The cold rolling strip surface defect on-line inspection system based on machine vision

In order to inspect the cold rolling strip main surface defects in while product line running normally and control manufacturing process more efficiently, a novel cold rolling strip surface defect on-line inspection system is designed. Modularization design idea is adopted, the whole system is made up of light source module, image collection module, defect image processing module and server module. For making sure real-time characteristics of the system, high-speed linear CCD camera and relational high-speed fiber image collection card, high-speed FPGA image processing card designed by ourselves are adopted in hardware, RT-Linux real-time operation system is adopted in software. Aiming at low contrast and high noise of defect image, homomorphic filtering algorithm based on PDE is proposed for image preprocess; considering of the exist of defect image transition region, image segmentation algorithm based fuzzy-set and information entropy theory is adopted to accomplish defect abstraction. Based on these, kinds of features of defect are extracted, the defect classification based on fuzzy support vector machine is designed and classification experiment is finished, better classification effect is obtained.