Study On The Source And Channel Coding Theory And Techniques Of Video Communication In Error-prone Environments

The compressed video data, in which bits are highly correlated and variable length codes (VLCs) are liable to cause error propagation effect, are very sensitive to channel bit errors and erasures. Usually, the video communication quality can not be guaranteed under the condition that the channel bit error rate (BER) is around 10^-4 without any bit-error resilience measures taken. But, to decrease channel BER only by enhancing the power of the error correction codes will inevitably lead to high computation complexity and large coding delay, which is not suitable for real-time video communication.As the VLC symbol error rate (SER) is the main factor that degrades the video communication quality, in this dissertation, we concentrate our research on the error correction of VLC symbols in both source and channel coding. The ability of the VLC symbol error correction is increased, and, equally to say, the sensitivity of video streams to channel bit errors is decreased and the power and redundancy of channel coding can be reduced. In channel coding researches, we also study the low complexity and low delay channel coding algorithms which are suitable for video communication.The main contributions of this dissertation are as follows. In source coding research, a novel VLC, called alternate VLC (AVLC), is proposed, which decreases the VLC symbol error propagation effect. In channel coding research, we find that the randomness of the code structure of the randomly constructed irregular low-density parity-check (LDPC) codes complies with the binomial distribution. And a method for constructing low-density convolutional (LDC) codes is presented. These achievements have typically applied value in the fields of low complexity and low delay video communication. Finally, we develop LDC codes over rings, which can be directly applied to the high order modulation. The LDC codes over rings can also be used to protect the VLC symbols by 'equal space sampling' method, which greatly decrease the VLC SER and solve the essential problem in the error correction of video streams.In real application, we suggest that both bit error correction codes and symbol error correction codes are combined to protect the video streams, i.e., the channel bit errors are corrected first, then the VLC symbol error-correction method is used, which is an effective way to improve the video communication quality.