利用改进分形特征对SAR图像目标检测方法的研究  

Study on Target Detection of SAR Image Using Improved Fractal Feature

作　　者：承德保[1] 胡风明[2,3] 杨汝良[2] 

机构地区：[1]北京航空航天大学经济管理学院,北京100191 [2]中国科学院电子学研究所,北京100190 [3]中国科学院研究生院,北京100190

出　　处：《电子与信息学报》2009年第1期164-168,共5页Journal of Electronics & Information Technology

摘　　要：改进分形特征是用指数小波在一个尺度上对检测图像滤波,针对特定大小目标用能量关系函数求得各像素点的分形特征。该文研究了利用改进分形特征对SAR图像进行目标检测的方法,分别使用改进特征与扩展分形特征对单一背景和复杂背景条件下的SAR图像进行目标检测,结果表明:改进分形特征能够在这两种背景条件下以更低虚警率检测出全部特定大小的目标,目标空间可分辨性好、位置指示准确;但在复杂背景条件下的检测虚警率比单一背景下的检测虚警率有所上升。The improved fractal feature of each pixel can be extracted based on the filtered image using the exponential wavelet at one scale and energy functions. This paper studies the method of the improved fractal feature for SAR images target detection. The results of target detections using the improved fractal feature are compared with that using the method of Extended Fractal （EF） for SAR images in the simple and complex backgrounds respectively. Results state the method using the improved fractal feature can not only detect all size-fixed targets but also have lower false alarm rates, spatial resolution of the detected targets are higher and the locations of the detected targets are more accurate in both of the backgrounds, but worse false alarm rates in complex background than in the simple background using the improved fractal feature.

关 键 词：合成孔径雷达 目标检测 指数小波变换 改进分形特征 扩展分形特征 

分 类 号：TN958[电子电信—信号与信息处理]