基于嵌入式Linux的视觉伺服系统设计-《计算机技术与发展》

文章信息/Info

Author(s):: CHAO Yan-kai; XU Yu-fin; YANG Yong-huan; School of Mechatronics Engineering and Automation, Shanghai University

Keywords:: embedded visual servo; object recognition and tracking; OpenCV ; moment invariants

摘要:: 文中通过分析目前视觉伺服系统的实现方式，建立了一个基于嵌入式Linux的视觉伺服系统，用于室内相对简单背景下的运动目标识别与跟踪。文中首先完成了开源视觉库OpenCV到嵌入式Linux的移植，搭建了硬件平台、软件平台及交叉编译环境。基于OpenCV完成图像处理，采用图像的区域特征一不变矩作为特征信息，实现运动目标的形状识别。通过判断目标轮廓在图像中的像素偏移量，计算出机械臂末端的空间坐标值，调节机械臂各关节角度，通过不断逼近的方式，实现对运动目标的跟踪。实验证明，文中设计的视觉伺服系统能够实现设定的目标

Abstract:: Through the analysis of the present visual servo system implementation, it concerns the establishment of a visual servo system based on embedded Linux mainly used for recognition and tracking of the moving objects in a relatively simple background of the house. At first, studied the successful transplant of the open source visual library OpenCV to the embedded Linux and building of the hardware/ software platform and cross-compiling environment. The image of the moving object is processed via OpenCV. The shape recognition of the moving object is achieved according to the feature information of provincial characteristics of image-moment invariants. The spatial value of manipulator end is calculated by estimating the offset of the moving project＇ s profile, and then the angles of the mechanical arms＇ joints are adjusted accordingly. These conditions would be approximated gradually to achieve the tracking of the moving object. Experiments show that this visual servo system can realize the target