摘要
目前,在步态识别技术中多数描述步态特征的方法在非侧面视角下识别效果一般都不够理想,通常会明显低于侧面视角,针对这一问题,文章提出一种以角度特征分量特征作为步态特征的识别方法,提高步态特征的分类能力从而提高识别率。在步态检测部分文章采用基于色度坐标的混合高斯来抑制阴影和消除噪声,步态识别部分使用支持向量机对所提取的角度特征分量特征进行训练和分类,最终在保证侧面视角识别率的情况下同时提高在非侧面视角下的识别效果。
At present,most gait feathers can not work very well but lateral angles of view in gait recognition technology,even worse obviously than other views usually.In order to solve the problem,a new method gait recognition based on components of angle feature is proposed in this paper to improve the classification capacity of gait feature.The paper adopts GMM based on chromaticity coordinates in gait detection and uses SVM to train and recognize gait by the components of angle feature.So while it will insure the recognition rate in lateral view,it will also improve it in other views.
出处
《计算机与数字工程》
2010年第3期135-138,共4页
Computer & Digital Engineering
