Strip Steel Defect Detection Based on Saliency Map Construction Using Gaussian Pyramid Decomposition

A novel detection algorithm for strip steel defect image based on saliency map construction using Gaussian Pyramid decomposition is proposed in this paper. Firstly, the acquired gray image of strip steel is decomposed into strips steel sub-images with different resolution by Gaussian Pyramid. Secondly, the saliency map is constructed by the central-surround differences operation of strips steel sub-images and image fusion of difference sub-images. Finally, we respectively calculated mean values of maximum value in image rows and columns, in which small mean is chosen as the optimal threshold segmentation of strip image, and then to segment surface defects of steel strip. Experiment results show that the proposed method is valid for inhibition of the image background and can be realized complete segmentation and accurate detection for strip steel defect.