Design And Implementation Of Active Vision System Based On ARM Embedded Platform

In recent years,with the rapid development of machine vision,intelligent monitoring and other fields,the technology related to active vision is becoming more and more mature.Aiming at the technology of target detection and tracking based on video sequence,this paper proposes an active vision system scheme based on ARM.The main work and results of this article are as follows:In the aspect of target detection,two kinds of typical moving target detection algorithms are studied.It includes the frame difference method and the background subtraction method.And the simulation experiment is carried out.The effect of background subtraction depends on the construction of the background,and the inter frame difference method can detect the moving target without the need of the background information.In target tracking,aiming at the situation of interference with similar color background,an improved Camshift algorithm is proposed,which combines FAST feature point detection and FREAK descriptor to modify the tracking results of Camshift algorithm.The experiment shows that the improved algorithm can effectively deal with this kind of interference and improve the tracking accuracy of Camshift algorithm.In addition,under the condition of low illumination,the color characteristics are not obvious,so the KCF algorithm is used to track.Aiming at the situation that the tracking target part is blocked,an improved KCF algorithm is proposed,which corrects the tracking error of KCF algorithm through FAST feature points and FREAK descriptors.The simulation experiments show that the improved algorithm has good tracking results for the partial occlusion problem.In the aspect of embedded hardware implementation,an embedded active vision platform with S5PV210 as the core and STM32 as a motion control module is built.In the visual aspect,the moving target is detected by the frame difference method,and the algorithm is tracked by the algorithm proposed in this paper.The transplant of embedded Linux system,OpenCV library and QT library has been completed.It mainly includes the design of human-machine interaction interface,image acquisition part based on V4L2 protocol,transplant of improved Camshift algorithm,transplant of improved KCF algorithm and so on.In terms of motion control,a wheeled vehicle based on STM32 is built,and the wheel type trolley control strategy is designed to track the moving target.Through the research of the algorithm of the active vision system and the transplantation of ARM.The experimental results show that the active vision system can detect and track moving objects accurately in a certain environment,and achieve the desired effect.