A noisy image compression coding scheme is presented based on BayesShrink threshold and EZW algorithm. In order to compress the noisy image, the important idea is introduced that is implementing denoising and coding simultaneously. At first, this paper analyzes the noise influence on image compression, and then introduces the quantizer design. Through combining soft-thresholding function with dead-zone quantitative technology, experiment results show that it achieves much better reconstructed image quality in the same compression ratio comparing with the traditional EZW and JPEG coding scheme.