The Design And Realization Of Moving Object Detection And Tracking Algorithm In Near-shore

Our country is a maritime country, which has long coastline and numerous islands,coastal defense and security is in a very important position in the national securityagency. However, the present sea defense ability is not accustomed to the actual state ofthe national security. Therefore, establishing a coast defense visualization of remotesurveillance system is a kind of effective means to solve the current island management.In order to solve the difficulty of traditional video surveillance system to deal withcomplicated surveillance scene and behavior problems, it is necessary to design a set ofvideo surveillance system based on moving object tracking technology (behavioranalysis), simultaneously, Object detection and tracking technology is the core of thevideo surveillance system. This paper studies the technology of the near-shore movingobject detection and tracking in detail.According to the characteristics of near-shore moving object image (Its occupiedarea is more larger and the resolution is more higher) and the dynamic complex scene ofnear-shore sea surface, by combining with the advantages of inter-frame difference andbackground difference, this paper designs a hybrid difference method to realize thedetection of the near-shore moving object. First of all, by using the method of threeframe difference for the three consecutive frame image in video image sequences, thetracking system can obtain the movement pixels and background pixels of the movingobject; Secondly, to get the moving regions of the object, the tracking system usesbackground difference just for the area of the movement pixels; Finally, to obtain thecompletely moving object, this paper uses the method of morphological processing forthe moving regions obtained by the above method.The near-shore moving object detection is the precondition and the necessary stepof the object tracking. At present, there have been many object tracking algorithms inImage Processing and Computer Vision research field, but due to the advantage of thealgorithm of the particle filter that can deal with the nonlinear and non-Gaussian system,therefore, the algorithm of the particle filter is widely used in object tracking fields. According to the problem of the sea moving object tracking, the changes of the seabackground are difficult to predict and the feature of moving object will change overtime, for example, the color information of moving object will change with the changein lighting environment, the edge information of moving object is vulnerably effectedby the background edge information. Therefore, just by using a single feature is hard torealize the robustness in object tracking. In order to improve the accuracy androbustness of object tracking as far as possible, this paper designs an algorithm of theparticle filter based on a fusion of color and edge feature, and it is applied in the seamoving object tracking. This paper introduced the algorithm of object tracking based onthe particle filter from the following several parts: object’s prior knowledge (color, edgefeature), the state transition and object observation model of the object tracking system.Then, by combining with the fusion strategy of several features, the paper realizes thesea moving object tracking based on the particle filter fusing color and edge feature.Finally, the paper gives the tracking results of the sea moving object and analyze to theabove tracking results.Finally, this paper designs a set of near-shore video surveillance system, whichmainly realizes the following functions: the function of reading the video data, thefunction of moving object detection, the function of moving object tracking, thefunction of drawing the motion trail of the object. Meanwhile, the function of movingobject detection and moving object tracking is the core of this video surveillancesystem.