基于特征检测和小波变换的SAR图象的智能压缩

在地质普查、安全监控等众多任务中，合成孔径雷达成像系统产生着海量的图象数据．必须通过有限的通道传回地面，迫切需要一个能从图象的信息中自动检测目标或兴趣区域的智能压缩技术。为此，本文提出了一种基于标准特征检测方法的Bayes网络融合检测方法检测兴趣区域，并据此对图象的小波分解系数进行变化和编码，从而既实现高压缩，又保持了关键区域图象的质量。

SAR Image Intelligent Compression Based on Feature Detection and Wavelet Transfom

In many Earth observation and surveillance missions, a large amount data is collected by the on-board SAR, and must be transmitted to ground through a channel with the limited capacity; in this case, one algorithm of intelligent compression that can select region of interest or target automatically on-board is needed. In this paper, we propose an algorithm for ROI of on-board image selection, which can separate the underlying image features from the speckle noise based on fusion and Bayes Neural Network. And then, we can achieve a good compression ratios as well as making sure that the import targets and areas of interest are not removed and degraded.