Automatic defect inspection system for steel products with exhaustive dynamic encoding algorithm for searches

摘要

A technology to stabilize the process by inspecting fine cracks in advance before mass and fatal defects occur is important. We propose a vision inspection system for the edge cracks of cold-rolled steel strips. It is important to detect the edge cracks of cold-rolled steel sheets early because they can result in plate breakage. There are two major components to achieve a suitable defect inspection system. The first is an optical system design technique that generates an image that can easily distinguish a defect area from a nondefect area. The second is an automatic detection algorithm technique that can accurately detect the position and shape of a defect in an image generated from an optical system. The optical part is designed using a backlight technique, which places the camera on the top of the strip and irradiates light from the bottom to the top. The defect detection algorithm detects the defects based on morphological operations. It is important to design the structuring elements that determine the performance of the detection algorithm. We optimized the structuring elements using the exhaustive dynamic encoding algorithm for searches. Experiments were conducted on the obtained images by installing the proposed system in actual production lines. The experimental results show the effectiveness of the proposed system.