Back to articles
Volume: 17 | Article ID: art00040
Material Classification Using Color and NIR Images
  DOI :  10.2352/CIC.2009.17.1.art00040  Published OnlineJanuary 2009

Material classification is becoming more important in computer vision and digital photography applications, which require accurate classification of objects present in the imaged scene. This is a very challenging task because the sheer diversity of scene content and lighting conditions decreases the usefulness of many color- and texture-based features used in image classification. In this work, we investigate the potential offered by using information outside of the visible spectrum, specifically the near-infrared (NIR). The difference in the NIR images' intensities is not just due to the particular color of the material, but also absorption and reflectance characteristics of the colorant. This relative independency of NIR and color information makes NIR images a prime candidate for classification. The database, on which the training and testing were conducted, consists of textile, tile, linoleum and wood samples. To classify the materials, visible and NIR images were analyzed according to their lightness, texture, and color. The analysis results were the input to a classifier in form of feature vectors. The results show that our database is classified almost exactly. Comparing with visible-only features, wood and textile samples were better classified due to the additional information the NIR images provide.

Subject Areas :
Views 7
Downloads 1
 articleview.views 7
 articleview.downloads 1
  Cite this article 

N. Salamati, C. Fredembach, S. Süsstrunk, "Material Classification Using Color and NIR Imagesin Proc. IS&T 17th Color and Imaging Conf.,  2009,  pp 216 - 222,

 Copy citation
  Copyright statement 
Copyright © Society for Imaging Science and Technology 2009
Color and Imaging Conference
color imaging conf
Society of Imaging Science and Technology
7003 Kilworth Lane, Springfield, VA 22151, USA