Charles Yaacoub

Feedback Channel Suppression in Distributed Video Coding with Adaptive Rate Allocation and Quantization for Multiuser Applications

We present a novel rate allocation technique for distributed multiuser video coding systems without the need for a permanent feedback channel. Based on analytical calculations, the system unequally distributes the available bandwidth among the different users, taking into account the actual amount of movement in the transmitted video as well as the transmission conditions of each user. On one hand, the quantization parameters are dynamically tuned in order to optimize the decoding quality. On the other hand, a frame dropping mechanism allows the system to avoid unnecessary channel use, when the analytical estimations show that the successful decoding of a given frame is not possible because of very high motion or bad channel conditions. A significant gain in the system performance is noticed compared with the case of equal allocation of channel resources and constant quantization parameters.

New Adaptive Algorithms for GOP Size Control with Return Channel Suppression in Wyner-Ziv Video Coding

We present novel algorithms for adaptive GOP size control in distributed Wyner-Ziv video coding, where an H.264 video codec is used for intracoding of key frames. The proposed algorithms rely on theoretical calculations to estimate the bit rate necessary for the successful decoding of Wyner-Ziv frames without the need for a feedback channel, which makes the system suitable for broadcasting applications. Additionally, in regions where H.264 intracoding outperforms Wyner-Ziv coding, the system automatically switches to intracoding mode in order to improve the overall performance. Simulations results show a significant gain in the average PSNR that can reach 3 dB compared to pure H.264 intracoding, and 0.8 dB compared to fixed-GOP Wyner-Ziv coding.

Content adaptive gop size control with feedback channel suppression in distributed video coding

This paper presents a novel algorithm for content adaptive GOP size control in distributed video coding. The GOP size is dynamically varied along the sequence, depending on motion activity. Automatic mode selection allows the system to switch between H.264 intra-coding and Wyner-Ziv coding modes to optimize the overall performance. Furthermore, the encoder determines a suitable compression ratio for the Wyner-Ziv frames without the need for a feedback channel. Simulation results show significant improvement in the average system performance, compared to fixed GOP Wyner-Ziv and H.264 intra-coding.

Side information enhancement using an adaptive hash-based genetic algorithm in a Wyner-Ziv context

Side information construction in Wyner-Ziv video coding is a sensible task which strongly influences the final ratedistortion performance of the scheme. This side information is usually generated through an interpolation of the previous and next images. Some of the zones of a scene however, such as the occlusions, cannot be estimated with other frames. In this paper we propose to avoid this problem by sending some hash information for these unpredictable zones of the image. The resulting algorithm is described and tested here. The obtained results show the advantages of using localized hash information for the high error zones in distributed video coding.

Texture Synthesis Using the Structure Tensor

This paper proposes a two-stage texture synthesis algorithm. At the first stage, a structure tensor map carrying information about the local orientation is synthesized from the exemplars data and used at the second stage to constrain the synthesis of the texture. Keeping in mind that the algorithm should be able to reproduce as faithfully as possible the visual aspect, statistics, and morphology of the input sample, the method is tested on various textures and compared objectively with existing methods, highlighting its strength in successfully synthesizing the output texture in many situations where traditional algorithms fail to reproduce the exemplars patterns. The promising results pave the way towards the synthesis of accurately large and multi-scale patterns as it is the case for carbon material samples showing laminar structures, for example.

Joint Source-Channel Wyner-Ziv Coding in Wireless Video Sensor Networks

In this paper, we show an application of Wyner-Ziv video coding in a wireless video sensor network. A rate estimation technique that permits to determine the compression rate without the need of a feedback channel is proposed. This technique allows the system to assign unequal transmission rates for different video sensors. It employs a cross-layer approach that takes into account the amount of motion in each of the captured video scenes and the transmission conditions of each sensor. Simulation results show an improvement in the average system performance obtained with the proposed rate allocation method compared to the traditional rate allocation technique where all sensors are assigned an equal bandwidth. This improvement will motivate further research on the implementation of Wyner-Ziv video coding in Wireless Video Sensor Networks.

Structure tensor based synthesis of directional textures for virtual material design

Exemplar-based texture synthesis schemes are promising for virtual material design. They provide impressive results in many cases, but fail in difficult situations with large and multi-scale patterns, or with long range directional variations. Since a prior synthesis of a geometric layer may help in the synthesis of the texture layer, a two-stage structure/texture synthesis algorithm is proposed. At the first stage, a structure tensor map carrying information about the local orientation is synthesized from the exemplars data, and at the second stage, the synthesized tensor field is used to constrain the synthesis of the texture. Results show that the proposed approach not only yields better synthesized textures, but also successfully synthesizes the output texture in many situations where traditional algorithms fail to reproduce the exemplars patterns, which paves the way towards the synthesis of accurately large and multi-scale patterns as it is the case for pyrolytic carbon samples showing laminar structures observed by Transmission Electronic Microscopy.

Improving hash-based Wyner-Ziv video coding using genetic algorithms

This work aims at improving the quality of the side information in distributed video coding. A novel technique for the fusion of multiple side information is developed for this purpose. Based on genetic algorithms, the proposed frame fusion technique incurs no additional complexity at the encoder side, in case of hash-based distributed video coding. Simulation results show a significant improvement in the side information quality compared to other interpolation techniques available in the literature, which greatly improves the rate-distortion performance of a distributed video codec, where the gain in PSNR can reach 7 dB.

Impact of light field compression on focus stack and extended focus images

Light Fields capturing all light rays at every point in space and in all directions contain very rich information about the scene. This rich description of the scene enables advanced image creation capabilities, such as re-focusing or extended depth of field from a single capture. But, it yields a very high volume of data which needs compression. This paper studies the impact of Light Fields compression on two key functionalities: refocusing and extended focus. The sub-aperture images forming the Light Field are compressed as a video sequence with HEVC. A focus stack and the scene depth map are computed from the compressed light field and are used to render an image with an extended depth of field (called the extended focus image). It has been first observed that the Light Field could be compressed with a factor up to 700 without significantly affecting the visual quality of both refocused and extended focus images. To further analyze the compression effect, a dedicated quality evaluation method based on contrast and gradient measurements is considered to differentiate the natural geometrical blur from the blur resulting from compression. As a second part of the experiments, it is shown that the texture distortion of the in-focus regions in the focus stacks is the main cause of the quality degradation in the extended focus and that the depth errors do not impact the extended focus quality unless the light field is significantly distorted with a compression ratio of around 2000:1.

A Cross-Layer Approach for Dynamic Rate Allocation in H.264 Multi-User Video Streaming

In this paper, we propose a cross-layer approach for the dynamic allocation of unequal transmission rates between several users requesting video from a streaming server. Our allocation technique takes into account distortions caused by source coding, channel impairments and error concealment in order to assign users different priority levels. These priorities are then used to serve users in such a way that the highest levels correspond to the worst performing users. Variable bitrates are achieved by unequal error protection of the different video streams using rate-compatible punctured turbo-codes. Simulation results show a significant improvement in the overall system performance compared to a traditional system where all users are allocated equal channel resources.