Textile Image Segmentation Using a Multi-Resolution Markov Random Field Model on Variable Weights in the Wavelet Domain

Abstract

This paper proposes a new texture image segmentation algorithm using a Multi-resolution Markov Random Field (MRMRF) model with a variable weight in the wavelet domain. For segmentation on textile printing design, firstly it combines wavelet decomposition to multi-resolution analysis. Secondly the energy of the label field and the feature field are calculated on multi-scales based on variable weight MRMRF algorithm. Finally new segmentation results are obtained and saved. Compared with traditional algorithms, experimental results prove that the new method presents a better performance in achieving the edge sharpness and similarity of results.