Research On Human Action Recognition Algorithms Based On Depth Information

Human motion recognition is an important research direction in the field of computer vision.It has great application value in intelligent monitoring and advanced human-computer interaction scenarios.Traditional human motion recognition algorithms based on two-dimensional cameras are vulnerable to changes in light intensity and texture.Depth sensors have been introduced,and in particular,the well-performing and inexpensive Kinect series released by Microsoft has enabled depth-based research to be widely implemented.Information largely overcomes these problems based on twodimensional images.This paper introduces the research background and significance of human motion recognition technology based on depth information,introduces the research methods of human motion recognition algorithm based on depth information at home and abroad,and analyzes its advantages and disadvantages,as well as the commonly used depth-based information.The public data set was introduced.Then based on the depth information,a method of human motion recognition is proposed and optimized.The specific work is summarized as:(1)proposes a feature descriptor that can represent human motion based on the skeleton joint point data,which will not be affected by changes in perspective and scale.After quantifying the feature vector extracted from the action sequence,The quantified results are input to the classifier as observation sequences for identification.In this paper,Hidden Markov Model(HMM)is used as a classifier.Simulation experiments were performed on the MSR-Action3 D dataset and the UTD-MHAD dataset respectively,and good results were obtained.(2)The human motion recognition algorithm proposed in this paper is optimized,and the depth map is used to solve the noise interference a nd selfocclusion problems,so as to extract more reliable skeleton joint point data and further optimize the algorithm's recognition effect.Simulation experiments were also conducted on the MSR-Action3 D dataset and the UTDMHAD dataset,respectively,and the experimental results were analyzed.(3)A scheduling strategy is proposed to solve the partitioned task scheduling problem for multi-objective detection.This algorithm considers the energy consumption constraints and is named WALECC.Firstly,mathematical modeling of the problem and energy constraint conditions was carried out.Then the solution to the constraint condition was proposed based on the classical scheduling algorithm HEFT.Finally,the simulation experiment was performed and compared with the current representative algorithm to verify the feasibility and supriority of the algorithm.