Design And Implementation Of Target Detection And Tracking System

In recent years,with the rapid development and popularization of video surveillance systems,there has been an explosive growth in image and video data.It has become a research hotspot to implement detection and real-time tracking of captured images and video data.This paper mainly studies the design and implementation of target detection and tracking system based on Camera Link interface.The main content includes the following aspects:1.A target detection and tracking system architecture based on image acquisition was designed.The system shoots video through high-definition industrial cameras and transmits it to the PC via an image capture card.The upper software implements efficient image acquisition,target detection and real-time tracking,and image display.The system mainly includes three modules: image acquisition,image processing,and image display.The image processing module consists of three core algorithms: image preprocessing,target detection,and target tracking.2.Image preprocessing,target detection and tracking algorithms are studied.First of all,before the target detection,the weight graying algorithm and the median filter algorithm are used to preprocess the image to improve the detectability of the target.Secondly,the target detection algorithm is compared and analyzed,and the initial target is detected using the Speeded Up Robust Features(SURF)detection algorithm.This algorithm is robust to background motion and scale changes.Finally,the target tracking algorithm based on detection and the target tracking algorithm based on correlation filtering are analyzed.Kernelized Corretion Filters(KCF)with fast speed and high precision are selected as the core algorithm of the target tracking module.3.In view of the shortcomings of the KCF algorithm applied directly in the system,this paper proposes an improved tracking algorithm.There are two main problems in the traditional KCF algorithm.First,there is no redetection mechanism when the target is occluded or lost during the tracking process.As a result,the system cannot continue to track the target.Second,the tracking frame scale cannot achieve adaptive estimation of the target scale,leading to background.The introduction of information or the loss of foreground information,the target model will accumulate error information and improve the tracking failure rate.To solve the problem of target occlusion or loss,this paper proposes a target tracking algorithm based on SURF feature detection.According to the problem of scale change,a target tracking algorithm based on multi-scale estimation strategy is proposed.Tests are performed on the Object Tracking Benchmark(OTB)data set.The experimental results show that the improved algorithm is more robust to target occlusion and scale changes than the KCF algorithm,and the tracking success rate is greatly improved and can be achieved Real-time tracking.4.This article uses Microsoft Foundation Classes(MFC)to build a target detection and tracking system based on image acquisition.First of all,the data acquisition and the layout of the interactive interface are realized.Secondly,image preprocessing,target detection and tracking algorithm are implemented.Finally,the system performance was verified.Real-time video capture through industrial cameras,the data obtained by the acquisition card transmitted to the host computer,the user through the upper software interface configuration parameters,the ultimate goal of video data detection and tracking.Experimental results show that the system can achieve efficient image acquisition,target detection and tracking,and can achieve real-time tracking for low-resolution video,which has a certain practical application prospects.