Image processing techniques for leather hide ranking in the footwear industry

Abstract

The subjective assessment of finished leather hide has resulted in much argument, wasted costs and disruption of production schedules in both the tannery and leather footwear industry. Leather has attracted much research work worldwide; however, to date no objective quality assessment methodology is available, let alone an automatic assessment system. In this project, a machine-vision-based approach to grading leather hide has been developed. The image of a given castle hide was acquired using the developed machine vision system. A map of the different types of defect contours contained in the hide was extracted. Algorithms were then formulated enabling the computerization of the quarter rule, a standard method for grading leather hide in the footwear industry. Typical results from the developed grading process include the leather cuttable areas within a given hide, taking into account various types of defects, and the contour map of defects, which in fact is a vital input to any computerized shoe-part nesting system. The success of automatically grading leather hide in this project will not only provide the tanner with a means of accurately defining the price of a leather hide but also greatly assist the footwear manufacturer in the optimal nesting of irregular shoe parts within a given hide.