Research Of Moving Object Detection And Tracking Techniques Based On OpenCV

Moving target detection and tracking is one of the main topics of computer vision,which is combined with image processing, pattern recognition, artificial intelligence,automatic control and other fields of advanced technology. It is used widely in videoearly-warning, secure monitoring, machine navigation, industrial automation, TVguidance and so on. Therefore, the research has the important theory significance andthe practical value. This paper focuses on moving target detection and trackingtechniques.It is used an improved Surendra algorithm in this paper. It acquisitiones thresholdadaptive by OTSU algorithm and obtaines the background update mask by framedifference method. It updates the background by the results and combines withmorphological dilation and erosion algorithm to obtain the foreground. It is used ancombination algorithm in this paper, which is based on the Kalman algorithm forprediction and CamShift algorithm for tracking. It is based on the foreground imagewith the moving targets and predictes the target location of the next frame by Kalmanalgorithm in this method.It determines the search window size and center cycle times bythe color features and the motion characteristics of moving target.The improve Surendra method can outlines the contours of moving objects reallyin the case of static background. It can be effective for a large position and velocity ofmoving target prediction by the algorithm combined with Kalman algorithm andCamShift algorithm. It’s good performance in the algorithm.The algorithm is good performance by experiments,in which the single frameaverage execution time achieves48.6ms, Dice coefficient reaches0.53, and thesuccessfully targeting number of the recall rate can achieve0.42.