Research On Accurate Recognition Method Of Microscopic Features In Precision Assembly

The key issue of the MEMS automatic assembly process is to solve the measurement problem of the positional deviation of the parts during the automated assembly process.Quartz flexible accelerometer is a typical micro-electromechanical system.Realizing its flexible assembly can greatly improve the assembly efficiency and output of the product.It has great practical significance for China's national defense industry.The micro-assembly system involved in this paper is oriented to quartz flexible accelerometer products.Based on the flexible assembly line of micro-parts independently researched and developed by the research group,the paper focuses on the precise identification of the welding surface features of the gold-wire ball joints during the welding process of the torque-bearing parts.The micro-feature identification method of the design of the software and hardware of the alignment detection system,the calibration method of the imaging system and the feature recognition method of the gold ball solder joint is studied.The feature recognition method based on the relative position relationship is proposed to improve the micro-miniature,The accuracy of the identification of small size features in parts.The main research work of the thesis is as follows:Analyze the task requirements of the micro-assembly system involved in this paper,determine the functional indicators of each subsystem of the machine vision alignment detection system;complete the selection and non-standard design of related devices according to the assembly task requirements;Complete the writing of image processing programs on the Visual Studio platform,call The API functions of visual function libraries such as MIL and Opencv.According to the actual situation of this subject,the calibration method of industrial camera and the prism position and orientation calibration method are studied for the error cause and the motion coordinate system error and the spectroscopic prism pose are corrected.The image processing algorithms commonly used in machine vision technology are introduced.Different image proeessing algorithms are used to process the part images.The image processing results are analyzed and the appropriate algorithm is selected to preprocess and extract the acquired images.By analyzing the features to be identified and the relative positional relationship of the main positioning features of the moment seat is proposed.The feature analysis and recognition method based on relative position is proposed to identify the characteristics of the gold wire ball joint.According to the main positioning feature of the lower moment seat part—the engraving feature is used to make the template,the feasibility of the contour feature matching method in the application of the main positioning alignment feature identification is verified by the experiment;the relative positional relationship between the feature to be identified and the main positioning feature is divided.In the region and in the region of interest,the welding plane characteristics of the gold wire ball joint are accurately recognized.At the same time,the detection accuracy of the machine vision alignment detection system is verified by the assembly experiment of the high-precision shaft-hole parts.