Graph Cut segmentation with automatic editing for Industrial images

Segmentation of Industrial image is an important step for automatic production and product inspection. There are many challenges for automatic industrial image segmentation, such as poor contrast at workpiece boundaries by complex illumination. Relying on information from the images does not work well in traditional segmentation. In this paper, we propose a novel automatic editing method for improving the performance of Graph Cut method in industrial image. Our method, proposed in this paper, is able to edit the segmentation results automatically using both the mechanical manufacturing environment´s information and the workpiece geometric shape´s information. The process can automatically revise the segmentation results efficiently while taking advantage of Graph Cuts´ global optimization. We adopt the method for the segmentation of bearing image from the assembly line. The experiment results reveal that our approach is effective and practical for the mechanical components´ segmentation.