Sensor based color sorting system for leather shoe components

Abstract

Component color matching is an important aspect in leather product manufacture and this is currently being carried out manually by experienced assorters. But errors do occur due to fatigue and subjectivity associated with manual color sorting of components and the productivity is also lower. The error in component color matching results in a product being rejected although all the quality parameters are otherwise met. Hence large leather product manufacturers are looking for an objective, cost effective and faster color sorting system. Reflectance spectrophotometer, which are used in quality control in color matching are too expensive and cumbersome for online color sorting. Color sensor based color sorting offers itself as an attractive option. Automatic color sorting based on color sensors has been screened for its application in fruit sorting but only one color parameter, either redness or yellowness is often used in such systems. In case of leathers, one parameter will be inadequate as the color should be the same with acceptable tolerance levels in terms of all three parameters, lightness or darkness(L), red or green(a), yellow or blue(b) in CIE coordinate system or L, c(chroma), and h(hue) in CMC coordinate system. This makes the sorting of leather components much more difficult and technologically challenging. In the present study, a color sensor based circuit has been designed and fabricated and investigated for use in leather color sorting. K-means clustering has been employed to sort the leather sample components based on L, a, b values into 4/5 clusters. The accuracy of the system has been validated and shows closer conformity with the expert opinion.