Improved Channel Optimized Image Coding Based on More Rational Source Model

Joint source channel coding (JSCC) has been proved a promising scheme for image communication in varying channel because of the properties of graceful degradation/improvement with limited coding delay and complexity, especially the combination of wavelet transform and JSCC is an easy way to achieve high-powered channel optimized image coding. For the purpose of attaining better performance, the source statistical characteristics including the intra-subband correlation of wavelet transform coefficients are considered, and then a Markov source modeling method capable of reserving and characterizing the correlation is developed, furthermore the utilization of the residual redundancy is also demonstrated within three referenced channel optimized image coders with fixed rate. Simulations show that the source model is more rational for wavelet transformed images and the improved channel optimized image coders utilizing the residual source redundancy performs better with no or negligible extra expense of bits and coding complexity.