Font Size: a A A

Research On Sheet Counting Technology Based On Machine Vision

Posted on:2022-07-28Degree:MasterType:Thesis
Country:ChinaCandidate:T P ZhuFull Text:PDF
GTID:2518306533994939Subject:Electronic information
Abstract/Summary:PDF Full Text Request
Plate counting is an inevitable problem for every plate manufacturer,which is related to the economic interests of the manufacturer.The traditional manual counting and mechanical counting methods are inefficient.Therefore,the development of plate counting technology based on machine vision is of great significance to improve the accuracy of plate counting and the labor productivity of plate manufacturers.In this paper,three methods are used to study the technology and method of plate counting:(1)In this paper,the method of counting the number of plates by skeleton extraction is studied,and the skeleton extraction method based on the maximum disc method is used.Firstly,the improved histogram equalization is used to enhance the image and enhance the linear features of the plate;then,before the skeleton extraction,the image is binarized,the maximum disc method is used to extract the skeleton of the plate,the angle threshold method is used to eliminate redundant pixels,the least square method is used to fit the skeleton into straight lines,and finally the skeleton lines of the plate are counted.(2)The method of wave peak detection is used to study the plate counting.Through detail enhancement,the influence of foreign body adhesion on the wave peak detection is eliminated,and the gray difference between the plate and the gap is enhanced.The gray projection method is used to extract the one-dimensional gray features of the plate.The pseudo wave peaks are eliminated by the wave peak screening conditions in this paper,and finally the plate counting is realized.(3)The full convolution neural network is used to realize the semantic segmentation of the plate end face image.The vgg16 network is used as the basic network to build the network model,and the self-made plate data set is trained to obtain the semantic segmentation model in this paper.The model is used to predict and segment the plate end face image,and the segmentation results of the model are counted to verify its accuracy.In this paper,based on machine vision,three methods are used to realize the problem of plate counting based on machine vision,and the experimental results of 0.8 mm plate with100% accuracy are achieved in the laboratory.
Keywords/Search Tags:Machine vision, Plate count, Skeleton extraction, Wave peak detection, Convolutional neural network
PDF Full Text Request
Related items