noise. Then we show that the estimation is correct only if it is made from the original image. Since this image is unknown, we need to compute
an approximation of sufficiently good quality to provide useful estimation results.
Such an approximation is provided by a wavelet-based deconvolution algorithm. Thus, a hybrid method is first used to estimate the space-variant parameters from this image and then to compute the regularized solution. The obtained results on high resolution satellite images simultaneously exhibit sharp edges, correctly restored textures and a high SNR in homogeneous areas, since the proposed technique adapts to the local characteristics of the data.
@article{ref4,
}
A. Jalobeanu, L. Blanc-Féraud, J. Zerubia: “Satellite image deblurring using complex wavelet packets” - International Journal of Computer Vision (IJCV), Kluwer, 51(3), Feb 2003
The deconvolution of blurred and noisy satellite images is an ill-posed inverse problem. Direct inversion leads to unacceptable noise amplification. Usually, the problem is either regularized during the inversion process, or the noise is filtered after deconvolution and decomposition in the wavelet transform domain. Herein, we have developed the second solution, by thresholding the coefficients of a new complex wavelet packet transform; the thresholding functions are automatically estimated. The use of complex wavelet packets enables translational invariance and improves directional selectivity, while remaining of complexity O(N). The results obtained exhibit both correctly restored textures and a high SNR in homogeneous areas. Compared to previous algorithms, the proposed method is faster, rotationally invariant and better takes into account the directions of the details and textures of the image, improving restoration. The images deconvolved in this way can be used as they are (the restoration step proposed here can be inserted directly in the acquisition chain), and they can also provide a starting point for an adaptive regularization method, enabling one to obtain sharper edges.
@article{ref5,
}
A. Jalobeanu, L. Blanc-Féraud, J. Zerubia: “Hyperparameter estimation for satellite image restoration using a MCMC Maximum Likelihood method” - Pattern Recognition (PR), Elsevier, 35(2), Feb 2002
The satellite image deconvolution problem is ill-posed and must be regularized. Herein, we use an edge-preserving regularization model using a phi function, involving two hyperparameters. Our goal is to estimate the optimal parameters in order to automatically reconstruct images. We propose to use the maximum-likelihood estimator (MLE), applied to the observed image. We need sampling from prior and posterior distributions. Since the convolution prevents use of standard samplers, we have developed a modified Geman?Yang algorithm, using an auxiliary variable and a cosine transform. We present a Markov chain Monte Carlo maximum-likelihood (MCMCML) technique which is able to simultaneously achieve the estimation and the reconstruction.
@article{ref8,
}
J. Zerubia, A. Jalobeanu, Z. Kato: “Markov Random Fields in Image Processing. Application to Remote Sensing and Astrophysics” - in New avenues for astronomical data analysis (J. Phys. IV France), A. Bijaoui and J.P. Rozelot ed., EDP Sciences, Jan 2002
Early vision directly deals with raw pixel data involving image compression, restoration, edge detection, segmentation, texture analysis, motion detection, optical flow, etc. Most of these problems can be formulated within a general framework, called image labeling, where we associate a label to each pixel from a finite set. The meaning of this label depends on the problem that we are trying to solve. For image restoration, it means a grey-level; for edge detection, it means the presence or the direction of an edge; for image segmentation, it means a class (or region); etc. The problem here is how to choose a label for a pixel, which is optimal in a certain sense.
Our approach is probabilistic: at each pixel, we want to select the most likely labeling. To achieve this goal, we need to define some probability measure on the set of all possible labelings. In real scenes, neighboring pixels have usually similar intensities; edges are smooth and often straight. In a probabilistic framework, such regularities are well expressed by Markov Random Fields (MRF). Another reason for dealing with MRF models is of course the Hammersley-Clifford theorem which allows to define MRF through clique-potentials. In the labeling problem, this leads to the following Bayesian formulation: we are looking for the Maximum A Posteriori (MAP) estimate of the label field yielding to the minimization of a usually non-convex energy function. [...]
@inbook{ref9,
}

Conference papers – peer-reviewed

A. Jalobeanu, G. Gonçalves: “On the boresight calibration of Airborne LiDAR Systems” - submitted to SilviLaser 2012: First Return (SilviLaser'12), Vancouver, Canada, May 2012
Within the AutoProbaDTM project, we plan to develop fast and fully automated techniques to derive topographic maps and error maps, from full-waveform airborne LiDAR data. A probabilistic approach is used in order to modelling surfaces and data acquisition, solving inverse problems and handling uncertainty. Bayesian inference provides a rigorous framework for unsupervised reconstruction of the DEM and error propagation from the data to the end result, treating all quantities as random variables. Automatic sensor calibration plays a major role in this project. In fact, the overall positional accuracy and uncertainty obtained from the LiDAR technology depends on the assembly and calibration of the three system components: the GPS (Global Positioning System), the INS (Inertial Navigation System) and the laser-scanner device.
In this paper we evaluate some of the principal boresight calibration methods and we propose a novel method based on the Bayesian inference to address this problem as well. Because this method don't use planar surfaces it is well suitable for forest areas with poor geometric planar features such as building roofs. The first contribution is to use not only the 3D points extracted from the raw waveforms but their uncertainty as well, and to apply a probabilistic surface matching with spatially variable point accuracy in order to obtain the attitude corrections. The second contribution consists of using all the flight lines, where most methods only use the calibration cross. In this way we can also estimate the attitude drift of the sensor platform and correct the LiDAR data for temporal attitude variations. Finally, we use the probabilistic framework for error propagation and we propose a probability distribution of the calibrated boresight angles.

