基于加权局部判别CCA的多视角步态识别方法

Multi-view gait recognition method based on weighted local discriminant canonical correlation analysis

为了解决步态特征提取难题和克服单一视觉的步态进行身份识别方法的局限性,提出一种加权局部判别典型相关分析(WLDCCA)算法。在此基础上,提出一种基于WLDCCA的多视角步态识别方法。该方法通过在WLDCCA中引入样本的类信息和局部信息,将不同视觉的步态特征有机地融合起来,提取的融合特征能够最小化同类样本之间的距离,同时最大化异类样本之间的距离,提高了步态识别的识别率和鲁棒性。在CASIA步态数据库上的实验结果验证了该算法的有效性和可行性。

To solve the problem of gait feature extraction and overcome the influence of single observation angle on gait recognition rate,a Weighted Local Discriminent Canonical Correlation Analysis(WLDCCA)algorithm is proposed and then a multi-view gait recognition method is proposed.The main characteristic of the proposed method is that the class information and local information is introduced into WLDCCA algorithm to effectively fuse the gait features of different views.The extracted fusion features can maximize the distance between intra-class samples and minimize the distance between inter-class samples at the same time,and then the recognition rate is greatly improved.The experimental results on the CASIA gait database verify the effectiveness and feasibility of the proposed algorithm.