Research on gait recognition based on K-means clustering fusion memory network algorithm

Medical research shows that human gait is unique, according to which human identity can be recognised. At present, convolutional neural network (CNN) is not stable enough in human gait classification, and most gait recognition methods based on contour or joint models have low accuracy. In order to solve this problem, this paper analyses gait features and deep learning technology, and proposes a gait recognition method based on K-means clustering algorithm and long short-term memory network (LSTM). According to the proportion of human height, the image after covering the wearing part is made by the network to extract the changing features of legs and the time dimension features of human gait cycle, so as to construct a new gait recognition model. Through experiments on the CASIA-B public dataset with multi-pose, multi-perspective and different coverage conditions, it is shown that the gait recognition model based on K-means clustering fusion memory network algorithm (KCFM) can significantly improve the clustering accuracy, can quickly adapt to the rapid changes in the planning trend, and has great advantages in mining the association of long-term series.