| 一种改进HOG特征的行人检测算法 |
| |
| 引用本文： | 田仙仙,鲍泓,徐成.一种改进HOG特征的行人检测算法[J].计算机科学,2014,41(9):320-324. |
| |
| 作者姓名： | 田仙仙 鲍泓 徐成 |
| |
| 作者单位： | 北京联合大学信息服务工程实验室 北京100101 |
| |
| 基金项目： | 本文受国家自然科学基金项目:高密度动态人群场景的多源图像融合研究(61271370),北京市教委科技项目(CIT&TCD20130513),北京市教育委员会科技计划面上项目(SQKM201411417010)资助 |
| |
| 摘 要： | 针对HOG特征检测准确率高、计算量大的特点,通过对HOG特征的结构进行调整,提出了使用Fisher特征挑选准则来挑选出有区别能力的行人特征块,得到MultiHOG特征。该算法结合线性SVM二值分类器,实现行人滑动窗口检测。用Inria标准数据集和自行拍摄数据集进行了测试,结果证明该算法较HOG在准确率及实时性上都有很大的提高。
|
| 关 键 词： | HOG特征 Fisher特征挑选 MultiHOG特征 SVM分类 |
| 收稿时间： | 2013/11/26 0:00:00 |
| 修稿时间： | 2014/3/18 0:00:00 |
Improved HOG Algorithm of Pedestrian Detection |
| |
| TIAN Xian-xian,BAO Hong and XU Cheng.Improved HOG Algorithm of Pedestrian Detection[J].Computer Science,2014,41(9):320-324. |
| |
| Authors: | TIAN Xian-xian BAO Hong and XU Cheng |
| |
| Affiliation: | Beijing Key Laboratory of Information Service Engineering,Beijing Union University,Beijing 100101,China;Beijing Key Laboratory of Information Service Engineering,Beijing Union University,Beijing 100101,China;Beijing Key Laboratory of Information Service Engineering,Beijing Union University,Beijing 100101,China |
| |
| Abstract: | In view of the HOG with large amount of calculation and the high detection accuracy,through the structural adjustment of the HOG features,the paper put forward that using Fisher feature selection criterion picks out the feature blocks full of ability to distinguish characteristics of pedestrian,and finally produces MultiHOG characteristics.Combined with the Lib-SVM classifier, the algorithm in the paper detects pedestrians with slided windows,and has a higher accuracy and real-time performance than HOG. |
| |
| Keywords: | HOG features Fisher feature selection MultiHOG characteristic SVM classification |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《计算机科学》浏览原始摘要信息 |
|
点击此处可从《计算机科学》下载全文 |
|
