This article is cited in 9 papers

Comparative study of texture detection and classification algorithms

P. P. Koltsov

Research Scientific Institute for System Studies, Russian Academy of Sciences, Nakhimovskii pr. 36-1, Moscow, 117218 Russia

Abstract: A description and results of application of the computer system PETRA (performance evaluation of texture recognition algorithms) are given. This system is designed for the comparative study of texture analysis algorithms; it includes a database of textured images and a collection of software implementations of texture analysis algorithms. The functional capabilities of the system are illustrated using texture classification examples. Test examples are taken from the Brodatz album, MeasTech database, and a set of aerospace images. Results of a comparative evaluation of five well-known texture analysis methods are described – Gabor filters, Laws masks, ring/wedge filters, gray-level cooccurrence matrices (GLCMs), and autoregression image model.

Key words: texture analysis, boundary points density, local extrema density, Gabor filter, Laws mask, ring/wedge filters, gray-level cooccurrence matrix (GLCM), autoregression image model.

UDC: 519.7

Received: 11.02.2011

 English version: Computational Mathematics and Mathematical Physics, 2011, 51:8, 1460–1466

Bibliographic databases:

• 
•