You will find on this web page illustrations of some results in relation to our past and on-going research in the areas of image, video, multi-view, light fields processing. The problems addressed include representation, compression, editing which requires addressing inverse problems such as inpainting, and super-resolution.

Light fields computational imaging

Computational light fields imaging ERC advanced grant CLIM
Light fields segmentation using a ray-based graph structure
Super-rays for efficient light field processing
Homography-based low rank approximation of light fields
Partial Light Field Tomographic Reconstruction From a Fixed Camera Focal Stack
Lytro Illum Light Field Dataset Lytro Illum Light Field Dataset
Lytro F01 Light Field Dataset Lytro F01 Light Field Dataset

High Dynamic range Imaging

AdaptiveRequantization HDR imaging: Tone Mapping using Re-quantization for Compression
Logo Not found HDR imaging: Rate-Distortion Optimized Tone Mapping for Compression
Logo Not found HDR imaging: Gradient-Based TMO for Rate-Distortion Optimized Backward-Compatible HDR Compression

Multi-view Video Processing

Depth maps extraction Depth maps extraction from Multi-view Videos
VRML Automatic reconstruction of video sequences from natural video sequences
VRML 3D reconstruction and applications
VRML Computed 3D Models for Very Low Bit-rate Video Coding and Video Manipulation

Image analysis for compression

Clustering based fast epitome construction
Epitome Inpainting for Image Compression
Scalable Image Coding based on Epitomes

Visual Attention

vam vam vam Computational model using high-level visual features
vam vam vam Visual fixation analysis and datasets
vam How visual attention is modified by disparities and textures changes?

Image and Video Inpainting

Examplar-based inpainting based on local geometry
inpaiting inpaiting Depth-based image completion for view synthesis
3D view synthesis with inter-view consistency
inpaiting Hierarchical super-resolustion-based inpainting
inpaiting Reduced complexity video inpainting: application to object removal and background estimation


vam Low-Complexity Single-Image Super-Resolution based on Nonnegative Neighbor Embedding
vam Single image super-resolution using sparse representations with structure constraints

Sparse representations, Convolutional networks, auto-encoders, learning for compression and denoising

quantizationAE Auto-encodeur optimisé au sens débit-distorsion
dicoLearning Stochastic Winner-Take-All Auto-Encoders
dicoLearning Structured dictionary learning for sparse representations and image coding
Oriented Wavelet Transform Oriented Wavelet Transform
clusteringLinearMappingNoiseRemoval Learning clustering-based linear mappings for quantization noise removal
Webmaster: Valid CSS! Valid XHTML IRISA
Last time modified: 2017-08-30