Font Size: a A A

Study And Implement On Key Technology Of Human Motion Detection And Abnormal Behavior Recognition

Posted on:2010-11-08Degree:MasterType:Thesis
Country:ChinaCandidate:S C FanFull Text:PDF
GTID:2178330332988562Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Human motion detection and abnormal behavior recognition is one of new arising high-tech application fields and front topics in Computer Vision. It spans many subjects including computer science, machine vision, image engineering, pattern recognition, artificial intelligence, etc. Human motion detection and abnormal behavior recognition can automatically analyze images in video sequence by the methods of computer vision and video analysis without artificial interference. This system can real-time detect, track and recognize the moving human body in special environments. Furthermore, it also can analyze and judge whether the behavior of human body is abnormal and send alarm to guide the action and make the corresponding decision if necessary.Motion objects detection, tracking and behavior recognition are three core modules of this system. This paper introduces the research status and tendency of human motion analysis in the world. On this foundation, the work and the main contributions of this paper are as follows:The algorithms of camera-fixed motion objects detection are studied in this paper. In the condition of simple background, a background updating algorithm combined with adaptive threshold selection which improves the performance of the detection algorithm by limiting the range of the threshold is proposed. In complex background, the moving object is extracted based on adaptive Gaussian Mixture Model which can reduce the outside influences such as lighting changes, trees swaying and so on.In moving object tracking module, region fusion algorithm that obtains the minimum enclosing rectangle of moving object is proposed to select the center and the speed of the object as a state vector to track the moving object.In behavior recognition module, template matching method for human activities recognition is adopted. This paper proposes a new method that combines Hu Invariant Moment with other image features to describe the moving objects which improves the rate of human abnormal behavior recognition in the experiment and realizes human abnormal behavior detection system.
Keywords/Search Tags:moving object detection, Gaussian Mixture Model, abnormal behavior recognition, Intelligent Monitor
PDF Full Text Request
Related items