Unsupervised perceptual model for color image's segmentation

被引：0

作者：

Sobrevilla, P ^{[1
]}

Gómez, D ^{[1
]}

Montero, J ^{[1
]}

Montseny, E ^{[1
]}

机构：

[1] Tech Univ Catalonia, Appl Math Dept 2, Barcelona, Spain

来源：

NAFIPS 2005 - 2005 ANNUAL MEETING OF THE NORTH AMERICAN FUZZY INFORMATION PROCESSING SOCIETY | 2005年

关键词：

segmentation algorithms; image classification; coloring problem; fuzzy sets; perceptual vision;

D O I：

10.1109/NAFIPS.2005.1548560

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Color Segmentation is a fundamental, step in image understanding. Moreover, for getting accurate color image's segmentation algorithms, human being's perception of color should be considered. In this line we propose an unsupervised segmentation algorithm that is based on a fuzzy graph coloring process for representing the fuzzy color similarity degrees among neighboring pixels from a perceptual point of view. As main goal is to detect and extract the regions explaining the image, we stress the role of coloring procedures for unsupervised segmentation and fuzzy classification by means of useful, comprehensive and simple enough fuzzy graphical representations.

引用

页码：349 / 354

页数：6