摘要
在机器视觉领域中,目标检测是一个至关重要的研究问题,提出一种新的目标检测算法。该算法在MB-LBP特征级联分类器的基础上再串联两个分类器,分别为基于SIFT特征分类器和基于SURF特征分类器。首先,当测试图像通过MB-LBP特征级联分类器时,所有的目标都没漏检,但是有部分非目标被误检为目标,接着依次通过基于SIFT特征分类器和基于SURF特征分类器,检测结果只剩下目标和极少数非目标。实验结果表明,由MB-LBP特征级联分类器、基于SIFT特征分类器和基于SURF特征分类器构成的新级联分类器可以有效提高查准率。
In the field of machine vision,object detection is a crucial issue in the research.A new algorithm of object detection is proposed in this paper.On the basis of MB⁃LBP feature cascade classifier,the algorithm is connected in series with other two classifiers(the classifier based on SIFT feature and the classifier based on SURF feature).When the images under test are sent through the MB⁃LBP feature cascade classifier,all of the objects are detected,which means there is no missing detection,however,some non⁃objects are mistakenly detected as the objects.Then they are sent through the classifier based on SIFT feature and the classifier based on SURF feature in sequence,and the result is that only the objects and a very few non⁃objects are accepted.Experimental results show that the new cascade classifier constructed by MB⁃LBP feature cascade classifier,the classifier based on SIFT feature and classifier based on SURF feature can effectively improve the precision ratio.
作者
熊凯龙
范方亮
汪保玉
吴建华
XIONG Kailong;FAN Fangliang;WANG Baoyu;WU Jianhua(Information Engineering School of Nanchang University,Nanchang 330031,China;Gongqing College of Nanchang University,Jiujiang 332020,China)
出处
《现代电子技术》
北大核心
2020年第1期48-52,共5页
Modern Electronics Technique
基金
国家自然科学基金资助项目（61662047）
