In order to solve the problem of edge and texture blurring of multi-spectral image caused by the wavelet transform and bit-plane encoder of traditional method not taking into account image characteristics, a compression algorithm of multi-spectral TDICCD image having few bands is proposed in the present paper. The proposed direction-adaptive lifting DWT can adaptively choose the best lifting direction and use the Lagrange interpolation technique to make prediction according to its local characteristics. This method makes good use of the image texture features. The proposed rate control algorithm can adaptively allocate the rate bits according to image content. The experiments results showed that the proposed compression algorithm has good compressive property. Compared with traditional approaches, the performance of our method corresponds to the conventional approach for smooth images. However, it was superior to others for images with rich texture and a large number of edge detail information. In the compression ration 4∶1, the average peak signal to noise ratio of the proposed compression algorithm could improve the average PSNR by 1.939 dB. They effectively protect the edge and texture detail information for multi-spectral image. It is very suitable for the application of space multi-spectral CCD image with rich texture and a large number of edge detail information.