The Research On Tracking And Counting Of Moving Ship Based On OpenCV

In order to create a good traffic environment, reduce the inland waterwaypollution and accidents, reaching the green shipping purposes, this paper hasconducted the research to the maritime surveillance. As the special environment ofinland waterway, this paper analysis and compare the improved algorithm and thetraditional algorithm to validate the superiority of improved algorithm in the specialcase of inland waterway.Among the algorithms of motion detection, this paper focuses on analyzing thethree classical target detection algorithms. In order to adapt to the dynamic change ofinland waterway, this paper analysis the Gauss mixture model background method.For the defects of slow convergence speed of background of the mixed Gaussbackground model method, So we proposes a ship motion detection algorithm basedon three frame differencing and Gauss mixture model. Firstly, we use the three framedifference method to detect the changes in the scene, Secondly, processing the Gaussbackground modeling improve the real-time of target detection, Finally, the algorithmis easily affected by illumination mutation or flow volatility. At the end extract theforeground dealt with the area threshold method for filtering the interferencedisruptors. The experimental results show that the improved algorithm. Theexperimental results show that the improved algorithm can better adapt to thedynamic changes of the scene, and has better performance in real-time.Among the algorithms of tracking of moving target, Firstly, we learn about thethought and basic principle of the Mean Shift algorithm, focuses on analyzing theCamShift target tracking algorithm, but this The algorithm is a semi-automatic targettracking algorithm based on the color information. Although this algorithm can adjustthe search area adaptively and track the moving targets accurately, but this methoddoes not describe the moving target accurately only based on color feature, especiallywhen the color of the moving target and background are closely, it will produce theserious deviation phenomenon. To solve the above problems, we proposed a newCam Shift algorithm based on the multi-feature information and we use of color,texture and edge three features instead of only the color feature to describe the target. To solve the defects of the feature weights unreasonable, we proposed an improvedalgorithm to distribution the weighted coefficient according to the contribution ofeach feature and describe the moving target more accurately. Finally we combined themoving target detection and the improved tracking target algorithm; realize theautomatic tracking of moving ship.In the motion of the ship traffic statistics, Firstly, we study the virtual linedetection method, virtual line and the target chain method; consider the inlandwaterway specific environment, the real-time and the accuracy comprehensively, WeFind the virtual line counting method has the better performance in ship trafficstatistics.At the end, combined with the VC++6.0software environment and opensource computer vision library, and implement the algorithm involves this paper, then,compare and analyze the experimental results. In the detection of ship motion, theimproved algorithm has better robustness and real-time features, and adapt to thedynamic changes in the scene. In the terms of moving ship tracking, the improvedalgorithm can improve the accuracy and stability in moving target tracking. In thetraffic statistics of the moving ship, the virtual line method can count the shipaccurately. After experimental test, the counting ship error rate is less than10%.