ISP Seminars

Phase Transition Analysis of Recently Proposed Compressed Sensing Recovery Algorithms

Adriana Gonzalez Gonzalez

16 February 2011

Compressed Sensing (CS) is a very important technique in signal processing that represents a change in the philosophy behind the acquisition process. By incorporating compression into the measurements, CS allows taking samples at a much lower rate than the Nyquist rate, which diminishes the amount of samples, the acquisition time and data dimension. This gives CS its own place in several applications of signal processing, such as imaging techniques. The main issue in CS is the recovery of the original signal, which has become a matter of solving underdetermined linear systems by having a sparse prior knowledge on the original signal. In this seminar two recently proposed CS recovery algorithms are presented and compared with IHT, one of the most used algorithms in this field due to its simplicity. The analysis is presented in the way of Phase Transition Diagrams, which allow observing the behavior of the algorithm regarding CS: the amount of measurements to take in order to recover a signal with a certain amount of information.

Ressources :

-- Wednesday, 16 February 2011 at 10:30 (45 min.)

Feature Point-Based 3D Mesh Watermarking that Withstands the Cropping Attack

Mireia Montanola

4 February 2011

State-of the-art robust 3D watermarking schemes already withstand combinations of a wide variety of attacks (e.g. noise addition, simplification, smoothing, etc). Nevertheless, there are practical limitations of existing 3D watermarking methods due to their extreme sensitivity to cropping. Spread Transform Dither Modulation (STDM) method is an extension of Quantization Index Modulation (QIM). Besides the simplicity and the trade-off between high capacity and robustness provided by QIM methods, it is also resistant against re-quantization. This paper focuses on two state-of-theart techniques which offer different and complementary advantages, respectively QIM-based 3D watermarking and feature point-based watermarking synchronization. The idea is to combine both in such a way that the new scheme would benefit from the advantages of both techniques and compensate for their respective fragilities. The resulting scheme does not make use of the original 3D model in detection but of some parameters as side-information. We show that robustness against cropping and other common attacks is achieved provided that at least one feature point as well as its corresponding local neighborhood is retrieved.

Ressources :

-- Friday, 4 February 2011 at 10:30 (45 min.)

{ "name":"Feature Point-Based 3D Mesh Watermarking that Withstands the Cropping Attack", "description":"State-of the-art robust 3D watermarking schemes already withstand combinations of a wide variety of attacks (e.g. noise addition, simplification, smoothing, etc). Nevertheless, there are practical limitations of existing 3D watermarking methods due to their extreme sensitivity to cropping. Spread Transform Dither Modulation (STDM) method is an extension of Quantization Index Modulation (QIM). Besides the simplicity and the trade-off between high capacity and robustness provided by QIM methods, it is also resistant against re-quantization. This paper focuses on two state-of-theart techniques which offer different and complementary advantages, respectively QIM-based 3D watermarking and feature point-based watermarking synchronization. The idea is to combine both in such a way that the new scheme would benefit from the advantages of both techniques and compensate for their respective fragilities. The resulting scheme does not make use of the original 3D model in detection but of some parameters as side-information. We show that robustness against cropping and other common attacks is achieved provided that at least one feature point as well as its corresponding local neighborhood is retrieved.", "startDate":"2011-02-04", "endDate":"2011-02-04", "startTime":"10:30", "endTime":"11:15", "location":"", "label":"Add to my Calendar", "options":[ "Apple", "Google", "iCal", "Microsoft365", "MicrosoftTeams", "Outlook.com" ], "timeZone":"Europe/Berlin", "trigger":"click", "inline":true, "listStyle":"modal", "iCalFileName":"Seminar-Reminder" }

Optimal Dense Disparity Map Quantization and Residual Prediction for Lossless Stereo Image Coding".

Amit Kumar K.C.

15 December 2010

This presentation is based on my thesis work, which is compression of stereo images.

Ressources :

-- Wednesday, 15 December 2010 at 10:30 (45 min.)

Integrated H.264 Region-of-Interest Detection, Tracking and Compression for Surveillance Scenes".

Ivan Alen Fernandez

1 December 2010

''Motion detection and tracking is an important vision topic for many applications such as video surveillance. When this process takes place during video encoding and transmission, Regions-of-Interest (RoIs) turn to be a very useful tool in order to favor the encoding of such regions compared to the fixed background. In this presentation, we show that, in conjunction with effective spatio-temporal filters, H264 Motion Estimation can be efficiently used for a robust and coarse-grain detection and tracking of moving objects. The integration of the compression and detection modules enables the prediction of ROIs positions from previous frames, offering therefore tracking at a very low computational cost. In the case of high resolution sequences affected by severe quality degradation (such as improper interlacing, light reflections and camera shaking), the global video compression ratio can be dramatically improved without damaging the ROI. This is especially the case when appropriate encoding options, i.e. appropriate Flexible Macroblock Ordering (FMO) types, are exploited. Different proposals are offered to maximize the quality of the RoI facing a dynamic constraint of the network bandwidth.''

