Paper web defection segmentation using Gauss-Markov random field texture features

In order to segment paper web defections effectively, texture features, based on the Gauss-Markov random field model, were used in this paper. By introducing the characteristics of paper web texture features, the maximum difference of the local texture parameters w₁, w₂, w₃ and w₄ of the Gauss-Markov random field model was used as a judgment index for web defection segmentation. A dirty spot defection was segmented by this method and its result shows that the paper web defections can be effectively segmented by the judgment index. The max difference of the local texture parameters of the Gauss-Markov random field model can be used as the judgment index to segment the defections in the paper web, which has the characteristics of nature texture.