Target Detection And Tracking System Design Based On Dynamic Context

With the rapid development of science and technology, more and more scholars want to have breakthrough in target detection and tracking for a video image as a hot research direction of the computer vision. People hope to be able to be done automatically by the camera movement on the target tracking out of human control. This thesis research detection algorithm, tracking algorithm, as well as the control of the stepper motor for target detection and tracking system based on dynamic context. Finally, complete the real-time target tracking system.A dynamic background target detection algorithm based on mutual information was proposed to improve the non-adaptive for dynamic background of the traditional target detection algorithm. This algorithm combined the concept of mutual information between consecutive images successfully. First, the image of once Mallat wavelet decomposition which is the approximate component, Pso and Powell hybrid optimization strategy was used to look for the maximum mutual information value of the image, then completed image matching based on space conversion model, bilinear interpolation at the same time to eliminate the impact of changes in background. Finally, used improved three difference method completing the target detection and proved the accuracy and feasibility of the algorithm through experiment.Target tracking include targeting and tracking, selected central method to locate the target. In this paper, the method of least squares was used to locate the position of the moving target and output trajectory of the target, lay the foundation to the target tracking. Selected the improvement of the Mean-shift algorithm, Camshift algorithm combined with OpenCV to prove the adaptability and accuracy of the Camshift algorithm tracking in different circumstances.For the whole tracking system, the thesis also focused on the stepper motor control algorithm, fractional thought was introduced to do Simulink simulation experiment to prove the superiority of the fractional PIλDμ control algorithm in regulation time and control accuracy.End of thesis, the connection of hardware facilities and debugging, PTZ movement control, parameters calculated, software interface of the target real-time tracking system were completed. Finally the target tracking system of the horizontal movement was completed successfully.