More information about the project: http://sites.google.com/site/autoprobadtm
@inproceedings{ref113,
}
A. Jalobeanu, G. Gonçalves: “The full-waveform LiDAR Riegl LMS-Q680i: from reverse engineering to sensor modeling” - American Society of Photogrammetry and Remote Sensing Annual Conference (ASPRS'12), Sacramento, CA, USA, Mar 2012
The development of new data processing methods requires access to the raw data. Unfortunately some LiDAR manufacturers do not provide information about the format and the users can only rely on proprietary software to do their processing. Even if using black boxes might be sufficient for some simple applications, it might be an impedi- ment to scientific research, as the processing would be limited to state of the art methods and their current imple- mentation. In existing full-waveform LiDAR software there is a lack of error propagation methods that might be an issue when making quantitative measurements of topography, reflectance, or vegetation parameters. This problem can only be addressed at the lowest level by working directly on the waveforms. Moreover, to improve range meas- urement and feature extraction techniques, and compute error bars correctly, one also needs an instrument model de- scribing the data acquisition process.
Here we focus on the Riegl LMS-Q680i airborne LiDAR sensor. We acquired 200 km2 of data (nearly 100 GB of undocumented binary files). We performed a reverse engineering to understand how the timestamps, look angles and waveforms were stored. Then we developed a model of this particular sensor: the two nonlinear detector chan- nels, the asymmetric amplifier impulse response and the ringing effect. Assuming this model, not only were we able to match the output of the proprietary software, but we also managed to compute the range uncertainty, and we opened the way to new methodologies to improve the reliability and the accuracy of echo extraction.
@inproceedings{ref107,
}
A. Jalobeanu: “Predicting spatial uncertainties in stereo photogrammetry: achievements and intrinsic limitations” - 7th International Symposium on Spatial Data Quality (ISSDQ 2011), Coimbra, Portugal, Oct 2011
We present a new probabilistic method for digital surface model generation from optical stereo pairs, with an expected ability to propagate errors from the data to the final result, providing spatial uncertainty estimates to be used for quantitative analyis in planetary or Earth sciences. Existing stereo-derived surfaces lack rigorous, quantitative error estimates, and we propose to address this issue by deriving a method of error prediction, rather than error assessment as usually done in the area through the use of reference data. We use only the information present in the available data and perform the prediction using Bayesian inference. We start by defining a forward model, using an adaptive radiometric change map to achieve robustness to noise and reflectance effects. A priori smoothness constraints are introduced to stabilize the solution.!Solving the inverse problem to recover a surface from noisy data involves fast deterministic optimization techniques.!Though the reconstruction results look satisfactory, we conclude that uncertainty estimates computed from two images only are unreliable, which is due to major limitations of stereo, such as non-Lambertian reflectance and incorrect spatial sampling, which violate our underlying assumptions and cause biases that cannot be accounted for in the predicted error budget.
@inproceedings{ref102,
}
M. Petremand, C. Collet, A. Jalobeanu: “Fusion bayésienne d’images hyperspectrales astronomiques” - XXIIIe Colloque GRETSI (GRETSI'11), Bordeaux, France, Sep 2011
The hyperspectral astronomical imagery gives the ability to study spectrally and spatially resolved objects. In the case of the next integral-field spectrograph (IFS) MUSE (Multi Unit Spectroscopic Explorer), each survey will produce large hyperspectral raw observations (1.2 GB of sensor data each), centered on the same sky area. The joint analysis of observations at different times remains problematic because of varying acquisition conditions (atmosphere, shifts, etc.). In this paper, we present a method aiming at fusing a set of several raw observations with respect to varying acquisition parameters. We propose to use the Bayesian framework which ensures a rigorous approach while yielding uncertainties associated to each fused pixel. This new algorithm is validated on small-size simulated data cubes, and its extension to large scale data is under process.
@inproceedings{ref108,
}
A. Jalobeanu, C. Gama, J.A. Gonçalves: “Probabilistic surface change detection and measurement from digital aerial stereo images” - IEEE International Geoscience & Remote Sensing Symposium (IGARSS'10), Honolulu, Hawaii, USA, Jul 2010
We propose a new method to measure changes in terrain topography from two optical stereo image pairs acquired at different dates. The main novelty is in the ability of computing the spatial distribution of uncertainty, thanks to stochastic modeling and probabilistic inference. Thus, scientists will have access to quantitative error estimates of local surface variation, so they can check the statistical significance of elevation changes, and make, where changes have occurred, consistent measurements of volume or shape evolution. The main application area is geomorphology, as the method can help study phenomena such as coastal cliff erosion, sand dune displacement and various transport mechanisms through the computation of volume changes. It can also help measure vegetation growth, and virtually any kind of evolution of the surface.
We first start by inferring a dense disparity map from two images, assuming a known viewing geometry. The images are accurately rectified in order to constrain the deformation on one of the axes, so we only have to infer a one-dimensional parameter field. The probabilistic approach provides a rigorous framework for parameter estimation and error computation, so all the disparities are described as random variables. We define a generative model for both images given all model variables. It mainly consists of warping the scene using B-Splines, and defining a spatially adaptive stochastic model of the radiometric differences between the two views. The inversion, which is an ill-posed inverse problem, requires regularization, achieved through a smoothness prior model.
Bayesian inference allows us to recover disparities as probability distributions. This is done on each stereo pair, then disparity maps are transformed into surface models in a common ground frame in order to perform the comparison. We apply this technique to high resolution digital aerial images of the Portuguese coast to detect cliff erosion and quantify the effects of weathering.
@inproceedings{ref88,
}
A. Jalobeanu: “Predictive Spatial Accuracy of Digital Elevation Models Generated from Stereo Pairs” - Ninth International Symposium on Spatial Accuracy Assessment in Natural Resources and Environmental Sciences (Accuracy'10), Leicester, UK, Jul 2010
A new method for reconstructing digital elevation models (DEM) from optical stereo pairs is proposed. The main originality is the ability to propagate errors from the observed data to the final result, providing all the spatial accuracy estimates required for the use of topography in planetary or Earth science applications. In general, stereo-derived DEMs lack quantitative error estimates. This can be a major issue when the result is used to derive physical measurements in areas such as hydrology or geomorphology. We aim at performing error prediction, rather than error assessment as usually done in the community through the use of reference data sets. Indeed, the goal is to use only the information present in the available data to predict the errors, since we believe it is the only way to build relevant spatially adaptive accuracy maps. Existing techniques usually provide only a global accuracy measure after a validation procedure, or at best propose to predict the local behavior of accuracy from morphological indicators, failing to capture the dependence upon the image content. We think that predictive accuracy computation shall replace the error assessment step, thus allowing for a fully automated DEM generation with relevant error maps. A Bayesian approach is used, which provides a rigorous way of estimating uncertainties and various parameters. We start by defining a forward model, consisting of warping the observed scene through a disparity map and assuming a spatially adaptive radiometric change map to achieve robustness to noise and reflectance effects. An a priori smoothness prior model is introduced in order to stabilize the solution. Solving the inverse problem to recover the disparity map from noisy measurements requires to optimize an energy function. We employ fast deterministic techniques to recover an posteriori probability density function (pdf) of the disparity map. Finally, the disparities are converted into a DEM through a geometric camera model. Combining disparity and camera calibration errors allows for a comprehensive error propagation from the input to the final DEM.
@inproceedings{ref95,
}
A. Jalobeanu, M. Petremand, C. Collet: “Bayesian fusion of hyperspectral astronomical images” - Proc. of 30th workshop on Bayesian Inference and Maximum Entropy methods (MaxEnt'10), Chamonix, France, Jul 2010
The new integral-field spectrograph MUSE will acquire hyperspectral images of the deep sky, requiring huge amounts of raw data to be processed, posing a challenge to modern algorithms and technologies. In order to achieve the required sensitivity to observe very faint objects, many observations need to be reconstructed and co-added into a single data cube. In this paper, we propose a new fusion method to combine all raw observations while removing most of the instrumental and observational artifacts such as blur or cosmic rays. Thus, the results can be accurately and consistently analyzed by astronomers. We use a Bayesian framework allowing for optimal data fusion and uncertainty estimation. The knowledge of the instrument allows to write the direct problem (data acquisition on the detector matrix) and then to invert it through Bayesian inference, assuming a smoothness prior for the data cube to be reconstructed. Compared to existing methods, the originality of the new technique is in the propagation of errors throughout the fusion pipeline and the ability to deal with various acquisition parameters for each input image. For this paper, we focus on small-size, simulated astronomical observations with varying parameters to validate the image formation model, the reconstruction algorithm and the predicted uncertainties.
