Optimized Fuzzy C-means Clustering Methods for Defect Detection on Leather Surface

Abstract

In this paper, captured images are segmented for the defective part, that is used for the further process of grading the quality of the products using automated inspection systems employed in industries such as leather, fabrics, textiles, tiles... etc.. These industries are the greatest conventional industries that need automatic detection systems as a basic part in diminishing investigation time and expanding production rate. Initially in this work, the input image is wet blue leather fed into a contrast enhancement process that improves the visibility of the image features. This contrast-enhanced image is employed with segmentation process that utilizes Fuzzy C-means algorithm (FCM) technique. This paper proposes two different optimization techniques, Grey Wolf Optimization (GWO) & Monarch Butterfly Optimization (MBO) for executing centroid optimization in FCM and results are compared with Modified Region Growing with GWO of leather segmentation method. The results exemplify that incorporation of optimization technique with FCM has a quite evident impact on segmentation accuracy of 96.90% over context techniques.