Research On Detection Algorithm Based On Human Behavior

The research of human behavior detection algorithm has been the focus of computer vision and pattern recognition in recent years.The research in this field is of great significance to accelerate social development and improve people's quality of life in the era of artificial intelligence.Many application fields,such as intelligent monitoring system,smart city and human-computer interaction,cannot be separated from the study of human behavior.There are many aspects in the research of human behavior detection algorithm.This paper selects two representative aspects to study,namely pedestrian detection algorithm and interactive behavior recognition algorithm.In this paper,dynamic and static pedestrian and human behavior detection and identification classification,the main work of the paper is as follows:This paper starts from pedestrian detection and human behavior recognition respectively,this paper expounds the research background,significance and the development status of the research content at home and abroad,and indicates the increasingly important status of human behavior research in the field of computer vision.It discusses various mainstream research directions,and obtains the main content of this paper through analysis.This paper briefly introduces the feature extraction method and feature descriptor of pedestrian detection and behavior recognition,and introduces the algorithm and implementation process used in this paper.The algorithm used in the pre-processing stage of pedestrian detection is mainly Gaussian mixture model.The Gaussian mixture model is used to obtain the area of interest,and then pedestrian detection is carried out for the area of interest.The gradient direction histogram feature is used in pedestrian feature extraction,the dimensionality reduction is processed by principal component analysis,and the current mainstream classifier is used in classification SVM The experimental results are obtained by using the Open CV file video sequence and INIRA pedestrian dataset.The low rank thinning algorithm based on Sc SPM and RPCA is used in human behavior recognition stage.The texture feature,optical flow feature and edge feature of human RGB image are extracted in the early stage.The static feature is weighted fusion,then the static fusion feature is encoded and fused with the dynamic optical flow feature by Sc SPM model.Then the low rank matrix of fusion feature is extracted by RPCA algorithm,and finally the human behavior is classified by linear SVM classifier.The experimental verification is carried out on the data set CAD-60 and MSR Action Pairs,and the confusion matrix of the experimental results is obtained.The above two experiments show that the two experimental algorithms in this paper have achieved good experimental results in pedestrian detection and human behavior recognition,and prove that the experimental algorithm is feasible.