-- Wednesday, 1 December 2010 at 14:00 (45 min.)

{ "name":"Integrated H.264 Region-of-Interest Detection, Tracking and Compression for Surveillance Scenes'.", "description":"''Motion detection and tracking is an important vision topic for many applications such as video surveillance. When this process takes place during video encoding and transmission, Regions-of-Interest (RoIs) turn to be a very useful tool in order to favor the encoding of such regions compared to the fixed background. In this presentation, we show that, in conjunction with effective spatio-temporal filters, H264 Motion Estimation can be efficiently used for a robust and coarse-grain detection and tracking of moving objects. The integration of the compression and detection modules enables the prediction of ROIs positions from previous frames, offering therefore tracking at a very low computational cost. In the case of high resolution sequences affected by severe quality degradation (such as improper interlacing, light reflections and camera shaking), the global video compression ratio can be dramatically improved without damaging the ROI. This is especially the case when appropriate encoding options, i.e. appropriate Flexible Macroblock Ordering (FMO) types, are exploited. Different proposals are offered to maximize the quality of the RoI facing a dynamic constraint of the network bandwidth.''", "startDate":"2010-12-01", "endDate":"2010-12-01", "startTime":"14:00", "endTime":"14:45", "location":"", "label":"Add to my Calendar", "options":[ "Apple", "Google", "iCal", "Microsoft365", "MicrosoftTeams", "Outlook.com" ], "timeZone":"Europe/Berlin", "trigger":"click", "inline":true, "listStyle":"modal", "iCalFileName":"Seminar-Reminder" }

Gene network reconstruction from expression microarray data

Jérôme Ambroise

17 November 2010

In this seminar, I will present several algorithms enabling to reconstruct gene regulatory network from gene expression data. A gene regulatory network focuses on interactions between transcription factors and their target genes. Unsupervised methods will be first presented including relevance networks and Gaussian Graphical Models. If the number p of variables (genes) is much larger than the number n of microarrays experiments, standard approaches to compute Gaussian Graphical Models are inappropriate. Suitable alternative based either on regularized estimation of the inverse covariance matrix, or on regularized high dimensional regression will be briefly introduced. Then, I will present the supervised algorithm recently proposed by F. Mordelet and J.-P. Vert. Using this supervised approach conducts to an improvement of predictive performances but this method cannot be used to predict interactions involving an 'orphan' transcription factor. Finally I will present the 'TNIFSED' method that infers gene regulatory network from the integration of correlation and partial correlation coefficients with the gene functional similarity through a supervised classifier. Compared to the supervised SIRENE algorithm, TNIFSED performed slightly worse when transcription factors are associated with a wide range of yet identified target genes. However, unlike SIRENE, predictive performance of TNIFSED does not decrease with the number of target genes, a feature which makes TNIFSED suitable to discover target genes associated with ’orphan’ transcription factors.

Ressources :

-- Wednesday, 17 November 2010 at 11:00 (45 min.)

{ "name":"Gene network reconstruction from expression microarray data", "description":"In this seminar, I will present several algorithms enabling to reconstruct gene regulatory network from gene expression data. A gene regulatory network focuses on interactions between transcription factors and their target genes. Unsupervised methods will be first presented including relevance networks and Gaussian Graphical Models. If the number p of variables (genes) is much larger than the number n of microarrays experiments, standard approaches to compute Gaussian Graphical Models are inappropriate. Suitable alternative based either on regularized estimation of the inverse covariance matrix, or on regularized high dimensional regression will be briefly introduced. Then, I will present the supervised algorithm recently proposed by F. Mordelet and J.-P. Vert. Using this supervised approach conducts to an improvement of predictive performances but this method cannot be used to predict interactions involving an 'orphan' transcription factor. Finally I will present the 'TNIFSED' method that infers gene regulatory network from the integration of correlation and partial correlation coefficients with the gene functional similarity through a supervised classifier. Compared to the supervised SIRENE algorithm, TNIFSED performed slightly worse when transcription factors are associated with a wide range of yet identified target genes. However, unlike SIRENE, predictive performance of TNIFSED does not decrease with the number of target genes, a feature which makes TNIFSED suitable to discover target genes associated with ’orphan’ transcription factors.", "startDate":"2010-11-17", "endDate":"2010-11-17", "startTime":"11:00", "endTime":"11:45", "location":"", "label":"Add to my Calendar", "options":[ "Apple", "Google", "iCal", "Microsoft365", "MicrosoftTeams", "Outlook.com" ], "timeZone":"Europe/Berlin", "trigger":"click", "inline":true, "listStyle":"modal", "iCalFileName":"Seminar-Reminder" }

Image Denoising using Non-Local Wavelet Bases

Laurent Jacques

4 November 2010

None

Ressources :

-- Thursday, 4 November 2010 at 14:00 (45 min.)

ISP Seminars

Previous Seminars