Landsat-TM光谱图像数据压缩技术的研究

A Research on Compression Technique of Landsat(TM) Image

采用提升格式的整数小波变换 ,对Landsat -TM图像数据进行小波分解。在分析Landsat-TM图像整数小波变换后系数特点的基础上 ,将小波变换压缩技术中的零数编码推广到多光谱图像压缩中。采用整数 97小波变换同时去除空间与谱间冗余 ,对单波段图像各个子带小波系数的能量、均值、方差进行了数据统计。由小波系数的相关性 ,采用 3DSPIHT对小波系数进行压缩编码。实验结果表明 ,该方法的去相关性能良好 。

In the paper TM image data was decomposed by using lifting scheme of Integer Wavelet Transform(IWT).On the base of analyzing characteristics of transformed wavelet coefficient , zero-tree coding compression of wavelet transform was extended to the compression of multispectral images .Integer 97 wavelet transform was applied to reduce spatial decorrelation and inter-band correlation. The statistic characteristic of wavelet coefficient was discussed after transform. The statistics of wavelet coefficient after transformed for every TM band were particularly analysed, and their statistics of energy, average and var were completed. 3D SPIHT is used in processing the transformed wavelet coefficient.The results of experiment indicate that the performance of removing correlation is better and effect of the compression is greatly improved.