@inproceedings{ref87,
}
M. Petremand, C. Collet, A. Jalobeanu, V. Mazet, F. Salzenstein, M. Louys: “New bayesian fusion scheme and visualization tool for astronomical hyperspectral data cubes” - Astronomical Data Analysis VI (ADA), Monastir, Tunisia, May 2010
The goal is to perform data cube fusion of hyperspectral images obtained with MUSE which provides a hundred of images when deep fields are observed. We have to maximize not only the signal-to-noise ratio but also the spatial and spectral resolution while handling with varying PSF (each of the 100 images are obtained in different conditions). A similar problem concerns the combination of different locations to cover a very large field from multiple poses. The approach consists in combining all (or almost all) the information into a single model. In this case, the model is a 2D image having an arbitrary resolution and number of spectral bands. In practice, this depends on the amount of available data, thus leading to super-resolved results when many input images are used. We propose to use a Bayesian inference method, based on the inversion of a forward model which describes the image formation process for each observation. The originality of the work is in devising and implementing new methods to perform multi-image data fusion through Bayesian inference, with the ability to address both spectral and spatial superresolution and to perform model selection in order to automatically determine the resolution needed. Also, this involves to handle with the huge size of the data (first simulated then obtained from real observations from 2012), which can be done using parallelization and a recursive code where the cubes of data are added the ones after the others. The Bayesian framework provides a natural framework for the modeling of optimal fusion scheme and the propagation of the errors.
@inproceedings{ref93,
}
A. Jalobeanu: “Spatial Accuracy Assessment of Digital Elevation Models: A Probabilistic Approach” - American Society for Photogrammetry and Remote Sensing annual conference (ASPRS'09), Baltimore, MD, USA, Mar 2009
We propose a new method for the measurement of high resolution topography from an optical stereo pair. The main contribution is the ability to propagate errors from the imperfect observed data to the final result, providing all accuracy estimates required for the use of topography in planetary or Earth science applications. Indeed, digital elevation models (DEM) computed from images using state of the art methods usually lack quantitative error estimates. This can be a major issue when the result is used to measure actual physical parameters, such as slope or terrain roughness.
Thus, we propose a new algorithm to infer a dense bidimensional disparity map from two images, that also estimates the spatial distribution of errors. We use a probabilistic approach, which provides a rigorous way of estimating parameters and uncertainties. All the parameters are defined as random variables within a Bayesian framework. We start by building a forward model, which consists of warping the observed scene using B-Splines and using a spatially adaptive radiometric change map for robustness purposes. An a priori smoothness model is introduced in order to stabilize the solution. Solving the inverse problem to recover the disparity map requires to optimize a global non-convex energy function, which is difficult task. A deterministic optimization based on a multi-grid strategy, followed by a local energy analysis at the optimum, allows to recover the a posteriori probability density function (pdf) of the disparity, which encodes both the optimal solution and the related error map.
Finally, the disparity field is converted into a DEM through a geometric camera model. This model is either known initially, or calibrated using the estimated disparity map and extra data (existing low-resolution DEM or ground control points). Automatic calibration from uncertain disparity and topographic data allows for a comprehensive error propagation from the input data to the final elevation model.
@inproceedings{ref84,
}
M.V. Joshi, A. Jalobeanu: “A MAP estimation for Multiresolution Fusion in Remotely Sensed Images using an IGMRF Prior Model” - IEEE International Geoscience & Remote Sensing Symposium (IGARSS'08), Boston MA, USA, Jul 2008
In this paper we propose a model based approach for multi-resolution fusion of satellite images. Given the high spatial resolution panchromatic (Pan) image and a low spatial and high spectral resolution multi-spectral (MS) image acquired over the same geographical area, the problem is to generate a high spatial and high spectral resolution multi-spectral image. This is clearly an ill-posed problem, which requires a proper regularization. We model each of the low spatial resolution MS images as the aliased and noisy versions of their corresponding high spatial resolution images. A decimation (aliasing) matrix is estimated for each of the MS bands by using the available Pan and the MS image. The high spatial resolution MS images to be estimated are then modeled as separate Inhomogeneous Gaussian Markov Random Fields (IGMRFs) and the Maximum A Posteriori (MAP) estimation is used to obtain the fused images. The required IGMRF parameters representing the spatial correlation among high resolution MS pixels are estimated from the available high resolution Pan image and are used in the prior model during the regularization. Since the method does not directly operate on the Pan pixel values as most of the other methods do, the spectral distortion is minimum and the spatial properties are better preserved in the fused image as the IGMRF parameters are learnt at every pixel. We demonstrate the effectiveness of our approach by conducting experiments on synthetic data as well as on real images captured by the Quickbird satellite.
@inproceedings{ref81,
}
A. Jalobeanu, D.D. Fitzenz: “Inferring deformation fields from multidate satellite images” - IEEE International Geoscience & Remote Sensing Symposium (IGARSS'08), Boston MA, USA, Jun 2008
We focus on a geophysical application of image processing: the measurement of high resolution ground deformation from two optical satellite images taken at different dates. Disparity maps estimated from image pairs usually lack quantitative error estimates. This is a major issue for measuring physical parameters, such as ground deformation or topography variations. Thus, we propose a new method to infer the disparity map. We adopt a probabilistic approach, treating all parameters as random variables, which provides a rigorous framework for parameter estimation and uncertainty evaluation. We start by defining a generative model of the data given all model variables. This forward model consists of warping the scene using B-Splines and applying a spatially adaptive radiometric change map. Then we use Bayesian inference to invert and recover the a posteriori probability density function (pdf) of the disparity map. The method is validated on multidate SPOT 5 imagery related to the Bam earthquake (Iran), showing results compatible with INSAR measurements.
@inproceedings{ref80,
}
A. Jalobeanu, D.D. Fitzenz: “Robust disparity maps with uncertainties for 3D surface reconstruction or ground motion inference” - ISPRS Proc. of Photogrammetric Image Analysis (PIA'07), Munich, Germany, Sep 2007
Disparity maps estimated using computer vision-derived algorithms usually lack quantitative error estimates. This can be a major issue when the result is used to measure reliable physical parameters, such as topography for instance. Thus, we developed a new method to infer the dense disparity map from two images. We use a probabilistic approach in order to compute uncertainties as well. Within this framework, parameters are described in terms of random variables. We start by defining a generative model for both raw observed images given all model variables, including disparities. The forward model mainly consists of warping the scene using B-Splines and adding a radiometric change map. Then we use Bayesian inference to invert and recover the a posteriori probability density function (pdf) of the disparity map.
The main contributions are: The design of an efficient fractal model to take into account radiometric changes between images; A multigrid processing so as to speed up the optimization process; The use of raw data instead of orthorectified imagery; Efficient approximation schemes to integrate out unwanted parameters and compute uncertainties on the result. Three applications could benefit from this disparity inference method: DEM generation from a stereo pair (along or across track), automatic calibration of pushbroom cameras, and ground deformation estimation from two images at different dates.
@inproceedings{ref71,
}
A. Jalobeanu, J.A. Gutiérrez: “Inverse covariance simplification for efficient uncertainty management” - Proc. of 26th workshop on Bayesian Inference and Maximum Entropy methods (MaxEnt'07), Saratoga Springs, NY, USA, Jul 2007
When it comes to manipulating uncertain knowledge such as noisy observations of physical quantities, one may ask how to do it in a simple way. Processing corrupted signals or images always propagates the uncertainties from the data to the final results, whether these errors are explicitly computed or not. When such error estimates are provided, it is crucial to handle them in such a way that their interpretation, or their use in subsequent processing steps, remain user-friendly and computationally tractable. A few authors follow a Bayesian approach and provide uncertainties as an inverse covariance matrix. Despite its apparent sparsity, this matrix contains many small terms that carry little information. Methods have been developed to select the most significant entries, through the use of information-theoretic tools for instance. One has to find a Gaussian pdf that is close enough to the posterior pdf, and with a small number of non-zero coefficients in the inverse covariance matrix. We propose to restrict the search space to Markovian models (where only neighbors can interact), well-suited to signals or images. The originality of our approach is in conserving the covariances between neighbors while setting to zero the entries of the inverse covariance matrix for all other variables. This fully constrains the solution, and the computation is performed via a fast, alternate minimization scheme involving quadratic forms. The Markovian structure advantageously reduces the complexity of Bayesian updating (where the simplified pdf is used as a prior). Moreover, uncertainties exhibit the same temporal or spatial structure as the data.
