基于双重轮廓特征匹配的遮挡目标识别算法  

Occlusion target recognition algorithm based on double contour feature matching

作　　者：宋建辉 郝奕全 SONG Jianhui;HAO Yiquan(Shenyang Ligong University,Shenyang 110159,China)

机构地区：[1]沈阳理工大学自动化与电气工程学院,辽宁沈阳110159

出　　处：《通信与信息技术》2024年第6期54-58,共5页Communication & Information Technology

基　　金：辽宁省教育厅高等学校基本科研项目(项目编号:LJKZ0275)沈阳市中青年科技创新人才支持计划项目(项目编号:RC210247)。

摘　　要：针对传统目标识别算法对于遮挡目标匹配难度大、误匹配率高的问题,提出一种基于双重轮廓特征匹配的遮挡目标识别算法。首先,将形状轮廓分为采样点和轮廓有效片段两个层次,其中轮廓片段是利用离散曲线演化算法和分段优化算法获得的,用于表征形状的全局信息,而轮廓点的用于表征局部细节信息。其次,通过动态规整算法DTW(Dynamic Time Warping)衡量轮廓片段轮廓点之间的精确型傅里叶高度函数特征(Accurate Fourier Height Functions 2,AFHF2)的相似度,同时计算轮廓分段的置信度特征。最后,结合分段置信度再次匹配识别。实验结果表明,与常见的遮挡目标识别算法相比,所提出算法的检索精度有明显的提升。To address the problem that traditional target recognition algorithms have difficulty in matching occluded targets and high error matching rate,a target recognition algorithm based on dual contour feature matching is proposed.Firstly,the shape contour is divided into two levels of sampling points and contour effective segments,where the contour segments are obtained by using the dis-crete curve evolution algorithm and segment optimization algorithm to represent the global information of the shape,while the contour points are used to represent the local detail information.Secondly,the similarity between the contour segments and contour points is measured by the dynamic alignment algorithm DTW(Dynamic Time Warping)to calculate the similarity of the accurate Fourier height function features(Accurate Fourier height Functions 2,AFHF2),and the confidence feature of the segmentation is calculated at the same time.Finally,the segmentation confidence is used again for matching and recognition.The experimental results show that the pro-posed algorithm has a significant improvement in retrieval accuracy compared with common occluded target recognition algorithms.

关 键 词：双重匹配 遮挡目标 精确型傅里叶高度函数 特征置信度 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]