In the present paper, an adaptive LBG VQ algorithm is proposed based on the point to line model for the expansion of LBG VQ algorithm which was proposed in 2009, and using the algorithm proposed in this paper can achieve less error quantification, and can avoid the increase in bits of the larger index generated by the point to line expansion of LBG VQ algorithm. This new algorithm was used in lossless compression of interferential multi-spectral LASIS image, and experiment shows that the adaptive LBG VQ algorithm can remarkably improve the lossless compression performance of interferential multi-spectral image, and that if using these VQ algorithms above after Dual-Direction Prediction, the compression ratio generated by the point to line expansion of LBG VQ algorithm will be a little less than that generated by conventional LBG VQ algorithm, and the compression ratio generated by the algorithm proposed in this paper will be much larger than that generated by conventional LBG VQ algorithm.