Robust Wireless Video Transmission Scheme Combined With The Application Layer Error Resilience Tools

Along with the advancement of video encoding and the development of wireless network, video communication has become one of the primary services over wireless networks. However, there are still severe technical challenges in wireless video communications, because the error-prone and time-varying band-limited wireless channels always have difficulty in meeting the requirements for large-data-volume and high-service-quality video transmission. Therefore, the video transmission and related video processing techniques have become part of the key research areas in video communication. In this dissertation, the error control techniques in wireless video transmission are researched from three aspects, i.e., the error resilience tools in video encoding, video transmission over OFDM wireless channels and the effective integration of the above techniques for wireless video transmission.The major contributions of the dissertation are summarized as follows:1. Two content analysis methods based on motion characteristics are proposed, i.e., a new multi-metric scene cut detection method and a key frame extraction method based on acceleration. Based on the two methods, the importance of each frame can be classified, which further facilitates the design of adaptive group of pictures (AGOP) structure.2. Based on the two newly developed scene analysis methods, an adaptive GOP structure construction method is proposed. After performing scene content analysis, the frames in the video sequence can easily be classified according to their impacts on the entire coding efficiency and error resilience performance. Then different coding types are assigned to the pictures, which will facilitate the determination of the service priority to the coding bits of each picture in the transmission scheme using differential service strategy (e.g., unequal error protection). By employing the proposed AGOP video encoding, together with advanced error concealment methods, the quality of the received video can be further secured when transmitted over error-prone channels. Moreover, an improved rate control scheme is also devised for the AGOP encoding scheme to alleviate the fluctuation in the pictures'coding quality. Extensive experiments have shown that AGOP is an effective error resilient tool besides improving the coding efficiency. 3. The key issues of the adaptation of video bitstream and wireless channels are investigated. Based on the analyses on the characteristics of compressed video data and the wireless OFDM channel, a novel three dimensional interleaving technique, namely, spatial-frequency-temporal interleaving (SFTI), is proposed. By introducing frequency and temporal diversity, the proposed method reduces the degradation in video quality caused by the impacts of wireless fading and facilitates error concealment on the application layer at the receiver.4. Several error resilience techniques are integrated to develop a stream transmission scheme over OFDM wireless channels. The general idea of the proposed scheme is to effectively integrate the proposed SFTI with error resilience tools in video coding and transmission. By combining the SFTI with multiple reference picture selection, flexible macroblock technique, and the proposed AGOP, respectively, robustness of video transmission can be well consolidated, which is demonstrated by an average quality improvement of over 2 dB for the reconstructed video at the receiver.