Análise de técnicas de pré-processamento para a segmentação de imagens de vitiligo

Abstract

Vitiligo is a disease identified by the loss of skin pigmentation, due to the decrease or absence of melanocytes. In order to evaluate these vitiligo images, this work presents some pre-processing techniques with white indexes and related images with the k-means and Otsu algorithms. The evaluation of these techniques was performed using four metrics: accuracy, precision, f1-score, and recall. The best results of white indexes with associated k-means obtained values between 90 and 91% of accuracy and 90 to 94% of f1-score.