摘要
为了解决当前特征点匹配算法在前景遮挡、背景复杂的干扰下,存在匹配能力不足的问题,提出了基于改进Brief的目标鲁棒匹配算法。首先,对存在匹配干扰的图像展开分析,通过基于不变矩的特征点检测方法,改进breif特征点检测。然后扩大滤波范围来处理目标积分图像,通过基于特征点主方向的旋转矩阵,构建Brief特征描述算子;最后结合欧式距离和KNN,进行匹配运算,并编程实现算法,实验测试结果显示:与当前传统特征点匹配算法相比,在特征遮挡严重、背景极复杂的干扰下,本文算法拥有较强的匹配精度与鲁棒性。
In order to solve the problem of insufficient recognition, in the current target identification algorithm under the prospect of sheltering and complex background interference, this paper put forward the feature point robust match algo-rithm based on improved brief, goals. First of all, to identify interference image analysis, Brief, algorithm and improve ment ideas were put forward. And then, to improve the Brief feature point detection, fast robust Brief feature points de tection operator is put forward. Finally, by expanding the scope of the filtering processing target integral image, improve the Brief description operator. Algorithm is realized and test by using C language. The test results show that compared with the traditional target recognition algorithm, the target block, the interference of background is extremely complex, the algorithm has strong ability of recognition in this paper.
出处
《国外电子测量技术》
2017年第2期29-33,共5页
Foreign Electronic Measurement Technology
关键词
Brief特征
目标匹配
特征点检测
积分图像
特征描述
鲁棒
Brief
target identification
feature point detection
integral image
character description
robust
