Font Size: a A A

Research And Implementation On The Key Technologies Of Intelligent Video Surveillance System

Posted on:2008-03-25Degree:MasterType:Thesis
Country:ChinaCandidate:Y H JiFull Text:PDF
GTID:2178360212973994Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
Intelligent Video Surveillance System (IVSS) is an emerging and challenging research content in the computer vision fields. The meaning of "intelligence" is that the IVSS can automatically analyze the image sequence and make a timely response to the exceptional objects and phenomena appeared in the watched scene contrasted with the traditional surveillance system. And it can work out a prediction of the incident according to the current state and remind the watcher to be prepared for it, which can reduce losses to the minimum. The IVSS overcomes the deficiencies that the traditional surveillance system requires a watcher to gaze at the monitor day and night and evidences are disposed passively after taking place according to the kinescope records, etc.. IVSS can assist and even replace people to accomplish the surveillance or control assignments by utilizing its primary predominance as a real-time, active, round-the-clock, preventive system. The burden of the watchmen is greatly lightened, as well as the cost of surveillance is largely lowered. Therefore, IVSS has a wide application prospect in security protection, traffic control, military security, fire alarm, crowd control and many other fields.At present, IVSS has made great progress, but there are still many problems in theory research and applications. Large numbers of researchers devoted themselves to the area and have already achieved many fruits. Based on the achievements, the dissertation studied some key technologies of IVSS. The main work can be summarized as follows:Firstly, in the aspect of moving object detection, considering the factors of "holes" resulted in easily by the method of temporal difference and bad practical capability of the optical flow method, the paper attaches very great importance to adaptive background subtraction for object detection. The paper validated one by one the Non-parametric model, W4 model, Single Gaussian model, Mixture Gaussian model by large numbers of experiments and compared their advantages and disadvantages in order to offer much more appropriate tactics for application. In order to enhance the stability and training efficiency of the background model, an improved Mixture Gaussian model is proposed. The experiments results prove that the improved algorithm...
Keywords/Search Tags:intelligent video surveillance, moving object detection, morphology, object classification, object tracking, mixture Gaussian model, Kalman filter
PDF Full Text Request
Related items