Research On Moving Target Detection And TrackingAlgorithm In Intelligent Monitoring System

With the continuous development of science and technology, the improving of the intelligent monitoring technology, the application scope and demand is growing. Compared with traditional monitoring system, the function of a new generation of intelligent monitoring system is increasingly to be perfectly, can realize all-day, uninterrupted, low false real-time monitoring. Among them, target detection, tracking and extracting technology are the core issue of foreign and domestic scholars, they have great research value and broad prospects for development.The paper focus on target detection and tracking in intelligent monitoring system, experiment research is proceeded under the static state of monitoring background. Through the comparison and analysis of existing detection and tracking algorithm, combine with the advantages and disadvantages of these algorithms, put forward the improved algorithm. Adopt the method of theoretical analysis and experimental simulation, proved the feasibility of the improved algorithm, complete the algorithm optimization. In the part of target detection, adopts improved three frame difference, background subtraction division combine with the gaussian mixture model to realize the goal.By analyzing the forming principle of shadow, compare with the results of shadow detection under different color space, put forward a kind of shadow detection method combine texture feature and color space model, removal the shaded area by simulation experiment, to make the target detection more accurate. In the part of target tracking, adopts the method combination with color space, texture characteristics, Mean-Shift and Kalman filter, to achieve the goal. By kalman filter and similarity judgement to updated target model in real-time, the algorithm real-time performance and robustness has been optimized. When the target has been kept out, we use the Kalman filter to predict the target effectively, reduce the times of operation. When background and target color are similarly, reduce the risk of failure tracking, improved the effectiveness and reliability of target tracking.