基于正面视角的步态识别
首发时间:2010-04-23
摘要:现有的步态识别算法研究几乎全都是基于侧面步态的。本文提出一种基于正面视角的步态识别方法。首先归一化RGB颜色空间被用来检测和去除阴影并用背景减除法提取二值化人体轮廓。提出一种专门适用于正面步态的周期检测方法,提取周期关键帧后跟踪轮廓线并用改进的等角度采样法进行采样以减少计算量。简单高效的傅里叶描述子被用来提取特征向量,进行数据降维后构造步态模板。用最近邻和最近邻标本分类器分别进行分类。在CASIA数据库上的实验表明该算法不仅具有较低的计算量而且表现出较好的识别性能。
关键词: 步态识别 正面视角 去除阴影 傅里叶描述子 步态模板
For information in English, please click here
Front-view Gait Recognition
Abstract:Existing gait recognition methods are almost based on side view sequences, while in the paper a new method for front-view gait recognition is proposed. Firstly, a method of normalized RGB color space is used for shadow removal and binary silhouettes are extracted by background subtraction. Then a special method for cyclic gait analysis is performed to extract key frames, to whose body contours an improved sampling method is applied. Next, the sample points are processed by Fourier descriptors with data dimension being reduced and the gait exemplar vectors can be constructed. Finally, the recognition is achieved separately by NN (the nearest neighbor classifier) and ENN (the nearest neighbor classifier with respect to exemplar). Experimental results show that the proposed approach is not only computing efficient, but also has an encouraging recognition performance on CASIA database.
Keywords: gait recognition front-view shadow removal Fourier descriptors gait exemplar
论文图表:
引用
No.4228952399512719****
同行评议
共计0人参与
勘误表
基于正面视角的步态识别
评论
全部评论0/1000