Research On Inspection System Of Steel Strips Surface Cavity Based On Machine Vision

Vision detection has become an indispensable part of modern industry and manufacture. Machine vision, a current inspecting method, integrates image processing and analysis, mode recognition and artificial intelligence technology, has wide application in real-time inspection of industry products, especially the inspection of surface defects.Due to thin sheet and hard force on cold rolled line of steel strips, cavities, especially those on the edge, can easily lead the whole line to break down and cause fatal damage to manufacture. The aim of this subject is to meet the urgent need of inspecting surface cavity defects on steel strips.Main tasks are described as follow:(1)Basing on the principles of machine vision technology, we set up a surface cavity inspecting system which combines reflected light with transmitted light, and then introduce the hardware and software components. In the dissertation, ways to improve the efficiency and precision are proposed on the basis of existing achievements.(2) Gray level parameters of image are applied to determine whether or not there are suspected cavity defects on steel-strip surface. Then analyze the noise pattern and design a median filter to eliminate noisy spots. This filter can effectively reserve edge details and provide high resolution as well.(3) A new way of segmentation is proposed, which integrates traditional segmentation with clustering theory. This algorithm can separate target from background effectively. Use a morphological filter to wipe off isolated noisy spots caused by segmentation.(4) Feature parameters (perimeter, area and compactness included) are extracted. Classify locate the defects, and then show inspection results on the computer.(5) Program with Visual C++ language to test the algorithm.By looking up scientific literature, we have acquired some knowledge about recent development in relative area. It proves that machine vision technology can be well applied to the inspection system of steel strips surface cavity. The algorithms proposed in this thesis are studied both in theory and practice, thus they can well improve the performance of inspection system.