基于检测识别的实孔径雷达景象匹配定位方法  

A method of registration for real aperture radar scene based on airfield detection and recognition

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]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）》2005年第2期25-27,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：针对恶劣成像条件下的机场目标实孔径雷达图像匹配问题,提出了一种基于目标检测识别的定位方法.通过线特征滤波预处理和投影变换检测机场跑道线,进一步用模板匹配识别跑道端点以进行定位.对实际图像试验结果表明该算法能够准确定位灰度相关算法难以正确匹配的强噪声、低对比度实时图像,并且具有参考图制备简单,实时性强的特点.A method based on airfield detection and recognition was proposed, aiming to the registration for airfield under ill condition of real aperture radar imaging. Through the linear feature enhancement and Randon transform to locate the runway line, then using model matching to find the exactly position of the runway end for registration of the airfield. The experiment showed that this method could accurately registration the airfield, even for serious speckle and low contrast radar image. It has the benefit of simplified reference map preparation and less computation than the gray correlation algorithm.

关 键 词：实孔径雷达 目标识别 景象匹配 机场检测 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]