基于边缘类型比率特征的人体检测算法

来源期刊:中南大学学报(自然科学版)2011年第z1期

论文作者:顾爽 陈启军

文章页码:717 - 723

关键词:行人检测;边缘类型特征;立体视觉;线性SVM分类器

Key words:people detection; edge type feature; stereo vision; linear SVM classifier

摘    要:研究基于立体视觉的室内人体检测方法,提出了一种描述图像一定区域内边缘类型比率的特征。算法在离线阶段将不同边缘点根据其邻域内的边缘梯度直方图(EOH)量化为不同的类别。检测时,首先通过立体视觉分割出场景中若干兴趣区域(ROI);然后,将每个ROI划分为若干相互重叠的部分,统计每个部分中每类边缘点出现的比率,组成特征向量。该特征能够较好地体现不同物体轮廓之间的差异。最后,针对不同正负样本特征训练线性SVM作为检测系统的分类器。实验表明,算法具有较好的人体检测结果。

Abstract: A people detection algorithm based on the stereo vision is proposed and a new feature which describes the ratio of edge types in a certain area is introduced. During the offline stage, the edge points are quantized into different types based on their histograms of edge orientation. During the online stage, a stereo segmentation algorithm is used to select certain ROIs. The edge type ratio is then calculated in some overlapped blocks in each ROI to form the feature vector. The feature proposed here can describe the differences between the contours of different objects very well. At last, a linear SVM which is trained based on certain positive and negative samples serves as the classifier. Experiment shows that the proposed algorithm can get good result in human detection task.

相关论文

  • 暂无!

相关知识点

  • 暂无!

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号