@inproceedings{ref70,
}
A. Jalobeanu, E. Slezak, J.A. Gutiérrez: “Multisource data fusion and super-resolution from astronomical images” - Astronomical Data Analysis IV (ADA IV), Marseille, France, Sep 2006
Virtual Observatories give us access to huge amounts of image data that are often redundant. Our goal is to take advantage of this redundancy by combining images of the same field of view into a single object. To achieve this goal, we propose to develop a multi-source data fusion method that relies on probability and band-limited signal theory. The target object is an image to be inferred from a number of blurred and noisy sources, possibly from different sensors under various conditions (i.e. resolution, shift, orientation, blur, noise...). We aim at the recovery of a compound model "image+uncertainties" that best relates to the observations and contains a maximum of useful information from the initial data set. Thus, in some cases, spatial super-resolution may be required in order to preserve the information. We propose to use a Bayesian inference scheme to invert a forward model, which describes the image formation process for each observation, and takes into account some a priori knowledge (e.g. stars as point sources). This involves both automatic registration and resampling, which are ill-posed inverse problems that are addressed within a rigorous Bayesian framework. The originality of the work is in devising a new technique of multi-image data fusion that provides us with super-resolution, self-calibration and possibly model selection capabilities. This approach should outperform existing methods such as resample-and-add or drizzling since it can handle different instrument characteristics for each input image and compute uncertainty estimates as well. Moreover, it is designed to also work in a recursive way, so that the model can be updated when new data becomes available.
@inproceedings{ref60,
}
A. Jalobeanu, J.A. Gutiérrez: “Multisource data fusion for bandlimited signals: a Bayesian perspective” - Proc. of 25th workshop on Bayesian Inference and Maximum Entropy methods (MaxEnt'06), Paris, France, Aug 2006
We consider data fusion as the reconstruction of a single model from multiple data sources. The model is to be inferred from a number of blurred and noisy observations, possibly from different sensors under various conditions. It is all about recovering a compound object, signal+uncertainties, that best relates to the observations and contains all the useful information from the initial data set.
We wish to provide a flexible framework for bandlimited signal reconstruction from multiple data. In this paper, we focus on a general approach involving forward modeling (prior model, data acquisition) and Bayesian inference. The proposed method is valid for n-D objects (signals, images or volumes) with multidimensional spatial elements. For the sake of clarity, both formalism and test results will be shown in 1D for single band signals. The main originality lies in seeking an object with a prescribed bandwidth, hence our choice of a B-Spline representation. This ensures an optimal sampling in both signal and frequency spaces, and allows for a shift invariant processing.
The model resolution, the geometric distortions, the blur and the regularity of the sampling grid can be arbitrary for each sensor. The method is designed to handle realistic Gauss+Poisson noise.
We obtained promising results in reconstructing a super-resolved signal from two blurred and noisy shifted observations, using a Gaussian Markov chain as a prior. Practical applications are under development within the SpaceFusion project. For instance, in astronomical imaging, we aim at a sharp, well-sampled, noise-free and possibly super-resolved image. Virtual Observatories could benefit from such a way to combine large numbers of multispectral images from various sources. In planetary imaging or remote sensing, a 3D image formation model is needed; nevertheless, this can be addressed within the same framework.
@inproceedings{ref58,
}
A. Jalobeanu: “Multisource data fusion and super-resolution from astronomical images” - Statistical Challenges in Modern Astronomy IV (SCMA'IV), Penn State, PA, USA, Jun 2006
The goal is to combine multiple astronomical images of the same field of view into a single model, within the Virtual Observatory framework where the huge amounts of data often exhibit some redundancy. To achieve this goal, we propose to develop a multi-source data fusion method using probability theory. We want to infer an image from several blurred and noisy observations, possibly from different sensors and instruments under various conditions. We aim at the recovery of a compound object "image+uncertainties" that contains a maximum of useful information from the initial data set. In some cases, conserving information may require achieving super-resolution.
We propose to use a Bayesian inference scheme to invert a generative model that explains the image formation for each observation while taking into account a priori knowledge. Understanding the image formation process is crucial.
The originality of the work is in devising a new technique of multi-image data fusion that also addresses spatial super-resolution and recursive model updating. This involves both automatic registration and resampling, which are difficult inverse problems that are treated within a probabilistic framework. Our contribution outperforms state of the art methods in astronomy since it can handle different instrument characteristics for each input and provide uncertainty estimates as well.
@inproceedings{ref59,
}
D.D. Fitzenz, A. Jalobeanu, S.H. Hickman, N.H. Sleep: “Integrating Laboratory Compaction Data With Numerical Fault Models: a Bayesian Framework” - Proc. of 25th workshop on Bayesian Inference and Maximum Entropy methods (MaxEnt'05), San Jose, CA, USA, Aug 2005
When analyzing rock deformation experimental data, one deals with both uncertainty and complexity. Though each part of the problem might be simple, the relationships between them can form a complex system. This often leads to partial or only qualitative data analyses from the experimental rock mechanics community, which limits the impact of these studies in other communities (e.g., modelling). However, it is a perfect case study for graphical models.
We present here a Bayesian framework that can be used both to infer the parameters of a constitutive model from rock compaction data, and to simulate porosity reduction within direct fault models from a known (e.g. lab-derived) constitutive relationship, while keeping track of all the uncertainties. This latter step is crucial if we are to go toward process-based seismic hazard assessment. Indeed, the rate of effective stress build-up (namely due to fault compaction) as well as the recovery of fault strength determine how long it will take for different parts of the previously ruptured fault to reach failure again, thus controlling both the timing and the size of the next rupture. But deterministic models need to rigorously incorporate uncertainties it they are to be useful in creating probabilistic assessments of seismic hazard. It is therefore important to work within a framework able to assess model validity as well as use data uncertainties.
Our approach involves a hierarchical inference scheme using several steps of marginalization. Existing experimental data are rarely adequate to completely define a single constitutive relationship for given physical fault material parameters over temperature and effective confining pressures of relevance to actual fault zones. We therefore focus on one rather general, though experimentally derived, compaction law to illustrate how applying the proposed inference scheme on simulated data can help design compaction experiments to provide better constraints on creep parameters.
@inproceedings{ref3,
}
A. Jalobeanu: “Bayesian Vision for Shape Recovery” - Proc. of 24th workshop on Bayesian Inference and Maximum Entropy methods (MaxEnt'04), Garching-Munich, Germany, Jul 2004
We present a new Bayesian vision technique that aims at recovering a shape from two or more noisy observations taken under similar lighting conditions. The shape is parametrized by a piecewise linear height field, textured by a piecewise linear irradiance field, and we assume Gaussian Markovian priors for both shape vertices and irradiance variables. The modeled observation process, equivalent to rendering, is modeled by a non-affine projection (e.g. perspective projection) followed by a convolution with a piecewise linear point spread function, and contamination by additive Gaussian noise. We assume that the observation parameters are calibrated beforehand.
The major novelty of the proposed method consists of marginalizing out the irradiances considered as nuisance parameters, which is achieved by a hierarchy of approximations. This reduces the inference to minimizing an energy that only depends on the shape vertices, and therefore allows an efficient Iterated Conditional Mode (ICM) optimization scheme to be implemented. A Gaussian approximation of the posterior shape density is computed, thus providing estimates of both the geometry and its uncertainty. We illustrate the effectiveness of the new method by shape reconstruction results in a 2D case. A 3D version is currently under development and aims at recovering a surface from multiple images, reconstructing the topography by marginalizing out both albedo and shading.
@inproceedings{ref2,
}
A. Jalobeanu, F.O. Kuehnel, J.C. Stutz: “Modeling Images of Natural 3D Surfaces: Overview and Potential Applications” - Proc. of IEEE conf. on Computer Vision and Pattern Recognition, Graphical Model-Based Vision workshop (CVPR'04), Washington DC, USA, Jul 2004
Generative models of natural images have long been used in computer vision. However, since they only describe the statistics of 2D scenes, they fail to capture all the properties of the underlying 3D world. Even though such models are sufficient for many vision tasks, a 3D scene model is needed when it comes to inferring a 3D object or its characteristics. In this paper, we present such a generative model, incorporating both a multiscale surface prior model for surface geometry and reflectance, and an image formation process model based on realistic rendering, that accounts for the physics of image generation. We focus on the computation of the posterior model parameter densities, and on the critical aspects of the rendering. We also discuss how to efficiently invert the model within a Bayesian framework. We present a few potential applications, such as asteroid modeling and planetary topography recovery, illustrated by promising results on real images.
