EXTRACTING PRINTED DESIGNS AND WOVEN PATTERNS FROM TEXTILE IMAGES

The extraction of printed designs and woven patterns from textiles is formu lated as a pixel labelling problem. Algorithms based on Markov random field (MRF) optimisation and reestimatio n re described and evaluated on images from an historical fabric archive. A method for quantitative e valuation is presented and used to compare the performance of MRF models optimised using α−expansion and iterated conditional modes, both with and without parameter reestimation. Results are promising for potential application to content-based indexing and browsing.