Research On Surface Quality Inspection Of Thermostat Shell Based On Laser Vision

For the industrial product,it is an important factor to ensure the quality of production and processing are fast,accurate,and highly stable.At present,many traditional enterprises still use manual detection or simple physical methods to detect products.Its slow speed,poor consistency,low accuracy make it difficult to meet production capacity requirements and indirectly restrict the development of the company.With the rapid development of machine vision and image processing,visual inspection is rapidly improving automated production.Its non-contact detection advantage ensures the safety of personnel,improves the detection efficiency,fast and accurate.It has become an important direction of development and research focus in the process of rapid industrialization reform.This article takes the thermostat shell produced by the company as the research object,makes a deep analysis on the surface quality of the thermostat shell,and designs a set of thermostat shell defect detection system based on machine vision.Due to the insufficiency of the twodimensional detection itself,which has many limited detection requirements,such as inaccurate depth information acquisition.So threedimensional inspection is applied in this paper,and a blind hole depth measurement system based on triangulation is developed.Combined with Halcon software,a simple but fast,effective and accurate depth measurement is realized.The study is as followsFirst,as the inspection requirements proposed by the company,a system design proposal was put forward,and a two-dimensional detection system and a three-dimensional measurement system for the shell surface were designed and constructed.In combination with the product characteristics of the shell itself,selecting suitable hardware.High-definition image acquisition to prepare for subsequent image processing.Second,Gaussian filter for image preprocessing,in the surface detection program,the shell is divided into three areas corresponding to the three stations,respectively,the bottom outer circle of the shell,the bottom inner circle of the shell,top of the shell.For the station1,A multi-threshold segmentation is used to coarsely extract the outline of the outer circle,and then an accurate circle center coordinate and a radius are obtained through sub-pixel edge extraction and edge fitting.A circle is generated and a polar coordinate transformation is performed to rapidly find a defect portion.The method adopted by Station 2 and Station 3 is similar.First,the affine transformation is performed on the area to be processed by taking the parameters based on the shape-based template matching,so that the to-be-inspected area in each image can be accurately located,and then each In each area,a global threshold method was used to detect the defect area to determine whether it was qualified.Third,for three-dimensional blind hole measurement,we analysis the triangulation theory and the principle of small hole imaging.Maded a strict calibration with halcon,including camera internal and external parameter calibration,light plane calibration,moving pose calibration.After obtaining the calibration data,the three-dimensional reconstruction is performed,and the standard block is measured using the depth measurement method based on the Z coordinate gray value,and the blind hole depth of the shell is measured after the compensation error is obtained.The experimental result of the sample indicates that the algorithm structure of this paper achieves a versatile and accurate judgment of various surface defects,and has a good reference and promotion significance for other similar products.The blind hole depth measurement results show that the accuracy can be maintained within the 0.03 mm tolerance range,which basically meets the set measurement target.