基于假设检验的SAR图像机场跑道自动识别  

Automatic recognition of airfield runways based on hypothesis testing in SAR images

作　　者：杨卫东[1] 左峥嵘[2] 张天序[2] 

机构地区：[1]华中科技大学图像识别与人工智能研究所,湖北武汉430074 [2]华中科技大学图像信息处理与智能控制教育部重点实验室,湖北武汉430074

出　　处：《华中科技大学学报（自然科学版）》2006年第2期11-13,共3页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金重点项目(60135020)

摘　　要：针对复杂条件下合成孔径雷达图像中机场目标自动检测识别问题,提出了一种基于假设检验的机场跑道自动识别算法,利用雷达图像中跑道灰度特性和结构知识,通过迭代分割和形态学滤波提取感兴趣区域,抑制具有类似灰度特性的水域对跑道线检测的影响,并结合Hough变换和线段跟踪连接提取候选跑道,最后采用假设检验方法对机场跑道进行识别.试验结果表明该方法可快速有效地检测识别复杂背景下低分辨率、低信噪比合成孔径雷达(SAR)图像中的机场跑道.In consideration of the automatic detection of airfield and its recognition in synthetic aperture radar （SAR） images, a hypothesis-testing-based algorithm for automatic recognition of airfield runways in SAR images was proposed. The region of interesting （ROD was extracted through iterarive segmentation and morphology filtering and the impact of lakes or rivers on runway detection for the similar intensity property was restrained. The candidate runway was extracted by Hough transform and segment tracing. The true airfield runway was identified by hypothesis testing with the knowledge of the runway intensity property and its structure features. Experimental results showed that the proposed algorithm can automatically detect the airfield runway in low resolution and low sig- nal-to-noise （SNR） SAR image under complex background in time.

关 键 词：合成孔径雷达 目标识别 机场检测 假设检验 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]