Research On Surface Defects Detection And Identification For Cold-rolled Steel Strips Based On Image Processing Technology

With the development of modern industry,the cold-rolled strip steel has been widely used in machinery manufacturing,aerospace,petrochemical and so on.Strip quality is required higher and higher in the use of cold-rolled strip,especially the strip surface quality.In the strip steel rolling process,some defects are inevitably created on surface such as scratch,hole,scarring,skin oxidation,cracks and so on,which reduces the serious deficiencies of the anti-fatigue strength of steel,corrosion resistance, resistance high temperature,and wear resistance properties.Therefore,detection and control of the strip surface defect appears particularly important.This article studies:1.On the basis of analyzing the cold-strip steel surface defect,combined with the basic requirements,the whole program of the strip surface defect detection system is designed,and hardware structure and software flow of the detection system are described in detail;2.Image processing of the strip surface defect detection system is studied,for the strip scratch,hole,emulsion spot characteristics of three types of defect images,using Gaussian filtering technique filter out the noise in images and enhance images.and the advantages and disadvantages of several classical edge algorithms are analyzed. Finally,using a fuzzy C-means clustering algorithm extract image edge,get the accurate defect location information and complete the segmentation positioning of the defect;3.Image characteristics of scratch,hole,emulsion spot on the surface strip are analyzed and selected,and the feature extraction algorithms are studied.The experimental results show that the defect shape and texture features eigenvalues obtained are targeted and high accuracy;4.Through analyzing network structure,learning algorithm and combined with strip surface defect detection,the surface strip defects identification and classification based on BP neural network are studied and realized.The results show that this method can effectively detect three defects of the strip scratch,hole,emulsion spot and the detection rate is over 80%.