Research On Pedestrian Movement Trajectory Recognition Based On Machine Vision

In today’s society, the development of pedestrian motion analysis has become increasingly sophisticated.It also brought a lot of convenience for people’s life.It’s a new and promising area in the future.Some related areas such as pedestrian detection, object recognition,tracking also have great prospects for development.How to establish a set of efficient and general system,will be relation to whether it can get more extensive application.In this paper, on the basis of previous research, we improve some algorithms to achieve better operational results. An improved frame difference method combined with multi-feature cascade AdaBoost classifier is putting forward to detect and identify pedestrian in the paper. Particle filter is used to predict pedestrian’s movement.Then use the AdaBoost classifier to confirm the prediction result.Study of this paper is as follows:(1)The pedestrian detection is studied.Frame difference is the key research.Through the improved frame difference to complete the pedestrian detection after processing of image.The accuracy of the algorithm is improved..By extracting the motion areas of the frame difference’s result to reduce the scope of the subsequent operation, in order to improve the accuracy of the algorithm.(2)The recognition algorithm is studied.A detailed study of boosting algorithm has been down in this paper. By comparing the performance of various operators,a Haar features and edgelet features cascade AdaBoost classifier is used in the paper.A dynamic weight is uesd to improve the accuracy of the algorithm.(3)Tracking of moving object is studied.Particle filter has been studied heavily.Particle filter is improved in the paper for pedestrian tracking.Then ues the cascade AdaBoost classifier to confirm the prediction result. The accuracy of the algorithm has improved.(4)Writting a program to achieve the algorithm.Running the program on OpenCV. Showing the experimental results to find that the accuracy and speed of detection, identification and tracking are improved.By the study and experiment in the paper, a pedestrian movement analysis system has been formed to complete the pedestrian detection,identification and tracking.The results show that the algorithm can be applied.