基于BP神经网络的人体行为识别

Human action recognition based on back propagation neural network

针对人体行为识别问题,提出一种基于径向基函数(BP)神经网络的人体行为分类算法。首先,利用奇异值分解(SVD)算法提取视频每一帧的奇异值,将每一帧的奇异值按照行拼接起来即为一个视频的样本,样本按照行排成样本矩阵;然后,利用主成分分析(PCA)对得到的矩阵进行去相关并且降低维数,降低维数的矩阵再进行线性鉴别分析(LDA),使样本变得线性可分;最后,利用BP神经网络对样本进行分类;实验结果表明,与采用最近邻分类和K近邻分类(k NN)相比,所提算法具有更高的识别率。

For human action recognition, an algorithm for classifying human action based on back propagation（BP） neural network was proposed. Firstly, singular value decomposition（SVD） was used to extract the singular value of each frame of the video. Then each row of a matrix was composed of the singular value of each video. Every single row of the matrix is a sample of human action. Secondly, the principle component analysis（PCA） algorithm was proposed to remove correlation and reduce dimension. Then the linear discriminant analysis（LDA） algorithm was applied to matrix processed by PCA to make samples linearly separable. Finally, the back propagation neural network was used as a classifier. The experimental results show that the proposed method, compared with nearest neighbor classifier and K-nearest neighbor（k NN） classifier, has a higher recognition rate.