Research On Shaft-hole Parts Select System Based On Machine Vision

With the introduction of Industry 4.0 and technological innovation,the traditional manufacturing industry is facing tremendous changes,and the production and manufacturing system is facing challenges such as more precision,intelligence,and information integration,while the current main operating methods of the traditional manufacturing industry are still manual.Lord,this undoubtedly brings many problems.This subject mainly takes the main piston and guide plate of the valve parts as the research object,and designs a set of shaft hole parts selection system based on machine vision,which aims to solve the measurement accuracy and selection of the two parts in the manual selection process Problems such as efficiency,parts backlog and waste.According to the analysis of the actual working conditions,this topic designed the matching scheme for the main piston and the guide plate,and carried out the construction of the system,and studied and solved the two key problems involved.Aiming at the dimensional accuracy of the main piston and the guide plate of the valve parts,this topic mainly relies on machine vision and image processing related technologies to achieve the size of the two parts.Through the theoretical analysis and practical effect comparison of various image processing algorithms,in this paper,the median filter method is used to reduce the noise of the image.The Canny operator obtains the pixel edge of the image,and the sub-pixel edge detection method based on the fitting method obtains the sub-pixel edge of the image.Then,the main piston and the main piston are obtained by the least squares fitting method.The pixel size of the guide plate is obtained by Zhang Zhengyou calibration method to obtain the camera parameters required for imaging correction,and then the actual size of the shaft hole is obtained through the conversion of the coordinate system.Through repeated measurement experiments,the results show that the detection accuracy of the size measurement system in this paper reaches 0.03 mm,and the measurement efficiency is within 1s/piece,which has reached the expected design goal of the system.Aiming at the problem of manual matching of two kinds of parts,this paper uses the maximum number of matching as the criterion,and proposes three matching algorithms based on greedy algorithm,axis as reference,and genetic algorithm to perform batching of main piston and guide plate parts of known size.Matching and comparing and verifying the three algorithms.Experiments show that the matching algorithm based on genetic algorithm has a high matching success rate and high efficiency,which meets the actual application requirements.