利用平均粗糙度特征对SAR图像目标检测方法的研究

Study on Target Detection of SAR Images Using Average Roughness Feature

研究利用平均粗糙度特征对由MSTAR数据集构成的SAR图像进行目标检测的方法,分别使用平均粗糙度特征与CFAR方法对不同复杂度背景条件下的SAR图像进行目标检测。仿真结果表明,平均粗糙度特征能够在不同背景条件下以更低虚警率检测出特定大小的目标,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。

This paper studies the method of the average roughness feature for the SAR image target detection by MSTAR database.The results of target detections using the average roughness feature are compared with those using the method of CFAR for SAR images in different backgrounds.Simulation results indicate that the method using the average roughness feature can not only detect size-fixed targets but also have lower false alarm rates,its spatial resolutions of the detected targets are higher and the locations of the detected targets are more accurate in different backgrounds,but the false alarm rates in complex background are worse than those in the simple background using the average roughness feature.