Research On Multi-feature Target Detection And Tracking Technology Based On Fixed Platform

In recent years,with the development of computer science,technology of object detection and tracking based on video images has become a research hotspot in the field of computer vision and image processing.Visual object detection and tracking technology plays an important role in video surveillance,UAV bombing,missile guidance,mobile robot search,visual navigation and so on.However,in practical application due to the complexity of the environment and the scene appearance in scale,shape and so on,which is very easy to change,the visual target detection and tracking technology is still facing lots of problems.After analyzing the above problems,this paper studies the target detection and tracking algorithm of RoboMasters robot competition scene,which is based on the RoboMasters robot competition.First of all,aiming at the problem of robot target detection both red and blue in the robot competition,the use of color threshold algorithm based on color feature of the target object in the scene of the enemy robot contest were detected,and using improved Roberts edge detection algorithm for shape feature extraction of target detection based on,the paper uses color threshold algorithm based on color feature of the target object to detect enemy robot contest and improved Roberts edge detection algorithm to extract shape feature extraction of target detection.Then,in order to solve the problem of multiple color regions of the same object,the fusion algorithm is used to fuse the different color regions of the same object.By the above two methods,the target object can be effectively detected from the video.Secondly,for the target tracking algorithm,the paper uses the CMT target tracking algorithm with high reliability.However,the original CMT target tracking algorithm can not meet the requirement of real-time performance,which is necessary during RoboMasters robot competition,because of the low processing speed caused by the global key points detection in each frame image.In order to improve the processing speed of the algorithm,the method of local key points detection is used to replace the global key points of the image,which greatly improves the processing speed of the algorithm.At the same time,in order to improve the precision of the local key points detection,based on the improved algorithm processing speed,the algorithm of the Kalman filtering algorithm is integrated in to predict the location information of the target.In addition,in the course of the competition,often the target object is blocked and "disappeared",leading to the failure.Aiming at this problem,the method of restarting detection process after the object disappear is so as to avoid the failure of tracking caused by occlusion of target.Compared with the original algorithm,the improved algorithmcan effectively improve the processing speed of the algorithm in the basis of ensuring the reliability of the tracking.Finally,the servo tracking system based on the fixed platform is designed,and the target detection and tracking algorithm is transplanted to the fixed platform.This paper adopts the method of serial communication transferring changed information of target position to the control center on the fixed platform servo tracking system,and at the same time in order to ensure the function of servo tracking system,double loop PID control strategy is used to control the PTZ motor is fixed on the platform,according to information received.