Title :
Texture classification through level lines
Author :
Hamdan, H.M. ; Larson, L.M.
Author_Institution :
Dept. of Math., Louisville Univ., KY, USA
Abstract :
The paper presents a method for texture classification through level lines. Level lines provide a non-redundant and full representation of an image, and they can be regarded as a signature for the image. By using earth mover´s distance (EMD), we can classify texture images based on EMD between their level lines. Our method is a geometric approach for texture classification, which has the advantage of being contrast-, rotation- and scale-invariant.
Keywords :
computational complexity; computer vision; content-based retrieval; image classification; image representation; image retrieval; image texture; linear programming; set theory; trees (mathematics); computational complexity; computer vision; content-based image retrieval; content-based retrieval; earth mover´s distance; geometric approach; image representation; image signature; level lines; level set; linear-programming optimization; texture classification; texture image classification; Computer vision; Content based retrieval; Earth; Image retrieval; Image texture analysis; Layout; Level set; Mathematics; Pixel; Shape;
Conference_Titel :
Image Processing. 2002. Proceedings. 2002 International Conference on
Print_ISBN :
0-7803-7622-6
DOI :
10.1109/ICIP.2002.1039127
