Font Size: a A A

Research On Error Resilience Video Coding And Rate Control In Wireless Networks

Posted on:2008-02-01Degree:DoctorType:Dissertation
Country:ChinaCandidate:P LiFull Text:PDF
GTID:1118360242494059Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
Video coding and its related problems in real time communications over error-prone networks, especially over wireless communication networks, draw more and more attentions in recent years. To further improve error resilience, rate control, and video transcoding, several topics are studied in this thesis, including joint source channel coding, multiple description coding, basic unit level rate control and spatial downscaling transcoders.In the condition without feedback channel, a novel robust coding scheme is proposed, based on data partitioning and unequal error protection, for H.264 video transmission over wireless network. According to their effects to error propagation, the C type partition is further divided into several subtypes in the proposed scheme. Then unequal error protection is applied to all partitions, including A, B and C types. The model of rate allocation and an improved bidirectional iterative local search algorithm are also presented. The proposed method is robust to packet loss and superior to the conventional H.264 coding scheme.In the condition with feedback channel, a novel robust joint source channel coding method is proposed for MPEG-4 video over noisy channel, basing on the analysis of hierarchy and the significant characteristic of MPEG-4 bits stream. In the proposed method, the base layer bits streams are reordered according to their importance and then interlaced. Then, multiple sub-layers of texture information are generated based on their rate distortion function, where encoder adopts the joint optimization between the number of multiple sub-layers and the bitrate of channel coding based on the feedback information. Therefore, the proposed method suits the variable channel conditions better and exhibits higher performance.A new combined multiple description and layered coding method based on adaptive frame insertion (AEI-MDLC) for video bit stream transmission over wireless ad-hoc network is proposed. The main idea of the method is to adaptively insert the transition frames according to the relative motion between two adjacent frames. Then, the video sequence is divided into two descriptions with independent prediction loops and respective base layer and enhanced layer descriptions. The proposed method can help the decoder recover from packet loss more quickly compared with previous methods, consequently provides more stable and better reconstruction quality.For H.264/AVC Video Coding, an improved basic unit level rate control algorithm build on the JM80 framework is proposed in the thesis. The algorithm adopts a novel motion-based rate prediction model which is more accurate and direct than the reference algorithm. The bit allocation scheme is rate-distortion optimized by applying the Lagrange method. The proposed algorithm meets the target bit rate closely and achieves considerable PSNR improvement in all test scenarios.Building a H.263 transcoding platform is the last part of this thesis, in which a cascaded DCT-domain downscaling transcoder is realized. Several methods related to motion vectors re-sample, coding modes determination and downscaling in DCT-domain are tested and analyzed.
Keywords/Search Tags:joint source channel coding, unequal error protection, multiple description coding, rate control, video transcoding
PDF Full Text Request
Related items