Research On Machine Vision Based Position Algorithm Of Chip Mounter

As one of the core equipments of the surface mount system(SMT),the mounter integrates multiple disciplines such as machinery,control,information,and machine vision.Because of its high speed,high precision,low cost and other advantages,it has been widely used in the production of high-end electronic technology products,and has been specifically listed in the "Made in China 2025" plan as one of the core development technologies for equipment Manufacturing.As an important control input link of the placement machine,the vision system of the placement machine directly affects the placement speed and placement accuracy of the placement machine system.This article focuses on the realization of a complete placement machine vision system..According to the requirements of enterprise projects,the peper builds a software platform on the Windows system and the Open CV visual image library,and uses Python as the algorithm's main language to make algorithm experiments.And design and build the vision hardware experimental platform of the placement machine,to achieve the chip and PCB board image acquisition and data transmission functions.The specific research content is as follows:1.Determine the hardware solution of the vision system of the mounter.According to the detection accuracy requirements of the project and the estimation of the range of motion during actual placement,choose the camera parameters and select suitable lenses for the up and down camera system of the vision system,compare the quality of the images taken under different light sources to determine the light source parameters.2.For the high-precision detection of LQFP48 and SOP16 chips,an improved multi-linear fitting edge positioning algorithm based on Euclidean distance determination and chamfering is proposed.The original multi-linear fitting process is used to improve,analyze the impact of the lack of chip chamfering on the fitting result during the chip edge positioning process of the multi-linear fitting algorithm is analyzed.Therefore,an Euclidean distance determination and chamfering algorithm is added before fitting..The improved algorithm can not only improve the positioning accuracy of the algorithm but also improve the anti-noise ability of the fitting algorithm.3.Aiming at the identification and accurate positioning of Mark points on the PCB board,a least squares circle fitting contour center positioning algorithm based on bump detection is proposed.The algorithm improves on the original fitting algorithm,analyzes the deviation between the fitting center of the original fitting algorithm and the actual contour center when the Mark point is contaminated,by judging the defect direction of the contour and fitting in its defect direction Center offset compensation improves the positioning accuracy of Mark points,and improves the anti-noise ability of the algorithm due to considering the loss of Mark point information.4.Solve the internal and external parameters of the camera.Aiming at the problem that the captured image is distorted due to camera distortion a,affects the placement accuracy,Zhang Zhengyou calibration method is used to solve the internal and external parameters of the upward and downward looking cameras,thereby achieving correction function.