基于小波包分解的SAR图像压缩

SAR image compression based on wavelet packet transform

针对SAR图像含有丰富的中、高频信息,而基于小波变换的图像压缩方法会丢失高频细节信息,提出了基于小波包分解的SAR图像编码算法。小波包变换对SAR图像进行完全分解,再用与后续编码器相关联的代价函数进行最佳基搜索,然后根据各子带小波包系数的重要性进行加权,采用多级树集合分裂算法(SPIHT)编码。实验结果表明,该算法更好地保留了SAR图像的细节信息,获得了同压缩比下优于传统SPIHT算法的编码性能,更有利于后续图像处理。

In order to improve the compression efficiency of texture-rich synthetic aperture radar（SAR） images, this paper proposed wavelet packet decomposition based coding method to exploit middle and high frequency components. Firstly, applied wavelet packet transform to SAR images for a full decomposition, and then utilized a cost function related with the sequential coding scheme for best basis selection to improve the representation efficiency of SAR images. At last, weighted and coded wavelet packet coefficients in different subbands according to importance by set partitioning in hierarchical trees （SPIHT） algorithm. The experimental results show that the proposed coding method using wavelet packet transform compares favorably with the conventional wavelet-based SPIHT compression methods and keeps more texture information improving interpretability performance for further SAR image processing.