Research On Target Tracking And Recognition Technology Based On Information Fusion

The automatic tracking and recognition technology of dynamic target has many great advantages,such as getting a large number of target information,the strong ability to resist electronic interference,high measurement precision and large tracking field of view,so it is widely used in many fields including traffic management,bank monitoring,military border control and other security sensitive areas.This paper pedestrians are set as tracking and identifying target,by improving the algorithm of dynamic target recognition,the intelligentization and automation level of the monitoring system can be further enhanced,and the automatic tracking of the specific dynamic target in the key areas can be effectively achieved.In order to solve the problems of real-time,accuracy and anti-interference ability in automatic target tracking and recognition,this paper proposes the target tracking and recognition technology based on information fusion.Starting from the method and design of target tracking,by analyzing the principle of target tracking,this paper gives the design of target tracking platform,and researches the overall design method of automatic target tracking and recognition.Target detection is the precondition of image processing,considering the target imaging characteristics under slowly changing background,this paper researches the suitable target detection method and selects the mixed Gauss background modeling;in order to improve the real-time performance of target detection algorithm,proposes optimization scheme on adaptively adjusting the number of Gauss distribution,and extracts the RGB color weighted histogram and Sobel edge weighted histogram.Before the target automatic tracking,there is also a necessary link:target recognition,in order to ensure the accuracy of target recognition,this paper proposes the target recognition framework and method based on multi features information fusion;has a preliminary identification of the target through BP neural network;combining with the output of the network,makes use of D-S evidence theory for fusion decision;finally,the target fusion recognition effect of the system is verified by an example analysis.Target image tracking algorithm selected in this paper is particle filter,which is used to obtain the target image centroid pixel coordinates;this paper derives the implementation process of particle filter,analyzes the image features fusion method based on the fixed parameters,and gives the moving target tracking algorithm and modeling by image information fusion.According on the researches on target detection method,recognition method and tracking method,this paper through the platform for testing verification of the design theory,through the testing results,it can be seen from the experimental results that due to the influence of background change and occlusion,there is floating error in the pixel coordinates of the target center,and the target centroid pixel coordinates are basically locate in the effective field of view,the target tracking is relatively stable,the target automatic tracking and recognition system has better anti-interference ability.