@inproceedings{ref1,
}
A. Jalobeanu, L. Blanc-Féraud, J. Zerubia: “Natural image modeling using complex wavelets” - Proc. of SPIE, Wavelets X (SPIE-5207), San Diego, CA, USA, Aug 2003
We propose to model satellite and aerial images using a probabilistic approach. We show how the properties of these images, such as scale invariance, rotational invariance and spatial adaptivity lead to a new general model which aims to describe a broad range of natural images. The complex wavelet transform initially proposed by Kingsbury is a simple way of taking into account all these characteristics. We build a statistical model around this transform, by defining an adaptive Gaussian model with interscale dependencies, global parameters, and hyperpriors controlling the behavior of these parameters.
This model has been successfully applied to denoising and deconvolution, for real images and simulations provided by the French Space Agency.
@inproceedings{ref6,
}
A. Jalobeanu: “Fractal 3-D modeling of asteroids using wavelets on arbitrary meshes” - 1st Symp. on Interdisciplinary Approaches in Fractal Analysis (IAFA'03), Bucharest, Romania, May 2003
In this work, we study the 3D geometry of the small bodies in our Solar System in order to derive a probabilistic model of such objects. Images taken by various spacecrafts seem to exhibit a fractal behaviour, which we propose to investigate by using a multiscale approach. The idea is to look for a scale-invariant model that could simply describe the statistics of the asteroid surfaces. In order to access the different scales, we need either a Fourier or a Wavelet transform that could be applied to the triangular mesh defining the object to analyze. Since the former transform could not be easily constructed on meshes (because of their irregularity), we use a wavelet transform instead. This analysis tool is designed to capture both scaling and spatial information on the object. The main novelty w.r.t. existing wavelet transforms on meshes consists of providing a local estimate of the scale. This way, we show that the suspected fractal properties are actually an efficient modeling tool, and we build a statistical model of asteroids. A possible application of this model is the dense 3D reconstruction from multiple images, which is an ill-posed inverse problem. Using the fractal approach as a prior model within a Bayesian framework should enable us to get an accurate estimate of the asteroid shape.
@inproceedings{ref7,
}
A. Jalobeanu, R.D. Nowak, J. Zerubia, M. Figueiredo: “Satellite and aerial image deconvolution using an EM method with complex wavelets” - IEEE International Conference on Image Processing (ICIP'02), Rochester, NY, USA, Oct 2002
In this paper, we present a new deconvolution method, able to deal with noninvertible blurring functions. To avoid noise amplification, a prior model of the image to be reconstructed is used within a Bayesian framework. We use a spatially adaptive prior, defined with a complex wavelet transform in order to preserve shift invariance and to better restore variously oriented features. The unknown image is estimated by an EM technique, whose E step is a Landweber update iteration, and the M step consists of denoising the image, which is achieved by wavelet coefficient thresholding. The new algorithm has been applied to high resolution satellite and aerial data, showing better performance than existing techniques when the blurring process is not invertible, like motion blur for instance.
@inproceedings{ref73,
}
A. Jalobeanu, L. Blanc-Féraud, J. Zerubia: “Estimation of blur and noise parameters in remote sensing” - Proc. of Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP'02), Orlando, FLA, USA, May 2002
In this paper we propose a new algorithm to estimate the parameters of the noise related to the sensor and the impulse response of the optical system, from a blurred and noisy satellite or aerial image. The noise is supposed to be white, Gaussian and stationary. The blur kernel has a parametric form and is modeled in such a way as to take into account the physics of the system (the atmosphere, the optics and the sensor). The observed scene is described by a fractal model, taking into account the scale invariance properties of natural images.
The estimation is performed automatically by maximizing a marginalized likelihood, which is achieved by a deterministic algorithm whose complexity is limited to O(N), where N is the number of pixels.
@inproceedings{ref10,
}
A. Jalobeanu, N.G. Kingsbury, J. Zerubia: “Image deconvolution using Hidden Markov modeling of Complex Wavelet Packets” - IEEE International Conference on Image Processing (ICIP'01), Thessaloniki, Greece, Oct 2001
In this paper, we propose to use a hidden Markov tree modeling of the complex wavelet packet transform, to capture the inter-scale dependencies of natural images. First, the observed image, blurred and noisy, is deconvolved without regularization. Then its transform is denoised within a Bayesian framework using the proposed model, whose parameters are estimated by an EM technique. The total complexity of this new deblurring algorithm remains O(N).
@inproceedings{ref36,
}
A. Jalobeanu, L. Blanc-Féraud, J. Zerubia: “Estimation des paramètres instrumentaux en imagerie satellitaire et aérienne” - 17th GRETSI Symposium on Signal and Image Processing (GRETSI'01), Toulouse, France, Sep 2001
In this paper, a new method is proposed, enabling us to estimate the parameters of the noise of the sensor and the impulse response of the optical system, from a blurred and noisy satellite or aerial image. The blurring kernel is parametrized, and modeled taking into account the physics of the sensor; the natural scene is described by a fractal model. The estimation is performed automatically, by maximizing the joint likelihood, which is achieved by a deterministic algorithm.
@inproceedings{ref35,
}
A. Jalobeanu, L. Blanc-Féraud, J. Zerubia: “Estimation rapide du paramètre de régularisation en déconvolution d'images” - Congrès francophone de vision par ordinateur (ORASIS'01), Cahors, France, Jun 2001
La déconvolution des images satellitaires floues et bruitées est un problème inverse mal posé, qui peut être régularisé dans un cadre bayésien par l'utilisation d'un modèle a priori de la solution reconstruite. Des modèles basés sur une fonctionnelle de régularisation non quadratique ont été utilisés avec succès afin de restaurer des images exemptes de bruit tout en préservant les contours. Toutefois, ces modèles présentent des paramètres, dont la valeur influe très fortement sur la qualité de la solution obtenue. Nous avions déjà proposé une technique d'estimation de ces paramètres, toujours dans un cadre bayésien, qui donne de bons résultats mais qui nécessite un temps de calcul important, car il s'agit d'une méthode stochastique. Dans cet article, nous proposons une nouvelle méthode d'estimation du paramètre de régularisation, fondée sur une approximation gaussienne. Cette technique présente l'avantage d'être déterministe. De cette manière, l'estimation est rendue particulièrement rapide, quelle que soit la taille de l'image que l'on cherche à déconvoluer, car elle ne nécessite qu'une FFT et quelques opérations par pixel. L'estimateur que nous avons utilisé est le maximum de vraisemblance (MV) en données complètes. La technique proposée consiste à approcher les distributions a priori et a posteriori par une loi gaussienne, ce qui rend les fonctions de normalisation relatives à ces lois calculables analytiquement. Le paramètre estimé de cette manière correspond à une régularisation quadratique (qui ne préserve pas les contours), il est donc réajusté en conséquence afin de permettre l'utilisation d'une fonctionnelle non quadratique lors de la déconvolution. Les images déconvoluées de cette manière peuvent être utilisées telles quelles, lorsque la dégradation n'est pas trop importante. Elles peuvent également servir à l'estimation des paramètres adaptatifs dans un algorithme travaillant dans une base d'ondelettes, car elles présentent des bords francs et un bruit résiduel suffisamment faible.
@inproceedings{ref34,
}
A. Jalobeanu, L. Blanc-Féraud, J. Zerubia: “Estimation of adaptive parameters for satellite image deconvolution” - International Conference on Pattern Recognition (ICPR'00), Barcelona, Spain, Sep 2000
The deconvolution of blurred and noisy satellite images is an ill-posed inverse problem, which can be regularized within a Bayesian context by using an a priori model of the reconstructed solution. Since real satellite data show spatially variant characteristics, we propose to use an inhomogeneous model. We use the Maximum Likelihood Estimator (MLE) to estimate its parameters. We demonstrate that the MLE computed on the corrupted image is not suitable for image deconvolution, because it is not robust to noise. Then we show that the estimation is correct only if it is made from the original image. As this image is unknown, we need to compute an approximation of sufficiently good quality to provide useful estimation results. Such an approximation is provided by a wavelet-based deconvolution algorithm. Thus, an hybrid method is first used to estimate the space-variant parameters from this image and second to compute the regularized solution. The obtained results on high resolution satellite images simultaneously exhibit sharp edges, correctly restored textures and a high SNR in homogeneous areas, since the proposed technique adapts to the local characteristics of the data.
@inproceedings{ref54,
}
A. Jalobeanu, L. Blanc-Féraud, J. Zerubia: “Satellite image deconvolution using complex wavelet packets” - International Conference on Image Processing (ICIP'00), Vancouver, Canada, Sep 2000
The deconvolution of blurred and noisy satellite images is an ill-posed inverse problem. Donoho has proposed to deconvolve the image without regularization and to denoise the result in a wavelet basis by thresholding the transformed coefficients.
We have developed a new filtering method, consisting of using a complex wavelet packet basis. Herein, the thresholding functions associated to the proposed method are automatically estimated. The estimation is performed within a Bayesian framework, by modeling the subbands using Generalized Gaussian distributions, and by applying the Maximum A Posteriori (MAP) estimator on each coefficient.
Compared to real wavelet-packet-based algorithms, the proposed method is shift invariant, provides good directionality properties and remains of complexity O(N).
@inproceedings{ref53,
}

