期刊论文详细信息
PATTERN RECOGNITION 卷:53
Fuzzy aura matrices for texture classification
Article
Hammouche, Kamal1  Losson, Olivier2  Macaire, Ludovic2 
[1] Univ Mouloud Mammeri, Lab Vis Artificielle & Automat Syst, Tizi Ouzou, Algeria
[2] Univ Lille Sci & Technol, Lab CRIStAL, CNRS, UMR 9189, Cite Sci Batiment P2, F-59655 Villeneuve Dascq, France
关键词: Fuzzy aura set;    Fuzzy aura matrix;    Texture classification;    Spatially variant neighborhood;   
DOI  :  10.1016/j.patcog.2015.12.001
来源: Elsevier
PDF
【 摘 要 】

The aura concept has been developed from the set theory and is an efficient tool to characterize texture images. It is based on the notion of aura set and on the associated aura measure that involve the neighborhood of each image pixel. In this paper, we propose to extend this concept to the framework of fuzzy sets in order to take the imprecise nature of images into account. We define the notions of fuzzy aura sets and of aura measures to compute fuzzy aura matrices as texture descriptors. Fuzzy aura measures assume no restrictions about the neighborhood shape, size, and spatial invariance. Extensive tests of texture classification on Outex benchmark datasets show that fuzzy aura matrices computed with spatially variant neighborhoods often outperform other powerful texture descriptors on both gray-level and color images. (C) 2015 Elsevier Ltd. All rights reserved.

【 授权许可】

Free   

【 预 览 】
附件列表
Files Size Format View
10_1016_j_patcog_2015_12_001.pdf 1712KB PDF download
  文献评价指标  
  下载次数:0次 浏览次数:0次