| Along with the development of information technology,the appearance of deep camera represented by Kinect makes it easier for people to obtain the human motion related features such as the coordinate position of the human skeleton,which also provides a new data source for the comparison and analysis of human motion.Action recognition is a kind of problem which focuses on identifying and classifying human daily movements,but the focus of this paper is to compare the human motion data,analyze the similarity and finally evaluate the quality of the motion,which involves more analysis of professional movements(such as dance,sports,etc).In other words,the motion analysis and evaluation result is given by comparing with the standard motion.The main research work of this paper includes:At first,the characteristics of Kinect acquisition of bone data are studied and analyzed,and according to the requirements of this paper,the Kinect device is used to collect and build the motion database,then the data results were analyzed and tagged,which is applied to the experiment test of motion comparison and analysis.In addition,in the aspect of comparison and analysis of human motion,the characteristics of variable and asynchrony motion sequence are combined.Firstly,the motion comparison and analysis method based on the Dynamic Time Warping(DTW)algorithm is introduced.On this basis,the motion comparison and analysis method based on the improved Longest Common Subsequence(LCSS)algorithm is proposed,which is used to solve the problem that the efficiency of the longer motion sequence of the DTW algorithm is low.Through the experiments and tests on Taichi-3D database and 3DLife/Huawei database,we compare the different motion characteristics and comparison analysis algorithms.In the end,on the basis of the method of motion comparison analysis,a software platform based on MFC framework platform for motion analysis and scoring is designed and implemented.According to different application requirements,we designed the real-time and off-line two platforms working mode,they can complete the function of motion similarity between the real-time acquisition motion sequence or the existing storage motion sequence and the set standard motion. |