基于单目视频多特征融合的人头识别算法
首发时间:2009-03-02
摘要:提出一种采用固定单目垂直摄像视频序列的人头检测与识别算法。首先通过meanshift算法分割出人头部候选区域,在分析运动人头部区域的轮廓具有近似圆形以及人头顶部发色具有聚类性两个关键特征的基础上,提出结合候选区域内部颜色特征和边界轮廓特征的人头部目标区域识别算法。实验结果表明:提出的算法能有效地抑制光照的影响和消除与发色发布类似区域的伪目标。
关键词: meanshift 人头识别 特征融合 发色分布 头部轮廓
For information in English, please click here
Head detection based on multiple features merging in video sequences captured with single camera
Abstract:A head detection algorithm is proposed in video sequences vertically captured with fixed single camera. Firstly, meanshift--based image segmentation algorithm is induced to derive candidate head components in images. Furthermore, by referring to two features that the contour of human head regions is approximate round and the hair color distribution is clustered, algorithm combining the contour information and inside color information of candidate head components is presented to implement the task of head recognition. The experimental results show that the proposed algorithm is effective to reduce the light interfere and eliminate fake target whose color information is similar to hair color distribution.
Keywords: meanshift head detection feature merge hair color distribution head contour
基金:
论文图表:
引用
No.2979440551812359****
同行评议
共计0人参与
勘误表
基于单目视频多特征融合的人头识别算法
评论
全部评论0/1000