Research On The Key Technologies For H.264 Video Terminal And DSP Implementation

With the coming of the 21st century,man is entering into a new multimedia age. Video communication is becoming people's new demand. ITU-T and ISO joined to establish and release a new generation international video coding standard: H.264/AVC. Compare with the former standards,the new technique offers higher compression ratio and better video quality, and will be applied in many fields.However, there are still some technical problems in the application of H.264 standard. How to ensure image quality in video transmission with error-prone networks and how to realize real-time image coding on embed platform such as DSP are hotspots which are researched recently.An adaptive intra refresh method for robust transmission in error-prone networks of H.264 is proposed first in this dissertation. In this method, an intra-refresh matrix is created to record the dependence between the macroblocks in the current frame and the reference frame. Based on the intra-refresh matrix, an end-to-end rate distortion model for error-prone networks is proposed to estimate the motion mode of each macroblock in the following frame. As experimental results shown, this method can improve the decoded video quality with a constant bit rate in error-prone networks.Next, a generic FEC(forward error correction) payload structure are given in the paper. With this new structure, the tornado code is used to realize the unequal protection of the H.264 video stream. The simulation results show that the proposed technique can improve the decoded video quality greatly in channels with high packet loss rate.At last, the hardware structure and software programming methods of TMS320 DM642 are introduced in this paper, and the implement of H.264 encoder on it is also given. The methods are adapted to optimize the H.264 encoder based on DSP platform, such as packing data, intrinsic, linear assembly, and direct memory access. According to DSP optimization, real-time encoding for video sequence in CIF format is realized. Compared with the encoder without optimization, the image quality and bit-rate have little change. Furthermore, the methods can be used in practice.