A method of steel strip image segmentation based on local gray information

Defect images segmentation results play an important role in the precision of classification in the automatic strip surface inspection system (SIS). In order to obtain a better result, a method based on local defect gray distribution information is presented in this paper. Though defect objects show different appearance and complicated gray level distribution, gray level of background in local area is scattered regularly. Gray scale of background in defect region of interest (ROI) is analyzed to obtain the mean value and its distribution. Using these features, background part can be separated from defect object in the region of interest. The experimental results show that the method can be employed to segment steel strip surface defect image effectively, and to separate bright and dark defect objects at the same time.