Automated woven fabric texture periodicity extraction by spectral analysis and patch-DMF

Abstract

Texture provides important visual information for object identification, and texture representation is still a challenging problem in texture analysis. Fabric texture is a typical structural texture, whose periodic behavior plays a vital role in texture discrimination. In this work, an approach to extract the periodicity of fabric texture is proposed, in which the classic spectral analysis method Fourier transform is applied to select underlying frequency peaks corresponding to the texture periodicities given its efficiency and global property; then the patch-DMF based on the distance match function is built to evaluate the best match texture periodicity. The experiments on near-regular textures (plain weave fabric) demonstrate that the performance of the proposed method shows accurate results with near and regular fabric textures compared to others, and being practical usefulness for texture description.