Research And Application Of The Moving Target Detection Algorithm Based On Video Image

Now days, Intelligent surveillance technology has been widely used in people’s life and work, This technology includes many key steps involved moving target detection, tracking, behavior recognition, etc. Among of these steps, moving target detection is the basis and the key for intelligent monitoring equipment capable of monitoring. So it is very necessary to study the algorithm of moving target detection.There are many algorithms about moving target detection currently, some classic algorithms such as Frame-difference method、Optical flow method and Background subtraction method. But these methods have some shortcomings. The method of Background subtraction is sensitive to the change of illumination, the Optical flow method had a large computation and the Frame-difference could not extract complete information of moving target. How to overcome the above drawbacks to achieve the intelligent monitoring in kinds of scene is a important question which more people concern about. This paper was mainly about the study of moving target algorithm, the main work in this paper are the following:1.This paper have researched and analyzed the methods of moving target、 tracking and image preprocessing, and proposed a new detection algorithm based on The three-frame difference method. Compare to other methods, experiments show that this method could extract the edge information of moving target availably and not sensitive to the change of illumination. Its principle was simple and easy to apply and could realize the real-time detection.2.Because of the Fall detection need extract the completely information of moving target, this paper combined the improved method of moving target and GMM method to extract the moving target, then use the new method in paper [52] to remove the shadow. Experiments show that this method could adapt the change of outdoor environment, meanwhile, it could extract the complete information of moving target.3.Studying the methods of fall detection recent years proposed and using the combined method in fall detection, then got the results of detection outdoor. Experiments show that the combined method this paper proposed could achieve the complete region of the moving target, based on the step above, according to some of standards of fall detection, it could realize the detection of falling behavior. So it show that this method is useful in the fall detection system.