Automatic Recognition of Woven Fabric by Using Gray Level Co-occurrence Matrix

摘要

This paper proposes an approach to texture analysis that can be used to extract the features of woven fabric images and recognize the woven fabric structure. 2-D wavelet transform is utilized to decompose the original image, in order to gain the 1/16 size of the low-frequency sub image. Then gray level co-occurrence matrix is employed to extract the feature parameters: energy, contrast, correlation and entropy. Lastly, learning vector quantization neural network is adopted to classify three basic weaves: plain weave, twill weave and stain weave. The experiment results demonstrate that gray level cooccurrence matrix can extract the features well, and learning vector quantization neural network can recognize three basic woven fabric structures automatically and accurately.