Satellite multi-spectral image series are produced by image location, so there is strong correlation between neighbor images. The difference from usual image series is that there are apparent shift features between the image series. In order to use this feature to improve the compression efficiency and decrease the complexity of the coding system, a novel multi-spectrical image series coding algorithm is proposed. When the relative shift between two images is detected by the match algorithm in wavelet field, the error image between the images is coded in stead of the following image coding such that the whole coding efficiency is improved. The algorithm is simpler than the one based on three-dimensional wavelet transform, and the disadvantage with large memory requirement and big coding delay is overcome. The simulation results show that better effect is achieved than the one based on three-dimensional wavelet transform.