基于整数小波变换的多光谱遥感图像压缩技术

Compression of Multispectral Remote Sensing Image Based on Integer Wavelet Transform

在分析多光谱图像小波变换后系数特点的基础上,提出了一种基于整数小波变换的3维集合分裂嵌入块编码(3D SPECK)压缩方法。该方法将小波变换压缩技术中的零树编码推广到多光谱图像压缩中,采用整数小波变换去除空间冗余,对单波段图像,采用2D SPECK编码,对多波段图像,谱域上构成的小波矢量采用离散余弦变换(DCT)进行变换,对变换后的系数进行3D SPECK编码。实验结果表明,该方法硬件实现简单,编码解码时间快,对内存要求低。

3D SPECK compression of multispectral remote sensing image is presented, which is based on integer wavelet transform （IWT） by analyzing the characteristics of transformed wavelet coefficient. By the method the zero-tree coding of the compression of wavelet transform is extended to the compression of multispectral remote sensing image. Integer wavelet transform is applied to reduce the spatial decorrelation, and DCI＂ is applied to reduce the inter-band correlation. 3D SPECK is used in processing the transformed wavelet coefficient. The results of experiment indicate that hardware is easier to realize, the time of coding and decoding is less, and the requirement of EMS memory is lower.