视频中基于LW-PGD和SVM的头肩部检测  

Head-shoulder detection based on LW-PGD and SVM in video

作　　者：钦爽[1] 谢刚[1] 饶钦 郭旭[2] 张文慧[3] 

机构地区：[1]太原理工大学信息工程学院,太原030024 [2]西北工业大学机电学院,西安710072 [3]中国人民大学信息学院,北京100872

出　　处：《计算机应用研究》2014年第3期949-952,共4页Application Research of Computers

基　　金：太原市科技项目人才专项基金项目(120247-28)

摘　　要：针对视频中的行人检测问题,提出了LW-PGD(locating windows based on the pixel gradient direction of the top of head)快速定位头肩部和基于融合特征检测的方法。首先利用头顶像素点的梯度方向具有固定范围这一特性在前景中找出头顶候选点,根据该点快速确定人体头肩部区域,将其作为待测窗口;然后提取待测窗口的方向梯度直方图(histogram of oriented gradient,HOG)特征和HSV(hue saturation value)颜色特征;最后采用支持向量机(support vector machine,SVM)训练得到人体头肩部的分类器。实验表明,与传统的滑动窗口搜索方法相比,根据头顶点可以快速选取含有人体头肩部的待测窗口,提高了检测的效率;HOG和HSV多特征融合提高了检测的精确性,从而提出的算法有助于后续的行人分析。This paper proposed a method of locating head-shoulder quickly and detecting with the fusion feature to solVe the problems of pedestrian detection in video. It firstly selected the candidate pixel points in the foreground, for the pixel gradient direction of the top of the head had a fixed scope. Then it located the areas of human head-shoulder quickly by these points, which defined as the windows to be tested. Next, it extracted the HOG features and HSV features of the windows. Lastly, it trained the head-shoulder classifier by support vector machine （SVM）. Experimental results show that based on the pixel gra- dient direction of the top of the head, the windows which contain head-shoulder can be located more quickly than the tradition- al method, sliding window, which improves the efficiency of the detection. Furthermore, the accuracy of detection is also im- proved by the fusion feature of HOG and HSV. Thus, the proposed method performs well and can help the subsequent analysis of the pedestrians.

关 键 词：像素点梯度方向 方向梯度直方图 HSV颜色特征 支持向量机 头肩部检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP274.5[自动化与计算机技术—计算机科学与技术]