No abstract

This paper presents the results obtained in the framework of the first part of the French ANR TOSCANE project. This project aims at developing efficient solutions for adaptation of video contents to be delivered over audiovisual networks. We present here a complete transrating scheme for incoming H.264/AVC compressed video bitstreams. The proposed solution is based on frequency selective filtering and has the main advantage to be drift-error free. Experimental results show that the algorithm gives satisfying results in terms of visual quality compared to conventional transrating methods. It has been validated using the JM reference software and then implemented into a dedicated FPGA architecture. The use of the scalable extension of the H.264/AVC standard is finally considered as the following part of the project.

No abstract

One solution to increase the number of users enjoying High Definition-IPTV services and more particularly Video On Demand (VOD) is to adapt video bitrates to meet the user line bitrate constraint. SNR scalability provided by Scalable Video Coding (SVC) is an easy and reliable way to achieve this goal. In this paper, we propose a global solution to adapt compressed HD SVC bitstreams in order to extend the coverage of IPTV VOD services over the DSL link. The novelty of the presented approach relies on the hybrid combination of SNR scalability and base layer trans rating. The proposed solution is designed to optimize the reception video quality for the end-user. Simulation results show that this solution leads to an extension of VOD coverage over DSL link of more than 1000 meters with very satisfying video quality.

In this paper, we propose to compare two different H.264/AVC intra transrating architectures based on frequency position selectivity and frequency level selectivity respectively, using intra high-definition video bitstreams. We evaluate these solutions in terms of bitrate reduction and visual quality using both PSNR and Video Quality Metric (VQM) experiments. Simulation results show that the frequency position selectivity has a slight lower computing complexity while providing better or similar visual quality than frequency level selectivity.