Comparison of Methods in Skin Pigment Decomposition

TL;DR

The paper addresses the challenge of decomposing skin pigment into melanin and hemoglobin distributions from images to aid diagnosis. It compares HSV-based color-space methods, a physics-informed optical-density model with PCA/ICA, and a nonlinear-manifold approach using Isomap before ICA. The results indicate that Isomap can outperform PCA on nonlinear pigment manifolds, but its scalability is a key limitation, highlighting trade-offs between intuitive color models and physics-based modeling. The work guides method selection based on data complexity and scale, with potential for more accurate pigment maps in medical contexts.

Abstract

Decomposition of skin pigment plays an important role in medical fields. Human skin can be decomposed into two primitive components, hemoglobin and melanin. It is our goal to apply these results for diagnosis of skin cancer. In this paper, various methods for skin pigment decomposition are reviewed comparatively and the performance of each method is evaluated both theoretically and experimentally. In addition, isometric feature mapping (Isomap) is introduced in order to improve the dimensionality reduction performance in context of skin pigment decomposition.

Figures (7)

• Figure 1: Overview of Kim’s method.
• Figure 2: Skin color model in the optical density domain.
• Figure 3: Comparison of dimensionality reduction performance between Isomap and PCA. (c), (d): cumulative contribution ratio on (a), (b).
• Figure 4: Experimental inputs: (a) white circle: pimple (b) black circle: freckle; red circle: pimple.
• Figure 5: Pigment decomposition results of regional facial skin by Kim’s method: (a), (b); Tsumura’s method: (c), (d); Proposed method (Isomap+ICA): (e), (f). From left to right: hemoglobin map and melanin map. Red circle in (b) locates a mistake.