The EZW(Embedded Zerotree Wavelet)-like image compression algorithms lack

the error-resilience ability in noisy transmission environments. We propose the channel-

optimized source coding scheme to improve the robustness of them. First, a block-based

method is adopted to localize the error effects. Then we assign bits to each block by

applying dynamic bit allocation to the block-based EZW algorithm based on the rate-

distortion functions computed from the channel noise models. The performance of

our method was evaluated on both the binary symmetric channel and burst noise channel

models.