Shoe Upper Feature Point Detection Based On Vision System

The feature point punching of textile shoe uppers is one of the key steps in the high-end shoemaking process.The quality of the punching will directly determine whether the subsequent shoe sample materials can accurately match the transfer paper.At present,most shoe-making companies in my country still use manual punching in this link,which is still far behind the automated punching in developed countries.In recent years,with the continuous innovation and development of domestic computer information technology,machine vision technology has become more and more widely used in the industrial field.It is imperative to use machine vision technology to realize automatic punching of shoe upper feature points,which is greatly improving shoe companies.At the same time,it reduces the production cost of the enterprise and gets rid of the shortcomings of traditional manual craftsmanship.This paper designs a machine vision system to complete the automatic punching of shoe upper feature points.First,according to the analysis of different shoe upper characteristics,a shoe upper classification network with Mobile Net V2 as the core is proposed,which has a faster processing speed while ensuring the accuracy of classification.Through the investigation of the entire project,combined with the actual production situation,the production equipment was selected and built,and the camera calibration was completed.Visual algorithms are used to preprocess the images collected by the industrial camera to filter out some external interference,and then extract the overall outline of the shoe upper,reducing the amount of calculation in subsequent steps.Finally,based on template feature matching,it is proposed to perform feature matching extraction on the shoe upper twice to realize the positioning and punching of the feature points of the shoe upper.This article mainly uses Visual Studio 2015 as the platform and C# as the programming language to realize the acceptance of data signals and the sending of coordinate data,and build a corresponding visual human-computer interaction interface for the recognition of shoe uppers and the positioning of feature points.In the image processing module,the professional industrial vision software Vision Pro is mainly used to realize the image processing,which simplifies the programming workload of the researchers.Finally,the experimental verification is carried out to verify the accuracy of the algorithm and experimental system proposed in this paper.The shoe upper feature point detection based on the vision system studied in this paper comes from the actual production line of shoe companies,and has certain practical application value.