Pedestrian Tracking Algorithm Research Based On Agent Model

In today’s rapidly changing information age, lots of information makes people haveno time to take in all the information, it’s an inevitable requirement for the futuredevelopment trend to change the way of information focus. There is a big challenge forthe changing of information focus way: how to extract useful information from the vastamounts of data, especially the video data, thus arises at the historic moment, is"Intelligent video analysis technology". Intelligent video analysis technology is the keylink to make the video surveillance, the vehicle assistant driving as well as the intelligenttraffic management more intelligent. The key technology of intelligent video analysisincludes target detection, target tracking, behavior recognition and understanding.Target tracking is the foundation and key technology of intelligent video analysis, aswell as behavior analysis and behavior understanding, a good tracking result determinesthe success of subsequent research content. In this paper, aiming at the instantaneity andveracity of pedestrian tracking problem, we applied the least-square track predictiontechnology to target tracking, and in view of the problem that traditional tracking methodsoften tend to ignore that the state of pedestrian movement would be affected by thesurrounding environment and change the trajectory, only when the collision happened willthese methods adjust the tracking position. We make a certain depth of exploration of thepedestrian tracking for more intelligent, put forward a more accord with human thinking,more intelligent tracking method to improve the tracking algorithm, while combining theleast squares prediction and tracking algorithm, we fused a Intelligent obstacle avoidancemodel. This article main research content is as follows:(1) According to current research status of target tracking, mainly focused onlearning the Mean Shift tracking algorithm, programming algorithm and setup trackingframework, collected experimental data and verified the algorithm performance. Aiming atthe shortcomings of the Mean Shift algorithm tried to make an improvement.(2) Use least square for linear fit and predict the next place, regard the predictedposition as the next frame’s initial position for search, improved the instantaneity andveracity of Mean Shift algorithm. (3) Taken factors into account that the pedestrian movement would influenced by theobstacles in the surrounding environment and the pedestrian psychology and habit itself,the Intelligent obstacle avoidance model are constructed and can intelligently predict thetarget’s true location after change the path.Analysis shows that the method can make the tracking method conforms to the wayof human thinking which makes the tracking method more smart, it improved the trackingalgorithm’s accuracy and intelligence.