Computer Vision Inspection And Classification System In Glassware Production

In the process of glassware production,the method of checking the quality of glassware is still based on the engineer's experience.The precision and efficiency of glassware production are difficult to guarantee.It is difficult to culture the objectivity during the production.The quality inspection based on computer vision lays a solid foundation to improve the production efficiency and the product's stability in glassware enterprises.Under this background,this subject designs a set of glassware quality inspection system.This system use the computer image processing and real data detect main contour dimensions of the glassware online.The glassware quality inspection system includes shape,size,and surface defects.The main tasks are:Use a water cup as the sample.According to computer vision develop a quality inspection system of glassware.The hardware module of the system includes PC,optical lens,camera,light source,rotating mechanism,culling mechanism and machine adjusting structure.This paper mainly studies the matching relation between the hardware involved in image acquisition.An improved filtering method is used to pick out the pictures of the measured objects.Using two-dimensional Otsu method to divide the image threshold,and then the mathematical morphology and binarization of the divided photos are processed.This step lays the foundation for edge extraction.After analyzing and comparing several common boundary detection methods,it uses the Sobel detection operator method is used to divide the image threshold.In the process of photo recognition,the improved fast Hough transform line detection method is used to detect the boundary line.As the result,the circle contained in the image is detected by using Hough transform method,and the diameter and roundness of the circle are calculated,and the size of glassware and the size of defect are obtained.The software part of glassware quality inspection system should uses application of MATLAB to construct the computer vision and classification system.The final result of the actual test show that can achieve the established goals.