Comparison between EM and FCM algorithms in skin tone extraction

TL;DR

This paper compares EM and FCM algorithms for skin tone extraction across RGB, HSV, and YCbCr color spaces, finding EM in HSV yields the best results despite discarding luminance components.

Contribution

It provides a comparative analysis of EM and FCM algorithms for skin tone extraction in different color spaces, highlighting the effectiveness of EM in HSV space.

Findings

EM algorithm performs best in HSV color space.

Gaussian mixture model effectively models skin tones.

EM outperforms FCM in skin tone extraction tasks.

Abstract

This study aims to investigate implementing EM and FCM algorithms for skin color extraction. The capabilities of three well-known color spaces, namely, RGB, HSV, and YCbCr for skin-tone extraction are assessed by using statistical modeling of skin tones using EM and FCM algorithms. The results show that utilizing a Gaussian mixture model for parametric modeling of skin tones using EM algorithm works well in HSV color space when all three components of the color vector are used. In spite of discarding the luminance components in YCbCr and HSV color spaces, EM algorithm provides the best results. The results of the detailed comparisons are explained in the conclusion.