| Machine vision has been widely used in assembly location,product quality insp ection,object recognition and other fields because of its advantages of quick access,non-contact and high precision.In order to promote the intelligent development of radar assembly,solve the problems of low assembly efficiency,slow detection speed and low accuracy caused by manual assembly,a scheme based on machine vision to guide the assembly anddetection of radiation units of robotic manipulators is proposed to replace the current manual work.The ma in work is as follows:(1)Camera calibration.Build the camera imaging process with camera distortion.Acquiring the internal and external parameters of camera by Zhang Zhengyou checkerboard method,which helps determines the rotation and translation relations between the image pixel coordinate system and the world coordinate system.(2)Research on the algorithm for the extraction of the center point of radial unit.The thinking of algorithm that extracts the center point is calculating by locating the straight line of the outer contour of the radial unit.Firstly,we proved that,in industrial environment with lots of interference noise,Hough transform algorithm cannot extracts contour line of radial units stably.Then,an algorithm based on gray scale difference and polarity information is designed to pr ecisely fit the contour of the radial unit,and so as to reduce the running time and computational complexity of the algorithm,the algorithm initially carries on template matching of radial units based on shape information to preliminary positioning the u nit,it has reduced the ROI.At last,a number of radial units are taken for central point extraction experiment,and the feasibility of the algorithm is verified.(3)Simulating the configuration of radial units based on assembly accuracy.By analyzing the source of deviation and the way of deviation transmitting about radial unit,an algorithm is compiled to find the optimal arrangement for a particular set of radial units so as to minimize the assembly error between the radial units.Finally,a three axis platform driven by servo motor is built to carry out the installation experiment of radial unit.The feasibility and stability of the above image processing algorithms and simulation algorithms are verified from the point of view about real-time and accuracy.All about this paves the way for subsequent automated assembly of radar. |
