基于DTW的轮廓特征目标识别算法
Contour Feature Target Recognition Algorithm Based on DTW
宋建辉 1郝奕全 1刘晓阳 1赵亚威1
作者信息
- 1. 沈阳理工大学 自动化与电气工程学院,沈阳 110159
- 折叠
摘要
为提升传统目标轮廓识别算法的实时性,提出一种基于动态时间规整(dynamic time warping,DTW)的轮廓特征目标识别算法.该算法将质心高度增量特征描述符与DTW相似性度量算法相结合,首先对目标轮廓均匀提取采样点,并对目标图像以及模板图像轮廓点的质心高度增量特征进行提取,然后使用DTW算法寻找规整路径的方法对目标图像以及模板图像的特征矩阵进行相似性度量,最后定义形状复杂度,同时联合翻转目标的二次匹配得出最终识别结果.实验结果表明,所提出算法在MPEG-7、Kimia99 数据集中对待测形状能够在保证识别率优于大多数常见的传统目标识别算法的同时提升目标识别的实时性.
Abstract
In order to improve the real-time performance of traditional target recognition algo-rithms,a contour feature target recognition algorithm based on DTW(Dynamic Time Warping)is proposed.The algorithm combines the centroid height incremental feature descriptor with the DTW similarity measurement algorithm.It first extracts the sample points uniformly from the target con-tour,and then extracts the centroid height incremental features of the target image and the template image contour point.And then the DTW algorithm is used to find the regular path to measure the similarity of the feature matrix of the target image and the template image,define the shape com-plexity at the end,and combine the secondary matching of the flipped targets to obtain the final rec-ognition result.The experimental results show that the proposed algorithm can boost the real-time expression of target recognition in MPEG-7 and Kimia99 data sets,while ensuring that the recogni-tion rate is better than most common traditional target recognition algorithms.
关键词
轮廓点/质心高度增量特征/图像识别/动态时间规整Key words
contour points/centroid highly incremental feature/image recognition/dynamic time warping引用本文复制引用
基金项目
辽宁省教育厅高等学校基本科研项目(LJKZ0275)
沈阳市中青年科技创新人才支持计划项目(RC210247)
出版年
2024
NETL
NSTL
万方数据