CCD遥感图像的小波分解特性及编码压缩  

Wavelet transform characteristics and compression coding of remote sensing images

作　　者：赵洁红[1] 孙文军[2] 孟忠[2] 郝志航[2] 

机构地区：[1]长春工程学院信息工程系,吉林长春130021 [2]中国科学院长春光学精密机械与物理研究所,吉林长春130033

出　　处：《光学精密工程》2004年第2期205-210,共6页Optics and Precision Engineering

摘　　要：研究了遥感图像的小波变换系数统计特性及小波系数的相关性。通过试验得出小波变换系数分布的一般规律和遥感图像基本特征,即数据的局部相关性弱、纹理丰富复杂和低冗余度,不利于高保真压缩,而遥感图像的小波变换系数却具有很强的局部相关性。提出了插分预测编码(DPCM)与集合树多级分裂(SPIHT)算法相结合的遥感图像小波编码算法,用该算法对CCD采集的遥感图像进行了3级小波分解和重构,并与EZW算法、SPIHT算法进行了比较。结果表明,算法所得的遥感图像没有方块效应,图像恢复质量较高,在相同压缩比和比特率低的情况下,明显优于其他两种算法。The statistical characteristics of remote sensing image are studied by using wavelet transform and presented definition of similarity for wavelet coefficient. The general distribution rule of wavelet transform coefficient and basic characteristics of remote sensing images, such as lesser similarity, complicated and abundant texture, lesser redundancy degree, are indicated by experiments, which are not beneficial to compression in high fidelity. The wavelet transform coefficient of remote sensing image has stronger local similarity, hereby a version of combining of DPCM with SPIHT is presented. The experimental results show that the new compression scheme performs better than SPIHT arithmetic in terms of image reconstruction quality and encoding/decoding speed.

关 键 词：遥感图像 小波变换 子带相关 差分预测编码 集合树多级分裂法 图像压缩 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]