基于分层树集合分割的分布式干涉多光谱图像压缩  

Compression for Distributed Interferential Multispectral Image Based on Set Partitioning in Hierarchical Trees

作　　者：宋娟[1] 李云松[1] 吴成柯[1] 

机构地区：[1]西安电子科技大学ISN国家重点实验室,陕西西安710071

出　　处：《光学学报》2010年第2期358-363,共6页Acta Optica Sinica

基　　金：国家自然科学基金重点项目(60532060);国家自然科学基金(60672117;60802076);111基地项目(B08038);西安电子科技大学基本科研业务费项目"低复杂度分布式的光谱图像压缩编码方法研究"资助课题

摘　　要：提出了基于分层树集合分割(SPIHT)的分布式干涉多光谱图像压缩算法,将SPIHT和分布式信源编码(DSC)有效结合,以同时去除图像序列帧内、帧间及小波系数子带间的强相关性。根据DSC的特点,提出了基于边信息的辅助重构方法,以减少量化误差,提高重构质量。为了进一步保护光谱信息,采用基于加权提升系数的感兴趣区域(ROI)方法,对包含大部分光谱能量的强干涉区域优先编码。辅助重构方法在ROI保护方面也具有明显的优势,在保护强干涉区域的同时,能减少弱干涉区域的信息损失。实验结果表明,和变换域的DSC相比,算法的峰值信噪比(PSNR)提高了大约0.5 dB,拟合原始光谱曲线的性能较其它传统算法也有明显的提高。A distributed compression method for interferential multispectral image based on set partitioning in hierarchical trees （SPIHT） is proposed,which combines SPIHT with the framework of distributed source coding （DSC） to exploit the strong correlation within and among frames as well as different scales in the same orientation. And an auxiliary reconstruction method based on side information is presented at the decoder to decrease quantization error. To protect the spectral information efficiently,ROI method based on weighted lifting coefficients is performed to priorly encode the intensive interference area that contains most spectral energy. The auxiliary reconstruction method also takes an obvious advantage on ROI coding. The method not only protects intensive interference area but also reduces the loss of the weak interference area as much as possible. The experimental results show that the proposed algorithm obtains about 0.5 dB gains over DSC in transform domain and its ability to fit the original spectral curves is greatly improved compared with other traditional compression methods.

关 键 词：图像处理 干涉多光谱图像压缩 分布式信源编码 分层树集合分割 辅助重构 感兴趣区域 

分 类 号：TN919.81[电子电信—通信与信息系统]