Development Of Machine Vision Inspection System For Processing Quality Of Double-row Metal Parts

Metal parts are an important part of the automotive industry.With the improvement of people's living standards,the demand for automobiles is growing.Correspondingly,more metal parts need to be produced.At present,metal parts manufacturers are generally small and medium-sized enterprises.The lack of funds makes these companies only able to purchase advanced production equipment to improve production automation.Most enterprises can't afford expensive automatic testing equipments,so the quality of the product can only be detected by manual visual inspection or by means of simple gage inspection.The automation of the inspection process is still far from sufficient.Therefore,it is of great significance to study the automatic precision testing equipment for metal parts with independent intellectual property rights and high cost performance.In view of the above problems,this paper studies the automatic detection system of double-row metal parts based on machine vision technology.Firstly,the content and requirements of the processing quality of the two parts are analyzed.The overall solution and workflow are designed according to the testing requirements.On this basis,the main functional modules and correlations of the testing equipment are analyzed.A visual solution is designed based on the characteristics of the tested part,and the sorting device is designed to separate the defective parts.An image processing algorithm that automatically extracts the boundary of the part to be measured and detects the surface features from the feature image is studied.According to the feature image collected by the vision system,it is preprocessed to obtain the ROI,then the edge information of the feature quantity to be measured is extracted by the sub-pixel edge detection technology and the edge feature circle is fitted by the least squares method.The geometric template matching algorithm,the gray template matching algorithm improved by Gaussian image pyramid model,and the defect detection algorithm based on mask technology are applied to the image processing experiments to verify that these algorithms can accurately and quickly detect defect features.The camera calibration model applicable to the detection system was studied,and a method of performing calibration using a standard calibration plate was designed.The software for detection system is programmed based on C# and HALCON dynamic library,and carry out a series of actual measurement experiments using the developed automatic precision inspection equipment for double-row metal parts processing quality.By analyzing the experimental results,the measurement accuracy of the device,repeated measurement precision and measurement correct rate are verified to meet the requirements of the enterprise for automatic detection equipment.Finally,the possible sources of detection error are analyzed.