Source-channel Coding Research Of Image Compression And Transmission

With the development of science and technology, there is an increasing demand for multimedia communications. As a basic part of the multimedia communications, image communication has been extensively slighted. In real communication systems, it is of theoretical and practical significance to investigate the approaches of image compression and transmission using source-channel coding.In order to achieve the image compression and transmission over wireless channels, a series of error-resilient measures are adapted to protect the bit-stream produced by Set Partitioned Embedded Block Coder (SPECK). Three different source-channel coding mechanisms have been implemented and then the effect of Turbo on error protection has been detailed analyzed. Specifically, the research is as follows:1. A SPECK based Equal Error Protection scheme (EEP) has been proposed. Full considering the characteristics of SPECK bit-stream, the method used Reed-Solomon codes (RS) and cyclic redundancy check codes (CRC) to resist disturbing and get better performance than original SPECK.2. A SPECK based Unequal Error Protection scheme (UEP) has been proposed. First, a novel structure of three level lists is employed. Image blocks are coded according to their importance on reconstructing image which promotes efficiency. Second, jointed with UEP, the system perfectly yields good performance and high reliability. Experimental results show that the proposed scheme can obviously improve the image quality after transmitting even if the channels are in bad condition.3. We have proposed a joint source and channel remote sensing image compression system for image transmission over Binary Symmetric Channels (BSC). To avoid the failure caused by the loss of synchronization and the error propagation, we improve the error resilience of SPECK with an acceptable degradation of quality. The novel packets division permits to continue reconstructing image when uncorrectable errors occur. The system makes full use of the bit-stream and the simulations show the scheme is better than a Equal Error Protection scheme (EEP) and a Unequal Error Protection scheme (UEP).At last a new approach has been provided that is Turbo based error resilient transformation research. Efficient implementations are also developed to account for different situations. A new efficient decode-stop mechanism has been proposed. Research and simulation on 4 different decoding methods with Parameters provide a direction for the future work.