Font Size: a A A

Study On Moving Object Detection And Tracking Approach Based On Video Sequences

Posted on:2011-11-02Degree:MasterType:Thesis
Country:ChinaCandidate:L XiongFull Text:PDF
GTID:2178360308464699Subject:Traffic Information Engineering & Control
Abstract/Summary:PDF Full Text Request
Visual object tracking is a key problem in computer vision, and it is the basis of follow-up treatment, such as visual scene analysis, activity understanding and description. The technique of visual object tracking has a wide range of applications in intelligent video surveillance, traffic, human-computer interaction, visual navigation of robots, virtual reality, medical image processing, national defense, etc. The goal of visual target tracking is to imitate the motion sensibility of human vision, empower the machine with the ability of perceiving the target motion in the image sequence, and provide an important data source for visual analysis and understanding. Because of the complexity of the background environment, different light conditions, the complexity of different shapes and motion mode, occlusion and interference between objects, visual object tracking is still one of the most challenging tasks in computer vision.This paper mainly study on key modules of tracking framework such as object detection, object tracking and target recognition. This major research aspects involved include:1.In this paper, the section-distribution background model was proposed, using the appropriate background update mechanism, making video surveillance applications more adaptive for changing outdoor environment, meanwhile this approach could eliminate effects brought by the light intensity changes, the system disturbance or other reasons.2.Due to the traditional Kalman filter can not effectively track moving object, while the application of other higher robustness algorithms are complex and could not guarantee real-time performance, in order to consider both lower computational complexity and higher robustness of tracking, use linear subspace approach mapping a nonlinear tracking problem to piecewise linear space, and then use the Kalman filter to track, experiments prove its effectiveness.3.Take into account that the tracking algorithm always fail when dealing with complex scenes, this paper presents a general framework for tracking using context knowledge to aid tracking. Context of knowledge representation and reasoning decision-making tactics are generic, it can easily be used under different scenes, through setting the specific context information for particular scene and dynamic interaction between context module and other modules, can effectively deal with complex environments, such as occlusion by static object or interact between objects or caused by other factors.4.According to the statement of theory, this paper presents a general tracking framework and using experimental platform developed by VC++.NET and OpenCv library, combined with decision-making context information and reasoning, and experiments under different scenarios proved that the proposed methodology was reasonable.
Keywords/Search Tags:Moving Object Detection, Section-distribution Model, Object Tracking, Context Reasoning, Object Recognition
PDF Full Text Request
Related items