| With the growing sophistication of MEMS,micro-assembly has also widely aroused people’s concern.In recent years,the visual technology gradually is mature,also,it has been widely used in the field of parts detection with the advantages of high speed and precision and non-contact measurement.Byusing the multithreadingtechnology,micro-assembly system we have deceloped canautomaticallystart to realize the automatic assembly of miniatureparts.The concrete research content in this paper is as follows:1.The key technology has been studied for the problem of micro-assembly of miniatureparts,at the same time,the status at home and abroad on the micro-assembly system in recent years was analyzed and summarized.A study on the key technology of image recognition and automatic assembly for miniature parts also has been reached in detail.2.Basing on modular design of micro-assembly system,XML file is called in which the coordinates of miniatureparts has been saved to realize the three steps strategy with adsorption,posture adjustment and assembly with the cooperation of motor,visual system and other modules,so automatic assembly has been realized.Hardware of the visual recognition system was also described in detail.3.Firstly,the principle of camera calibration and the method Zhang Zheng you’s calibration was studied.Then,the experiment has been designed based on the online calibration of OpenCV library on which the internal and external parameters of camera is reached by method of subpixelcornerdetection,at the same time,iterative algorithm was also used to calculate the camera distortion.Finally,the distortion was corrected by calculating the distortion map.4.To begin with,anisotropic bilateral filtering was used to smooth target parts.Then,the target area was extracted with the combination of threshold segmentation and area denoising,and NMI feature matching and shape template matching was used to recognize the target part.Finally,accurate positioning of target parts has been realized with the least squares fitting.5.Mico-assembly software system is composed of eight parts by class encapsulation mode and OO technology.Multi-thread was used to monitor the operation state of assembly,so automatic-assembly was realized by the interface function provided by each module,the information of configuration,recognition and localization.6.Basedontheanalysis of image recognition and assembly precision error source,the experiment of light source,adsorption force,position adjustment and assembly also has been performed.Experimental results show that the spacing error of lateral and vertical in process of assembly is less than 0.007 mm,so it can meet the accuracy requirements of micro-assembly very well. |
