基于均值比率梯度增量符号相关的SAR景象匹配算法  

SAR Scene Matching Algorithm Based on ROA Gradient's Increment Sign Correlation

作　　者：杨朝辉[1,2] 陈鹰[2] 

机构地区：[1]苏州科技学院环境科学与工程学院,江苏苏州215011 [2]同济大学测量与国土信息工程系．上海200092

出　　处：《测绘科学技术学报》2011年第6期421-424,428,共5页Journal of Geomatics Science and Technology

基　　金：江苏省高校自然科学研究资助项目(10KJB42000);航天基金资助项目(0747-0540SITC2099-4)

摘　　要：基于均值比率梯度增量符号相关原理,提出了一种SAR景象匹配算法。首先利用ROA梯度算子提取实时图和参考图的边缘梯度信息;然后引入增量符号对梯度信息进行二值化编码并实现相关分析;最后在多尺度影像分解的基础上完成由粗至精的逐层匹配。实验结果表明,该算法对噪声与梯度变化具有较强的鲁棒性,具有较高的正确匹配率和实时性。SAR scene matching algorithm based on ROA（ratio of averages） gradient＇s increment sign correlation was proposc, d. Firstly, gradient information of reference image and realktime image were obtained with ratio of averages operator. Then, according to the increment and decrement information of ROA gradient, images were encoded to binary array and correlation similarity was calculated by increment sign correlation. Finally, SAR scene matching was performed from global to level--wise matching according to multi-scale theory. The experi- mental results showed that this method was robust under gradient change and noise conditions and has great matching probability and computational efficiency.

关 键 词：均值比率 多尺度 景象匹配 增量符号相关 合成孔径雷达 梯度 

分 类 号：P237.3[天文地球—摄影测量与遥感]