Indoor Abnormal Behavior Detection Based On Kinect

In recent decades,science and technology continue to improve and improve,the video surveillance system,especially the high definition system,has been popularized.Processing of HD video using computer vision technology,the technology is applied to the security monitoring field of computer vision,in order to improve the safety of the public,mainly through the detection of abnormal behavior of the public and the abnormal behavior appears to alarm people.Computer vision technology is widely used in the field of video surveillance,but also the need for efficient algorithms to solve real-time problems.In the field of computer vision,researchers constantly use the computer to identify and understand the "human behavior",from the perspective of target detection,target tracking and positioning,and finally understanding the behavior.Due to the influence of light,shadow,occlusion and noise,the understanding of the behavior of the video is also difficult.Thanks to Kinect,the depth of the image(RGB-D)into the people's attention,interference by K inect sensor is small,in the dark environment can also identify the target body,can obtain the skeletal features with spatial characteristics.Can be used for human behavior recognition,which led to a high degree of scientific research workers and stimulate new inspiration and ideas,based on the K inect platform for abnormal behavior detection.In this paper,the use o f Kinect devices to detect indoor abnormal behavior,RGB-D is to obtain the data information.In this paper,the abnormal behavior of the study is aimed at indoor scenes,which do not meet the expected behavior of people,usually abnormal behavior,including: fall,fight,chase,etc..And detect the abnormal alarm.This paper describes the algorithm and the features used by the three stages of human abnormal behavior detection,analysis of the advantages and disadvantages of these algorithms and features,a nd discusses the problems and difficulties of the research and the present situation,and analyzes the feasibility of using the image depth information and frame joint information.Secondly,it introduces the hardware and software architecture of K inect,and explains how to get the RGB-D information,and then describes the joint points of the skeleton.The joint information of the skeleton is extracted from the collected data,and the feature information is expressed by the joint angle information.Then,this paper introduces the mainstream human behavior recognition algorithm,this paper uses dynamic regularization algorithm to detect human behavior,and improve the algorithm to improve the operating efficiency.Finally,the paper summarizes the research work,and discusses the future work and development trend.