CLOTH TEXTURE CLASSIFICATION USING THE WAVELET TRANSFORM

Texture analysis plays an important role in several areas of image pro cessing and machine vision, such as biomedical imaging, textile manufa cturing, remote sensing, and military applications. In this study, it is shown that the wavelet transform is capable of providing features t hat may be used to discriminate between eight different cloth textures . The effectiveness of using wavelet transform features to classify te xture was compared with that of a commonly used method that extracts f eatures from spatial gray-level dependency matrices called co-occurren ce matrices. Texture features were input to a decision tree classifica tion algorithm to discriminate different textures. The wavelet transfo rm features correctly identified 86% of 64 cloth textures. The co-occu rrence features correctly classified 76% of the cloth samples.