机器人轴孔装配系统中的视觉目标识别与定位-机械电子工程专业论文.docxVIP

西安理 西安理工大学硕士学位论文 II II I I Abstract Title：TARGET RECOGNITION AND LOCATION BASED ON VISUAL IN THE ROBOT ASSEMBLY SYSTEM OF SHAFT AND HOLE ? Major：Mechatronic Engineering Name：Kai Zhang Signature: Supervisor：Prof.Weiping Fu Signature: Abstract With the rapid development of computer science, it is possible to realize intelligence in the field of mechanical manufacturing and processing. The teaching robot of mechanical assembly has a trend to intelligence, and the machine vision is an important approach to realize intelligence. In this paper, robot of shaft- hole assembly is be taken as an object to study the problem of target identification and positioning based on visual. First, we achieve binocular PTZ camera and monocular hand-eye camera calibration. By calibration robotic PTZ cameras and hand-eye binocular monocular camera, we take the parameters of each camera; we also compared the results with OpenCV and Matlab, both the maximum relative error is less than 0.6%. Last, the causes of error are analyzed. Secondly, the paper puts forward the method to obtain the pose information of shaft-hole targets. By the main image processing, including image threshold and contour matching, the complex background is removed in the work scene images, and the shaft/hole is extracted effectively. In order to get the centroid and generatrix of shaft, we use the method of the minimum circumscribed rectangle and Hough straight line detection. Using the least squares method to find the target shaft/hole space pose in the camera coordinate system. The experiment of indirect measurement method show that the final calculated errors of space point within 1 mm, and axis attitude within 1.4 °. Lastly, this paper researches target position of virtual reality technology based on Marker. Using the calculated target position and orientation combined with OpenGL 3D graphics tool to create a virtual model under Marker coordinates, and transforms the coordinates to camera coordinate system. Thro