文章基本信息

标题：Cyclic HMM-based Method for Pathological Gait Recognition from Side View Gait Video
本地全文：下载
作者：Hoang Xuan Hai ; Hoang Le Uyen Thuc
期刊名称：International Journal of Advanced Research in Computer Engineering & Technology (IJARCET)
印刷版ISSN：2278-1323
出版年度：2015
卷号：4
期号：5
页码：2171-2176
出版社：Shri Pannalal Research Institute of Technolgy
摘要：The analysis of human gait has become a popular and dynamic research on computer vision. An important application of gait analysis is to early detect the gait abnormalities, which may be caused by some diseases. In this paper, we present a video-based method to recognize common pathological gaits of the elderly such as ataxic, hemiplegic, limping, neuropathic, and Parkinson. There are three main processing steps in the system. In the first step, from each frame of the side view gait video sequence, we first separate the walking person from the background using adaptive background Gaussian Mixture Model. In the second step, we convert the extracted object into a seven-dimensional feature vector based on Hu's moments. In the final step, we analyze those extracted features to recognize different abnormal gaits using Cyclic Hidden Markov Model (CHMM). The Cyclic HMMs are trained with different values of parameters in order to achieve the best reliable recognition model for each disease. Experimental results on simulated abnormal gait database indicate the good performance of the proposed method in terms of low complexity and high recognition rate (about 83% for recognition and 93% for detection task).
关键词：Gait analysis; Pathological gait recognition; ; Gaussian Mixture Model (GMM); Hu's moments; Cyclic ; Hidden Markov Model (CHMM).