Research On Online Application Of Strip Surface Quality Inspection System Based On Machine Vision

With the increase of the economies of scale,the manufacture industry is developing dramatically,including automotive,home appliance,aerospace and etc.The demand of steel is increasing rapidly,which plays an important role in GDP.However,due to the complicated manufacture process,it is inevitable to have some defects on the surface of the strip steel,which includes cracks,scarring,roll marks,scratches,stains and rust red and etc.Also,the current weakness of the system is including the inaccurate way of checking defects of strip steel,discontinuous operation,non-automatized analysis of defects.Therefore,it is necessary to improve the real-time detection,automation,productivity and quality,which can enhance its competitiveness,wear resistance,anticorrosion.The research below contains three aspects:(1)The thesis describes the image acquisition and transmission system,and how the image is transmitted by Gigabit Ethernet.Additionally,it describes how to ensure the efficiency and effectiveness of the image transmission by way of the design and performance of the board,the design of image capture card,the process of image capture and transmission,and etc.(2)The thesis describes how to extract the image of defects from the surface of strip steel,which includes three methods,image feature extraction method,defects detection and extraction method,and feature detection and extraction method.Usually,the defect of strip steel contains morphological features,grey features and texture features.And the defect can be extracted and analysed based on the different features.(3)The thesis raises an improved Support Vector Machine,which is based on regression prediction.The research indicates that the method can increase the accuracy by 12.4% for the image recognition of the strip steel.