Comparative Analysis of Rule Based and K-Nearest Neighbour Approaches for Classification of Different Shades of Basic Colours in Fabric Images

Abstract

Proposed work presents comparative analysis of rule based and K-nearest neighbour approaches to classify the different shades of basic colours in fabric images. The mean and standard deviation of CIE Lab features of shades of red, green and blue colours are computed. A rule base is designed. An overall recognition rate of 97% is obtained in case of rule based approach. K-nearest neighbour classifier is designed taking into account, the average Lab values. The overall recognition rate of 94% is obtained in case of K- nearest neighbour approach.