基于轮廓片段匹配和图搜索的红外目标识别方法  

Infrared Target Recognition Algorithm Based on Contour Fragment Matching and Graph Searching

作　　者：苏娟[1] 杨罗 张阳阳 

机构地区：[1]第二炮兵工程大学907教研室 [2]96427部队气象室

出　　处：《兵工学报》2015年第5期854-860,共7页Acta Armamentarii

基　　金：国家自然科学基金项目(61302195;61132008)

摘　　要：针对前视红外(FLIR)目标识别中复杂背景干扰和目标轮廓片段缺失的问题,提出了一种基于轮廓片段匹配和图搜索的目标识别方法。提取基准图和实时图的轮廓特征并进行分段,利用片段的方位和朝向特征进行匹配并构建匹配图;在基准图轮廓片段间的空间关系约束下,利用图搜索策略遍历匹配图得到候选目标轮廓片段组合;结合全局相似性度量实现目标的匹配识别。仿真实验结果表明,该方法对复杂背景干扰和轮廓片段缺失的情况具有较好的适应性和鲁棒性。A target recognition method based on contour fragment matching and graph searching is proposed for complicated background interference and contour fragment missing in forward looking infrared （FLIR） target recognition.The contour features are extracted from the reference image and the real-time image,and decomposed into the fragments,and the orientation information of each fragment are utilized to realize the fragment matching and construct the matching graph.Under the constraint of the spatial relationship between the reference contour fragments,the graph searching strategy is used to traverse the matching graph to obtain candidate target contour fragment combinations.Finally,the global similarity measure is used to obtain the final recognition.Experimental results show that the proposed method has good adaptability and robustness to complicated background interference and contour fragment missing.

关 键 词：兵器科学与技术 目标识别 形状匹配 片段匹配 图搜索 空间关系 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]