Abstract: In this paper, a color embedded wavelet image compression algorithm based on human visual system and local image characteristics is presented. Firstly, the image is converted from RGB color space to YUV color space by using a reversible transformation to reduce the psychovisual redundancy, and the wavelet subbands of the decomposed color components are selected (to be encoded) according to energy distribution and human visual system. Then the embedded zerotree coding is performed. But unlike embedded zerotree wavelet (EZW) algorithm, the lowest frequency subband is coded separately from other highpass subbands, and the coefficients in high frequency subband is scanned adaptively in the order of human visual system importance. Finally, the complex context modeling is given by utilizing the correlation of wavelet coefficients and is adopted in the arithmetic coding. The experiment results show that the new color image compression scheme performs better than that of color embedded zerotree wavelet (CEZW) and color zerotree wavelet (CZW) in the aspect of recovery image quality and coding decoding time.