Research On Key Technologies Of Visual Object Tracking System

Visual object detection and tracking is one of the important research in the field of machine vision,through the analysis of video image,obtain the measured object space position,motion。This information is complete machine vision systems such as behavior analysis,and other high-level task,the basis of artificial intelligence is of great significance for the development of a higher level,in military weapons,aerospace,security monitoring,unmanned,and other fields has been widely used.In recent years,new technologies,such as electronics and communication,have become increasingly mature,providing new ideas for the design of machine vision object tracking system.After analyzing modern machine vision technology and many visual object tracking algorithms,this paper proposes a solution of moving object tracking task based on active vision theory and embedded technology,and completes the design,construction and test of tracking system.The main contents include the research and implementation of visual object tracking algorithm,the design and construction of active visual object tracking system,and the simulation analysis and experimental test of tracking system.Active visual object tracking system is composed of host computer and embedded tracking device.Raspberry Pi 4B was selected as the controller for the object tracking device,and real-time video images were acquired through the phase machine module.After the image was analyzed by visual detection and tracking algorithm,the object position information was obtained,and then the servo motion system was controlled to adjust the camera attitude to realize the dynamic object tracking.The device communicates with the upper computer via LAN,and completes tasks with high computational complexity,such as data analysis and processing,model training and updating,etc.According to the characteristics of embedded object tracking system,the software and algorithm are designed.On visual inspection of mainstream tracking algorithm research,on the basis of combined detection and tracking algorithm,using cascade classifier achieve object detection,using correlation filter for object tracking,under the tracking of the main program control,and real-time image acquisition procedures,match the servo motor control procedures,realizing the independent of target detection and tracking.Finally,this paper tests and analyzes the designed object tracking system and tracking algorithm.The visual object tracking algorithm designed in this paper is simulated,compared and analyzed on different computing platforms from the aspects of object model training effect,object detection and tracking accuracy,algorithm running speed and so on.Taking the author as the tracking object,the author designed an experiment to test the tracking effect of the moving object tracking system,and verified the feasibility of the design in this paper.