Abstracts, Posters, Preprints, Reports and Theses

A. Jalobeanu, C. Gama, H. Almeida: “DSM generation from stereo aerial images for the reconstruction of the sea-cliff retreat pattern controlled by gullying process, Costa da Galé and Melides sectors (Southwest of Portugal)” - 32th International Geographical Congress, Cologne, Germany, Aug 2012
The seacliffs evolution is an important aspect to be taken in account in the evolution of the world coastline. The seacliffs can suffer erosion induced by the storm wave incidence or subaerial erosion leading to the retreat of the coastline. However the amount of sediments that come from the cliff retreat represent an important sediment source to the coastal system. In some cases it is essential to include this volume in the sediment budget balance of the studied coastal area.
Many methods have been developed to monitor the evolution of seacliffs, most of them are supported by field measurements. In these work you propose the application of a new stereo photogrammetric method to reconstruct the cliff topography producing digital surface model (DSM) revealing the spatial distribution of the elevation errors. The model results are complemented by the acquisition of field data (GCP-ground control points) obtained using the DGPS (Differential Global Positioning System). This method also allows the generation of a coarse Digital elevation model (DEM) of the bottom of the seacliffs.
The field study was conducted considering two small stretches of the sandy embayed coastline between Tróia and Sines (Southwest of Portugal). In these sectors the backshore of the subaerial beach is limited landward by the presence of seacliffs that suffer subaerial erosion (gullying process). The seacliffs presents poorly consolidated sediments (sand, clay, granule and fine pebbles) that suffer subaerial erosion showing complex gully morphology between the top and the bottom of the cliff. The sediments eroded by this process are stored at the base of cliffs in the form of debris fans. During storm periods the subaerial beach significantly decreases its width and the sediments contained in debris fans suffers cut-off. The sediments are transported by the waves thereby entering in the coastal system.
Two data series of digital aerial images at 20 cm resolution, acquired in 2008 and 2009, were used to reconstruct cliffs digital surface models (DSM) and monitor the evolution of the complex gully system. A data set of 50 GCP was used to constrain the sensor location and orientation. The method was able to detect the presence of main areas of cliff displacement although the sensitivity of camera calibration prevented the absolute estimation of the displacement rate. New field surveys should help improve the results.
@misc{ref112,
}
A. Jalobeanu, G. Gonçalves: “Probabilistic topographic maps from waveform LiDAR data: first results and application to geomorphology” - IUGG Conference on Mathematical Geophysics, Edimburgh, UK, Jun 2012
The AutoProbaDTM project focuses on the development of new algorithms for digital elevation model (DEM) generation, using the latest full-waveform airborne LiDAR technology. This involves methodological development, software design, and validation over a 200 km2 test area in central Portugal. One of the biggest challenges is to develop efficient ways to process huge volumes of raw data without compromising the accuracy and the physical consistency of the result. In the long run, we wish to develop efficient LiDAR data processing techniques for high-accuracy and large-scale mapping.
Currently in the LiDAR industry it is not possible to deliver a quantitative accuracy map along with the reconstructed DEM. Standard quality control only allows to check the difference between the DEM and a set of reference points, lacking the ability to predict the accuracy related to elevations at arbitrary locations. Traditional surveys are currently essential for sensor calibration and quality control of LiDAR data. Although some degree of automation has been achieved for boresight alignment, the full automation of the external quality control is still an issue. We plan to provide fast automated techniques to derive elevation models and compute accuracy maps simultaneously, based on a probabilistic approach to modeling surfaces and data acquisition. Indeed, Bayesian inference provides a rigorous framework for error propagation, and allows us to fuse all sources of information optimally during the DEM generation process. In the future, the produced accuracy maps shall help geoscientists put error bars on DEM-derived physical quantities.
In June 2011, 200 km2 of data were acquired (100 GB of binary files representing half a billion waveforms), over a study area located in Portugal near Arraiolos (Alentejo region). We used a Riegl LMS-Q680i sensor flown by the French company IMAO. A total area of 140 km2 was scanned at a satisfactory spatial sampling rate, the ground spacing nearly equal to the footprint size (point density from 2.7 to 4 pt/m2, 50 cm footprint). We believe this is crucial for a correct processing as aliasing artifacts are significantly reduced, compared to common practice where the spacing is usually much larger than the footprint size. The scanner was flown at 1500 m above ground to minimize the cost per km2 (with an additional 3000 m test flight well beyond the manufacturer's specifications). Therefore the instrument was operated in the low signal to noise ratio regime, and the limits of state of the art processing methods are reached, hence the need for algorithmic development to achieve both a higher detection rate and an improved robustness.
We present preliminary results from the first stage of the project: a large gridded DEM (1 m post spacing) for the entire study area, including an accuracy map, i.e. an error estimate for each point. The area was mainly chosen for its geomorphological interest. Many fault traces are already known in this region. Neotectonic studies have already been conducted using an existing 10 m DEM (from digitized topographic maps) of questionable quality, investigating knickpoints in small rivers. The 1 m DEM brings unprecedented detail and sheds a new light on these studies, providing not only new information but also a confidence measure.
It is unclear whether the tectonic structure is still active, and the high resolution DEM shall help identify new fault scarps and riverbed displacements, possibly hidden under the canopy. Future work on bare ground filtering shall improve the results by further removing artifacts due to thick vegetation.
More information about the AutoProbaDTM project: http://sites.google.com/site/autoprobadtm
@misc{ref111,
}
A. Jalobeanu, G. Gonçalves: “Probabilistic topographic maps from raw, full-waveform airborne LiDAR data” - AGU Fall Meeting, San Francisco, CA, USA, Dec 2011
The main goal of the AutoProbaDTM project is to derive new methodologies to measure the topography and terrain characteristics using the latest full-waveform airborne LiDAR technology. It includes algorithmic development, implementation, and validation over a large test area. In the long run, we wish to develop techniques that are scalable and applicable to future satellite missions such as LIST (NASA Decadal Survey), to help perform efficient and accurate large-scale mapping.
One of the biggest challenges is to develop fast ways to process huge volumes of raw data without compromising the accuracy and the physical consistency of the result. Over the past decades, significant progress has been made in digital elevation model (DEM) extraction and user interaction has been much reduced, however most algorithms are still supervised. Topographic surveys currently play a central role in sensor calibration and full automation is still an unsolved problem. Moreover, very few existing methods are currently able to propose a quantitative error map with the reconstructed DEM. Traditional validation and quality control only allow to check the discrepancy between the product and a set of reference points, lacking the ability to predict the actual uncertainty related to elevations at chosen locations. We plan to provide fast and automated techniques to derive topographic maps and to compute error maps as well, based on a probabilistic approach to modeling terrains and data acquisition, solving inverse problems and handling uncertainty. Bayesian inference provides a rigorous framework for model reconstruction and error propagation, treating all quantities as random, and combining sources of information optimally. In the future, the uncertainty maps shall help scientists put error bars on quantities derived from the models.

