摘要
传统高光谱遥感图像压缩重构算法重点在信号传输与存储方面,采样频率要大于信号宽度的两倍,硬件实现较为困难,为此提出基于激光束匹配的高光谱遥感图像压缩重构算法研究。采用K-SVD算法训练冗余字典,稀疏分解高光谱遥感图像,得到遥感图像稀疏分解系数,以此为基础,量化处理遥感图像稀疏分解系数,通过JPEG-LS无损压缩算法压缩遥感图像,主要分为三个阶段:预测阶段、Golomb编码阶段与游程模式编码阶段,以得到的遥感图像压缩结果为依据,基于激光束匹配理论对遥感图像进行匹配追踪重构,实现了高光谱遥感图像的压缩重构。仿真实验结果显示:提出算法完成的图像压缩比例随着噪声程度的增加而下降,图像重构范围实际值80.00%~90.01%,超过最低重构程度限值75.56%,充分说明提出算法具备较好的压缩重构效果。
The traditional compression and reconstruction algorithms of hyperspectral remote sensing image focus on signal transmission and storage,and the sampling frequency is more than twice the signal width,so it is difficult to realize the hardware implementation.K-SVD algorithm is used to train redundant dictionary,sparse decompose hyperspectral remote sensing image,and get the sparse decomposition coefficient of remote sensing image.On this basis,the sparse decomposition coefficient of remote sensing image is processed quantitatively,and the remote sensing image is compressed by JPEG-LS lossless compression algorithm,which is mainly divided into three stages:prediction stage,Golomb coding stage and run mode coding stage According to the results,the hyperspectral remote sensing image is reconstructed by matching pursuit based on laser beam matching theory,and the compression reconstruction of hyperspectral remote sensing image is realized.The simulation results show that:the image compression ratio of the proposed algorithm decreases with the increase of noise level.The actual value of image reconstruction range is 80.00%-90.01%,which exceeds the minimum reconstruction degree limit of 75.56%,which fully shows that the proposed algorithm has good compression reconstruction effect.
作者
徐爽
刘贵珊
XU Shuang;LIU Guishan(Yinchuan Energy Institute,Yinchuan 750021,China;Ningxia University,Yinchuan 750021,China)
出处
《激光杂志》
北大核心
2020年第12期88-92,共5页
Laser Journal
基金
国家自然科学基金(No.31760435)
宁夏高校科学研究项目(No.NGY2018254)。
关键词
激光束匹配
高光谱遥感图像
压缩
重构
laser beam matching
hyperspectral remote sensing image
compression
reconstruction
