Development Of Surface Defect Detection System For Sanitary Electroplating Parts Based On Machine Vision

Machine vision insp ection technology h as been widely used in electronics,food,medicine and other industries,Due to factors such as mirror reflection of sanitary electroplating parts,it mainly relies on manual surface defect detection.Manual detection has risks such as high work intensity,low efficiency,missed detection,and artificial subjective judgment.Based on robotics and machine vision technology,this thesis deploys Halcon software for visual positioning and focuses on design of product imaging systems and development of different defect detection algorithms.Breaking through the problems of image reflection and partial exposure of sanitary electroplating parts,the detection of different defects of flat sanitary electroplating parts was successfully realized.1)A solution for a visual inspection system was designed.The hardware system includes a transmitting device,an image capturing device,and a robot vision positioning device,while the software system includes an HMI and an image processing module.2)The image was corrected for distortion using a calibration plate.For the visual positioning task,the transformation matrix of the pixel cooorsimates and the global oordinates of the sanitary electroplating parts by hands-eye calibration were determined;algorithms suitable for target location were proposed from the perspective of Blob analysis and shape-based teomplate snatching.It was found that the shape-based template matching method was better when the illumination was unstable.3)This project solved the imaging problem of sanitary electroplating parts.Based on the experimental comparison of the color of the light sourcemthe type of light source and the lighting environment.the imaging system was designed A blue light source with le ss dust generation was.used for imaging;by adopting a double-shaped light source with high c ontrast imaging,wide.and uniform illumination;Aiming at the problem of imaging reflection,the angle elimination strategy of changing the camera and product was proposed.When the angle was 80 degrees,the imaging effect was the b est.For local exposure problems,The dark room lighting environment was used to isolate the influence of external ambient ligh on the uniformity of light.4)For defect detection algorithims.the filtering effects of the median,mean,Gaussian and improved median filters were analyze and compared.An improved median filter with better image detail protection and faster calculation speed was selected.Histogram equalization solves image smoothing after filtering.For the scratches and pinhole defects,the Fourier transform algorithm with less influence of randomness of defects was adopted;For the flaws of the bump,a differential and morphological algorithm that minimizes the original shape of the defect was used.This work used C#language and Halcon vision software based on Visual Studio2013 development environment,joint programming and design of visual inspection software,with two flat sanitary electroplating parts for algorithm verification,It was concluded that the scratch accuracy rate was 95.4%the pinhole and bump accuracy rate was 96.9%and the detection time was controlled within 500ms.The visual inspection system has been successfully applied to the surface defect detection of the company's flat plating parts.