遥感图象的数据压缩 - Read.DOCVIP

遥感图象的数据压缩 目 录 1 摘要…………………………………………………………………………………………?1 2 引言…………………………………………………………………………………………?2 3 压缩编码理论及其设计……………………………………………………………………?3 3.1 K-L变换原理………………………………………………………………………3 3.2 小波变换原理………………………………………………………………………5 3.2.1 Fouier分析与小波分析……………………………………………………5 3.2.2有关小波的基本概念………………………………………………………7 3.2.3小波零树编码……………………………………………………………???11 3.2.4非线性量化编码…………………………………………………………???12 4 压缩编码软件具体实现……………………………………………………………………12 4.1 关于MATLAB工具的介绍……………………………………………………????12 4.2 K-L变换及小波变换编码的实现………………………………………………???13 4.2.1 16点K-L变换的实现…………………………………………………???14 4.2.2 64点K-L变换的实现…………………………………………………???14 4.2.3 小波分解及零树编码的实现…………………………………………?????15 4.2.4 小波分解及非线性量化的实现…………………………………………?17 5 实验结果和讨论…………………………………………………………………………???18 6 结束语……………………………………………………………………………………???24 7 参考文献…………………………………………………………………………………???25 附录　K-L变换实验结果…………………………………………………………………???26 1摘要 K-L变换是最优的正交变换，在遥感数字图象处理中主要用于压缩维数及去除相关性。实际运用中采用64点K-L变换保留4点可获得11.79倍的压缩率并且集中了图象的大部分能量。小波变换在时域和频域具有良好的局部化性能，与零树编码及非线性量化编码相结合能取得较大的压缩率且失真较小，对静止图象进行三次小波分解再扫描三次编码可获得25—30的压缩率，PSNR为25dB左右。MATLAB语言中提供了先进的图像处理工具箱，工具箱中包含有丰富的函数和命令。使用MATLAB语言实现K-L变换及小波零树编码及非线性量化编码两种压缩方法的结合，是本文进行的新尝试，由对64点K-L变换加后续静止图象压缩实验结果的观察可得，取四点恢复压缩率能达到187倍，PSNR为12—18dB左右，取得较为理想的效果。 remote sensing image data compressing Abstract:Karhunen-Leove Transformation(KLT) is the best method for multispectral image data compression.And it is usally applied in remote sensing image system for image compressing and remove the spectral correlation .Experiment results shows that remaining 4 points from 64-points-KLT can reach the ratio of 11.79 and keep most of the energy of the image.Because of the good localization performances of both time and frequency field, the Zerotree coding algorithm which is wavelet-based is applied to he compression of image data quantization of transform domain,algorithm of encoding,and almost lossless.While MATLAB has supplied advanced toolbox, where has abundant functions and demands. It makes our processing of image become easier and can ge