遥感超光谱图像是三维立体图像，随着成像光谱仪的发展超光谱图像的数据量越来越庞大，难以直接传输和存储，必须对图像进行压缩。由于遥感图像信息十分宝贵，应采用无损压缩或近无损压缩方法。到目前为止，一直没有形成一套成熟或标准的超光谱图像压缩技术。因此，对遥感超光谱图像压缩编码的研究具有重要的应用价值。 本文采用基于整数小波变换的编码方法对遥感超光谱空间图像进行编码压缩。考虑到遥感超光谱图像的特殊性和压缩算法的硬件实现，对遥感超光谱空间图像先采用D9/7整数小波变换，取得近无损的小波变换系数。图像经过小波分解后，嵌入式零树编码（EZW）就在当前阈值条件下，建立频带间小波稀疏的关系，来去除小波系数的冗余信息，从而以新的方式组织成为数据流。再通过算术编码器来完成最终的编码，将数据流压缩为带宽极小的压缩数据流。该算法为超光谱图像提供了很好的压缩解决方案：当对图像质量要求不高时，可以只传压缩码流中的较低层，当对图像质量要求较高时，继续传输较高层，实现了真正意义上的渐进传输，有效的节省了宝贵的链路资源，同时又保留了所需的原始图像数据信息。通过VC++仿真实验证明，在相同的压缩比下，基于整数小波变换的算法PSNR值优于离散小波变换压缩算法和JPEG算法。 本文以TI公司的新型数字多媒体处理器DM642为硬件平台，对基于整数小波变换的图像压缩算法进行了仿真。DM642的核心是C6416型高性能数字信号处理器，时钟频率为600MHz，指令执行速度高达4800MIPS，并且C64x还提供了一些特别适用于图像处理的指令，这为基于整数小波变换的超光谱图像压缩算法提供了可能性。实验通过CCS和TMS320DM642EVM板的结合使用，证明本文采用的基于D9/7整数小波变换的编码方法可以在DSP上实现超光谱图像的近无损压缩。

Remotely sensed hyperspectral image is a 3D stereoscopic image. With remotely sensed technology developing, the data of remotely sensed hyperspectral image is huge and it is hard to deliver and saving directly. So compression to the image is necessary. Because the remotely sensed hyperspectral image information is very precious, the lossless compression or nearly lossless compression method is needed possibly. So far, the super spectrum picture hasn't been becoming a set of mature or standard compress technique. Therefore the research of compress coding to the remotely sensed hyperspectral image has the important applied value. This paper is focus on the implemetation of the compress algorithm for two dimensioned hyperspectral remote sensing image, which mainly achieved by Integer Wavelet Transformation(IWT). Based on the character of the hyperspectral remote sensing image and the handware realization, the D9/7 of IWT is applied to the space hyperspectral remote sensing image, generating nearly lossless transformation coefficient. Subsequently, under current threshold, a sparse inter-frequency wavelet relationship was established, EZW would be invoked to neutralize the unwanted information in wavelet transformation coefficient and reorganize the data stream. These data stream were passed through the arithmetical coder, compressed datastream was finally achieved on very narrow bandwidth. By using this algorithm, in case of lower image quality was preferred, lower layer of the data stream would be transmitted; and the higher layer of the data stream are deliveried only if higher image quality was required, consequently, the link resource was saved, meanwhile, necessary information from original image could be secured. It has been proven by VC++ simulation that Integer Wavelet Transformation based compression could provide better PSNR value than Dispersed Wavelet Transformation based or JPEG compression. It would be a good compression solution for hyperspectral remote sensing image. A new type digital multimedia processor DM642 from Texas Instrument was adopted as the hardware platform to achieve the Integer Wavelet Transformation based image compression. The core of the DM642 is C6416 high performance digital signal processor with 600MHz clock frequency and 4800MIPS instruction execution. Moreover, C64x could provide a serial of instruction dedicated for image processing; therefore DM642 was selected as the hardware platform to achieve the Integer Wavelet Transformation based image compression. A lab test had been conducted on the combination of CCS and TMS320DM642EVM, and approaching lossless hyperspectral remote sensing image compression has been realized on DSP platform thanks to the D9/7 Integer Wavelet Transformation coding scheme depicted in this article.