In June 2011, 200 km2 of data were acquired (100 GB of binary files, half a billion waveforms) in central Portugal, over an area of geomorphological and ecological interest, using a Riegl LMS-Q680i sensor. We managed to survey 140 km2 at a satisfactory sampling rate, the angular spacing matching the laser beam divergence and the ground spacing nearly equal to the footprint (almost 4 pts/m2 for a 50cm footprint at 1500 m AGL). We believe this is crucial for a correct processing as aliasing artifacts are significantly reduced, compared to common practice where the spacing is larger than the footprint size.
A reverse engineering had to be done as the data were delivered in a proprietary, undocumented binary format, so we were able to read the waveforms and the essential timing and look angle parameters. An instrument model was developed to account for the overall impulse response and noise properties. The instrument was operated in the low signal to noise ratio regime to minimize the cost per km2, and the limits of state of the art processing methods are reached, hence the need for algorithmic development to achieve both a higher detection rate and an improved robustness.
We will present the latest results from the first stage of the project: a large DEM of the bare ground topogaphy for the entire study area, including an error estimate for each point, the major novelty being the spatial variability of uncertainty.

http://gi.cge.uevora.pt/AutoProbaDTM
@misc{ref103,
}
G. Gonçalves, A. Jalobeanu: “LiDAR boresight calibration: a comparative study” - AGU Fall Meeting, San Francisco, CA, USA, Dec 2011
Within the AutoProbaDTM project, we plan to develop fast and fully automated techniques to derive topographic maps from full-waveform airborne LiDAR data, based on a probabilistic approach to modelling surfaces and data acquisition, solving inverse problems and handling uncertainty. Bayesian inference provides a rigorous framework for unsupervised reconstruction of the DEM and error propagation from the data to the end result, treating all quantities as random variables.
Automatic sensor calibration plays a major role in this project. In fact, the overall accuracy and uncertainty obtained from the LiDAR technology depends on the assembly and calibration of the three system components: the GPS (Global Positioning System), the INS (Inertial Navigation System) and the laser- scanner device. Bore-sight angles are the angular offsets in X,Y and Z directions between the scanner frame and the INS frame measured at the centre of the INS body frame.
In this paper we evaluate some of the principal bore-sight calibration methods and we propose a novel method based on the Bayesian inference to address this problem as well. The first contribution is to use not only the 3D points extracted from the raw waveforms but their uncertainty as well, and to apply a probabilistic surface matching with spatially variable point accuracy in order to obtain the attitude corrections. The second contribution consists of using all the flight lines, where most methods only use the calibration cross. This way we can also estimate the attitude drift and correct for temporal attitude variations as well. Finally, we use the probabilistic framework for error propagation and propose a probability distribution of the calibrated bore- sight angles.
@misc{ref104,
}
C. Gama, A. Jalobeanu: “Evaluation of the short-term sea cliff retreat along the Tróia-Sines Embayed Coast (Costa da Galé sector), using stereo digital aerial images and Bayesian inference” - AGU Fall Meeting, San Francisco, CA, USA, Dec 2011
Monitoring the sediment budget of coastal systems is essential to understand the costal equilibrium, and is an important aspect to be considered in coastal management. Thus, the identification and the quantitative evaluation of sedimentary sources and sinks are the first steps towards a better understanding of the dynamics of coastal morphology.
The Tróia-Sines Embayed Coast (TSEC) in the southwest Portuguese coast corresponds to a continuous sandy beach that extends for approximately 65 km. It is limited at north by the Sado river estuary and at south by the Sines cape. Beaches are discontinuously limited landward by dunes (?42 km) and by sea cliffs (?18 km) made of poorly consolidated Plio-Plistocene detrital deposits. Cliff erosion by subaerial processes or gullying is a continuous phenomenon that contributes a significant amount of sediment to the TSEC coastal system, which is what we want to measure. Mainly due to winter rainfall, sea cliffs develop debris fans at the backshore inner limit, therefore we chose to make morphological measurements at one year interval. Thus, two series digital aerial images at 20 cm resolution were acquired in Oct 2008 and July 2009, supported by a collection of ground control points (GCP) to constrain the sensor orientation. Digital aerial stereo image pairs are used as main data source to reconstruct digital surface models (DSM). A new stereo photogrammetric method is used, based on dense disparity maps and Bayesian inference (Jalobeanu et al, 2010 and Jalobeanu, 2011). The originality of this method is in the computation of the spatial distribution of elevation errors in the DSM using stochastic modelling and probabilistic inference, which helps to detect the statistically significant changes in the estimated topography.
The difference between the two generated DSMs is used to characterize the variability of the main subaerial beach morphodynamics parameters, such as: i) the alongshore beach configuration; ii) the beach width; iii) the berm elevation and iv) the beach-face slope. Indeed, these are essential parameters for understanding the sedimentary dynamics of a coastal sector. Moreover, confidence intervals can be provided for quantitative parameters derived from the DSM, such as volumes of displaced material, slopes or various geometric parameters.
@misc{ref105,
}
D.D. Fitzenz, A. Jalobeanu, M. Ferry: “Toward Implementing Long-term Slip History and Paleoseismicity Into Active Fault Databases to Compute Effective Recurrence Models” - AGU Fall Meeting, San Francisco, CA, USA, Dec 2011
We developed a Bayesian framework for computing the combination of recurrence models that best represents our knowledge using prior beliefs updated by 3 distinct datasets: the catalog of large earthquakes (CLE), the earthquake geology data including dated cumulative offsets, fault length, and expected range of slip per full-segment events, and trench data. Our method determines the relative weights of the parameters of each model, and the relative weights of the models themselves, in an objective fashion, once the datasets and the priors are chosen, thus ensuring reproducible results.
We are interested in faults that are known to experience full-segment events each time they have a very large earthquake. We do not suppose that the slip is always the same, or that it propagates always in the same direction.
We report on the data that proved most influencial (both on the parameter space and on the relative weight associated to each model) in our case study of the Dead Sea Fault (DSF) in Jordan, to encourage their integration in the databases on active faults that are being developed.
Our first step was to couple a non-informative prior on each model parameter space to the catalog of large earthquakes. Going beyond the historical record thanks to archeological records proved very valuable;
We compared observed and synthetic dated cumulative offsets simulated using each candidate recurrence models, a distribution of plausible slip per event, and a minimum and maximum allowed slip rate. The 6 dated cumulative offsets (over a total of 47,000 yrs) proved very discriminant, in particular in cases where they evidence changes in fault slip rates over short time periods (e.g., by a factor of 3 in about 4 inter-event times, in the case of the Jordan Valley Fault);
We showed how a Bayesian analysis of the trench data incorporates the candidate model that is thought to explain the interval of time between consecutive events. It has to be built using an algorithm that goes from each ”observation”, i.e., the uncertain date of the event and its order in the sequence, to the ”true” inter-event time (that we do not know) and the next observation, through the recurrence model. In our method, we need either the ”raw” data, i.e., the radiocarbon dates of all samples used and the location of the event in the stratigraphy, or the probability density functions for the observed event dates, but only if they are reported in their full complexity and if they derive only from the calibration of one pair of radiocarbon ages.
In summary, we recommend that 1) historical and archeological data; 2) long cumulative slip histories; and 3) trench data, be reported with the uncertainties on the age and the slip when applicable (interval when defined by bounds, mean and standard deviation of known pdfs, or discretization of non-parametric pdfs, depending on the method to extract the age). In particular, raw radiocarbon layer ages can be reported as a mean and std dev, and complex calibrated earthquake dates (from oxcal or other softwares) should be sampled (e.g., every 5 years) and full tables should be stored together with the current version of the calibration curve.

