一种改进型双窗的SAR图像边缘检测器  

An Improved Bi-Window Edge Detector for SAR Images

作　　者：梁鑫[1] 徐慧[1] 

机构地区：[1]南京林业大学信息科学与技术学院,江苏南京210037

出　　处：《江汉大学学报（自然科学版）》2015年第2期177-181,共5页Journal of Jianghan University：Natural Science Edition

摘　　要：传统矩形双窗在检测非理想边缘时容易在边缘附近检测出错误边缘点,出现检测结果会偏移的问题。针对传统矩形窗函数平滑效果较差的局限性,设计了一种Gauss-Gamma形双窗函数算法,采用可接收操作特征(ROC)曲线来评价使用这两种双窗检测器在SAR图像中的性能。结果表明,新方法的检测有更好的边缘连接性,在边缘附近错误边缘较少,明显优于矩形双窗检测器。With traditional rectangular bi- window to detect non- ideal edge,iteasily detects an error about the edge point in the vicinity of the edge,and the test results will appear offset. Aiming at the limitation of traditional rectangular window that the function smoothing effect is poorer,designed a kind of Gauss-Gamma type double window function algorithm,by using the ROC curve to compare and evaluate the two kinds of dual-window detector performance used in SAR images.The results showed that the new method of detection had a better edge connectivity,the marginal error near the edge was less,and was significantly better than the rectangular bi-window detector.

关 键 词：矩形双窗 SAR图像 边缘检测 Gauss-Gamma形双窗 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TN911.73[自动化与计算机技术—计算机科学与技术]