针对星载多光谱图像压缩，提出了基于子带谱间变换的压缩算法。该算法首先对多光谱图像序列的每个波段分别进行空间二维小波变换，以此去除多光谱图像的空间相关性;为了去除多光谱图像的谱间相关性，将小波分解后的每一层子带作为整体，采用串行成对变换的方式对两个波段进行子带谱间KLT变换;最后,利用最优截断的嵌入式块编码算法对变换后的所有主成分同时进行最优率失真压缩。实验结果表明，该算法能够获得较好的压缩性能，同时具有较低的编码复杂度，适用于星载多光谱图像的压缩。

To the problem of spaceborne multi-spectral image compression，a compression scheme based on sub-band spectral transform is proposed. First，two-dimensional discrete wavelet transform is performed to each band of multi-spectral image sequences for removing the spatial correlation.Subsequently，in order to remove the spectral correlation，each level of the sub-bands resulted from wavelet transform are regarded as a whole，and sub-band spectral KLT is performed on the two bands by using serial pairwise transform. Finally，Embedded Block Coding with Optimized Truncation(EBCOT)algorithm is performed on all principal components to realize optimal rate-distortion compression. Experimental results show that the proposed algorithm not only provides better compression performance，but also has lower encoding complexity，which is suitable for the onboard compression of multi-spectral images.