http://gi.cge.uevora.pt/SourceMod
@misc{ref106,
}
A. Jalobeanu: “Predicting spatial uncertainties in stereo photogrammetry: achievements and intrinsic limitations” - Journal of Spatial Science, submitted, Nov 2011
We present a new probabilistic method for digital surface model generation from optical stereo pairs, with an expected ability to propagate errors from the data to the final result, providing spatial uncertainty estimates to be used for quantitative analysis in planetary or Earth sciences. Existing stereo-derived surfaces lack rigorous, quantitative error estimates, and we propose to address this issue by deriving a method of error prediction, rather than error assessment as usually done in the area through the use of reference data. We use only the information present in the available data and perform the prediction using Bayesian inference. We start by defining a forward model, using an adaptive radiometric change map to achieve robustness to noise and reflectance effects. A priori smoothness constraints are introduced to stabilize the solution. Solving the inverse problem to recover a surface from noisy data involves fast deterministic optimization techniques. Though the reconstruction results look satisfactory, we conclude that uncertainty estimates computed from two images only are unreliable, which is due to major limitations of stereo, such as non-Lambertian reflectance and incorrect spatial sampling, which violate our underlying assumptions and cause biases that cannot be accounted for in the predicted error budget.
@unpublished{ref110,
}
A. Jalobeanu, M. Ferry: “ASTER GDEM Validation report - Lower Tagus Valley (Portugal)” - internal report, Centro de Geofísica de Évora, Portugal, May 2009
We focus on an area comprising roughly a quarter of the 1°x1° GDEM tile due to the size of the available reference data (raster DEM provided by the Portuguese Geographic Institute, actually 15.3% of the tile area). Within this area the elevation ranges between 0 and 250 m and the slope never exceeds 25°. The differences between GDEM and reference DEM are analyzed in this report. We also present the result of a visual comparison with SRTM 90 m.
@techreport{ref100,
}
A. Jalobeanu: “Impact of DEM uncertainties on flood maps: vulnerability of the Portuguese coast to sea level rise” - 6º Simpósio de Meteorologia e Geofísica da APMG, Aldeia dos Capuchos, Portugal, Mar 2009
Flood maps are usually computed by thresholding digital elevation models (DEM) without taking into account errors on the topography. Even if scientists wish to do so in the future, the only information about DEM uncertainty available now is a RMS error at best. Thus, we propose to use our recent work on uncertainty estimation, allowing us to reconstruct a DEM and the spatial distribution of errors as well. Indeed, relevant flood maps can be derived rigorously if the elevation data comes with error bars. Flood probability maps could be directly computed, either for predefined sea levels, or for uncertain sea level rise predictions coming from global climate change models. The Bayesian framework allows for a rigorous management of various error sources so as to produce physically meaningful vulnerability maps. We plan to apply this methodology to several test sites on the portuguese coast using high-resolution digital aerial imagery.
@misc{ref83,
}
A. Jalobeanu: “Probabilistic Digital Elevation Model Generation For Spatial Accuracy Assessment” - AGU Fall Meeting, San Francisco, CA, USA, Dec 2008
We propose a new method for the measurement of high resolution topography from a stereo pair. The main application area is the study of planetary surfaces.
Digital elevation models (DEM) computed from image pairs using state of the art algorithms usually lack quantitative error estimates. This can be a major issue when the result is used to measure actual physical parameters, such as slope or terrain roughness.
Thus, we propose a new method to infer a dense bidimensional disparity map from two images, that also estimates the spatial distribution of errors. We adopt a probabilistic approach, which provides a rigorous framework for parameter estimation and uncertainty evaluation. All the parameters are described in terms of random variables within a Bayesian framework. We start by defining a forward model, which mainly consists of warping the observed scene using B-Splines and using a spatially adaptive radiometric change map for robustness purposes. An a priori smoothness model is introduced in order to stabilize the solution. Solving the inverse problem to recover the disparity map requires to optimize a global non-convex energy function, which is difficult in practice due to multiple local optima. A deterministic optimization technique based on a multi-grid strategy, followed by a local energy analysis at the optimum, allows to recover the a posteriori probability density function (pdf) of the disparity, which encodes both the optimal solution and the related error map.
Finally, the disparity field is converted into a DEM through a geometric camera model. This camera model is either known initially, or calibrated automatically using the estimated disparity map and available measurements of the topography (existing low-resolution DEM or ground control points). Automatic calibration from uncertain disparity and topography measurements allows for efficient error propagation from the initial data to the generated elevation model.
Results from Mars Express HRSC data are presented. A pair of images (including the nadir view) at 30m resolution was used to obtain a DEM with a vertical accuracy better than 10m in well-textured areas. The lack of information in smooth regions naturally led to large uncertainty estimates.
@misc{ref82,
}
D.D. Fitzenz, A. Jalobeanu, S.H. Hickman: “Integrating Laboratory Compaction Data with Numerical Fault Models: a Bayesian Framework” - Gordon Res. Conference on Rock Deformation, Big Sky, MT, USA, Sep 2006
@misc{ref76,
}
D.D. Fitzenz, A. Jalobeanu, S.H. Hickman: “Integrating Laboratory Compaction Data With Numerical Fault Models: a Bayesian Framework” - European Geosciences Union General Assembly, Vienna, Austria, Apr 2006
When analyzing rock deformation experimental data, one deals with both uncertainty and complexity. This often leads to partial or only qualitative data analyses from the experimental rock mechanics community, which limits the impact of these studies in other communities (e.g., modelling). However, it is a perfect case study for graphical models. We present here a Bayesian framework that can be used both to infer the parameters of a constitutive model from rock compaction data, and to generate porosity reduction within direct fault models from a known (e.g. lab-derived) constitutive relationship, and still keep track of all the uncertainties. This latter step is crucial if we are to go toward process-based seismic hazard assessment. Indeed, the rate of effective stress build-up (namely due to fault compaction) as well as the recovery of fault strength determine how long it will take for different parts of the previously ruptured fault to reach failure again, thus controlling both the timing and the size of the next rupture. But deterministic models need a measure of their robustness to become process-based earthquake-rupture forecast models. It is therefore important to work within a framework able to assess model validity as well as use data uncertainties. Our approach involves a hierarchical inference scheme using several steps of marginalization. We focus on one rather general, though experimentally derived, model of a compaction law, with a stress exponent, an apparent activation energy, and a porosity term as main parameters. We will first describe the method and show how it can help define the number and the duration of the experiments, as well as the range of conditions that would lead to a good determination of the physical parameters. We will then present an application to the Niemeijer et al (EPSL2002) data on quartz. Finally we will show how such creep laws can be implemented in direct models of pore pressure evolution using graphical models as a guide to propagate the uncertainties.
@misc{ref66,
}
D.D. Fitzenz, A. Jalobeanu, S.H. Hickman: “Integrating Laboratory Compaction Data with Numerical Fault Models: a Bayesian Framework” - AGU Fall Meeting, San Francisco, CA, USA, Dec 2004
@misc{ref77,
}
A. Jalobeanu, L. Blanc-Féraud, J. Zerubia: “Digital image processing method in particular for satellite images” - Patent #20040234162, Washington, DC, USA, Nov 2004
The invention concerns processing of digital images, captured by detection of electromagnetic waves, such as satellite pictures. The inventive processing consists in applying a parameterable fractal modelling to Fourier transforms of the pixels of the image and comparing (fig) the thus modelled transforms (aijq, wo) to the initial transforms to bring the parameters (q,w0) closer to the fractal model, and if required, the parameters of a transfer function of the instrument which has captured the image.
@unpublished{ref11,
}
A. Jalobeanu: “Modèles, estimation bayésienne et algorithmes pour la déconvolution d'images satellitaires et aériennes” - University of Nice Sophia Antipolis PhD Thesis, France, Dec 2001
Satellite or aerial images are corrupted by the optical system and the sensor. To reconstruct a good quality image from a noisy and blurred observation, one needs to perform a deconvolution.
First, we recall the principles of the acquisition chain, from optics to the sensor (visible or infrared), enabling us to model the degradation of the image.
In order to reconstruct the image without amplifying the noise, while preserving edges and textures, it is necessary to impose constraints on the reconstructed solution, which consists of choosing a prior model. We study satellite and aerial image modeling, which can be done within both probabilistic and variational frameworks, and using both discrete and continuous models. We propose new statistical models that take into account the fractal properties of natural scenes and their non-stationarity, using multiscale and adaptive approaches.
Next we study different techniques for estimating the model parameters, describing the properties of the images to be reconstructed. These techniques are developed within a Bayesian framework, and can be solved using either stochastic, or deterministic algorithms, depending on the problem.
Finally, we propose new fully automatic reconstruction algorithms. First, we suppose that the degradations (blurring kernel and noise statistics) are known, and we try to reconstruct the unknown image. Second, we consider the case where these degradations are unknown. We perform a blind deconvolution, in two steps, the first step consisting of determining the instrumental parameters, and the second of deconvolving the image with fixed degradation parameters.
Tests have been performed on remote sensing data such as satellite images (SPOT 5 and Pléïades simulations) and high resolution visible and infrared aerial images.
@phdthesis{ref12,
}

Who Am I?

Educational Background

Research Interests

Research topics & Applications I am interested in

Favorite Approaches & Methods

Research Projects (PI)

AutoProbaDTM (2010-2012)

SpaceFusion (2006-2008)

Selected Publications

Journal papers and Book Chapters – peer-reviewed

Conference papers – peer-reviewed

Abstracts, Posters, Preprints, Reports and Theses

Links