Research On Cable Surface Defect Detection System Based On Machine Vision

Wire and cable is a very important industrial processing materials.The surface quality is the first visual impression of the product,Which is an important measure of the product.The surface quality affect the consumer's satisfaction with the product directly,but also has a large extent affect to the final performance of the product.In recent years,with the rapid development of computer image processing technology and machine vision detection technology,machine vision inspection has been applied to various industries of life product quality testing process.Application of machine vision technology to cable surface defect detection,not only can achieve the whole process of real-time online detection,but also to a large extent improve the accuracy of detection.Thereby improving the quality and production of the cable.Two cores of machine vision system are image acquisition technology and image processing technology,In the high-speed production line,the image acquisition technology determines the quality of the collected image.High-quality images can not only eliminate the effects of some impurities,but also reduce the complexity of image processing algorithms.Image processing technology determines the accuracy of the entire detection system and the false detection rate and missed rate directly.A cable visual inspection system based on machine vision has been designed in this thesis.The main research contents are as follows.Firstly,an image acquisition system with 360° panoramic image of cable has been designed in this thesis.In this paper,three cameras in the same direction of the cable plane were 120° uniform layout of the design,to achieve 360° panoramic image acquisition.Choose the right industrial camera and lens according to the industrial site environment and the testing requirements.In order to obtain high-quality cable apparent image.A scientific lighting design has been conducted in this thesis.Experiments show that the image acquisition system can capture the panoramic view of the cable with radius of 1-20 mm when the production line is under 1.5 m/s.below the production line of less than 1.5 m / s.The image is clear and the distortion is small,which can meet the needs of industrial testing.Secondly,an image recognition algorithm for cable apparent defect has been designed in this thesis.Two different algorithms are used to detect the two types ofdefects,and the detection accuracy is as high as 94%.Thirdly,the system software part of the system has been designed in this thesis.The software design is based on the Qt4 graphical interface.In addition,the software part of the image processing algorithm has been completed in this thesis,the real-time display of the online image acquisition and the statistics of the defect information.Finally,the whole system has tested on the production line.Test results show that the detection level of the system to achieve the same level of detection of similar products.According to our access to the main results of the database,the system which is based on three cameras at the same time to achieve multiple types of cable surface defects on-line has not been reported.