Theapplication Of Machine Vision Method In The Detection Of Steel Strip Surface Defects

Metallurgical industry as an important industry of the national economy has beenrapid development in recent years, as one of the main products of the steel industry,strip steel has become an important raw materials for automotive, marine and aerospaceindustries. Its quality directly related to the quality and performance of subsequentproducts, so strip steel quality testing is of great significance.The thesis used machine-vision inspection system for strip surface defect detectionand classification, It analyzed and studied of on each step of image processing, andfound out the best analysis algorithm based on experimental data. Research result asfollows:1.The thesis used a special strip steel surface detection system, the system ismainly composed of the small strip transmission device and the machine-visiondetection system.2.According to the characteristics of the defect image noise, the thesis usedtraditional methods, wavelet threshold and wavelet packet algorithm for imagedenoising, The experimental results showed that wavelet packet algorithm can removethe noise, and retain the image details better.3.Aiming at the edge of the steel plate surface defect image of variety, complexedge, small target etc., used a method based on MAS wavelet transform to the steelplate surface edge detection, this method used the scale independent algorithm todistinguish step-like edge and roof edge, It can eliminate the interference caused bynoise edge, and the detected edge contour is continuous and clear.4.After extracting characteristic value of geometry, gray, texture feature on defectarea, then using Elman neural network to establish a network defect classifier, byidentifying the correct classification rate demonstrate the effectiveness of Elman neuralnetwork.