奇异值分解遥感图像压缩算法研究

Research on Compression Algorithm of Remote Sensing Image Based on Improved Singular Value Decomposition

研究遥感图像信息量大且不利于压缩的特点,针对目前一般遥感图像压缩算法的问题,为获得较大的CR(压缩比)和PSNR(峰值信噪比),提出了一种改进的奇异值分解图像压缩算法。算法主要是选取部分奇异值,然后利用奇异向量重构矩阵进行图像压缩。经过建模对于不同内容和纹理的遥感图像,在一定的压缩比下,均获得PSNR>34dB的恢复图像,在不损失最低频信息的同时较好地保持了遥感图像中丰富的高频信息,实现了高质量的图像压缩。经实验证明,与传统的奇异值分解相比,算法在相同图像压缩比的情况下,获得了更高的峰值信噪比,很好地完成遥感图像压缩的任务,为实际的星上应用提供理论依据。

In connection with the characteristics that remote sensing image has large information and is difficult to compress,an algorithm of image compression which used the improved singular value decomposition（SVD） was proposed so as to get large CR and PSNR because of the problems of present algorithm.The algorithm selected part of singular values,and then used singular vectors to rebuild the original matrix.Through the simulation modeling of MATLAB,the result shows that this algorithm acquires the restoration images of PSNR（Peak Signal Noise Radio） 32 dB for all images of different contents and texture with certain CR（compression ratio）,which keeps much high frequency information of the remote sensing image to realize the high quality image compression.The experiments show that the algorithm gets higher PSNR at the same image CR comparing with traditional SVD.The algorithm can accomplish the task of remote sensing image compression well and offer theory basis for practical star application.