Using Mean - Shift Algorithm Of Intelligent Tracking Control System

Target tracking technology is the key step to implement the intelligent control system,which combines the research results of many subject discipline such as image processing, pattern recognition, automatic control and so on, and uses the key technology to process and analysis the input image sequence, in order to achieve the target detecting and tracking. Target tracking can be divided into static background target tracking and dynamic background target tracking. Though static background target tracking has many mature theroy, tracking target under the dynamic background still has many problems. For example, it is difficult to achieve a balance between the positioning accuracy and real-time.Fortunately, an algorithm called mean-shift provides a new method to tackle this conflict, because the algorithm is a target search algorithm based on nonparametric density estimation, which uses the gradient steepest descent method to achieve the optimal position of the target iteratively, result of small amount of calculation; and this algorithm is not sensitive to the moving background, so it can be applied to the occasion of the moving background. This dissertation focuses on the mean-shift algorithm, and makes a number of improvements for its shortcomings, mainly includes the following aspects:(l)In-depth study of the nonparameteric density estimation, kernel function, gradient estimation and the choice of weighting function, which lays the foundation for improvement of the algorithm;(2)Aim at the problem caused by the fixed search window, this paper provides a method based on constant moment to adjust the search window adaptively, and uses Kalman filter to predict the size of the window, which makes the positioning more precision;(3)Algorithm will fail quickly when the target blocks, Kalman filter algorithm integration with the mean-shift algorithm is proposed to search the target, and Kalman filter residuals, target area and Bhattacharyya coefficient is combined to make judgment on the target block, which can effective track the block target; (4)Study the stepper motor technology and design the hardware driver circuit.In the side of automatic control, ARM embedded system combined with the stepper motor controls the camera to track the target that will leave the surveillance of the camera. And the stepper motor driver technology is studied to make the motor rotate smoothly, which can serve the tracking system better.