Detection Of Moving Objects With Complicated Conditions In Video

Nowadays, with the development of high technology, intelligent surveillance systems areused more and more widely. As the core technology in the intelligence surveillance, movingobjects analysis based on video consists of the moving objects detection and retrieval, actionidentification and analysis, the object classification, incident detection, and so on. Thedetection and retrieval of moving objects is the foundation and key of it. In computer visioncommunity, the higher levels such as moving object classification, tracking and behaviorunderstanding heavily depend on the results of moving object detection. Moving objectdetection is one of the fundamental tasks for video image analysis, which active topic widelyis researched in the last decade around the world.Although moving objects detection has been widely researched and many effectivealgorithms have also been proposed, there are still a lot of difficulties in moving objectsdetection in complex environments. In this dissertation, we make research on how to build abackground model based on gauss in complex environments and is focused on moving objectsdetection combined background subtraction with frame subtraction.Some common approaches in the field of object detecting are introduced briefly,including frame subtraction, background subtraction and optical flow in object detection. Allthose algorithms are analyzed and their advantages and disadvantages are concluded based onthe result of experiment.The main contributions in this paper are as follows:1. Several background modelingalgorithms are widely researched and analyzed, and the methods are compared in this paper.2.Gaussian background modeling including Single Gaussian background, mixture gaussianmodeling method, mixture gaussian model method based on block.3.A new mixture Gaussian-based algorithm is proposed. The improved mixture Gaussian-based background model isbased on block and fused with frame subtraction method.4.Experimental results show that thealgorithm in this paper improve the speed and detect moving object accurately. As the core technology in the intelligence surveillance, based on video, moving objectdetecting is more widely. The reason why the more and more people do more research worksis exactly that.