Comparative Study of Distance Metrics for Finding Skin Color Similarity of Two Color Facial Images

Abstract

This paper describes the comparative study of performance between the existing distance metrics like Manhattan, Euclidean, Vector Cosine Angle and Modified Euclidean distance for finding the similarity of complexion by calculating the distance between the skin colors of two color facial images. The existing methodologies have been tested on 110 male and 40 female facial images taken from FRAV2D database. To verify the result obtained from the existing methodologies an opinion poll of 100 peoples have been taken. The experimental result shows that the result obtained by the methodologies of Manhattan, Euclidean and Vector Cosine Angle distance contradict the survey result in 80% cases and for Modified Euclidean distance methodology the contradiction arises in 60% cases. The present work has been implemented using Matlab 7.