Stylized textile image pattern classification using SIFT keypoint histograms
Edutainment'11 Proceedings of the 6th international conference on E-learning and games, edutainment technologies
Hi-index | 0.00 |
This paper studies the methods of textile image segmentation which can be used for textile CAD (Computer-Aided Design). Based on the semi-supervised clustering, a new textile image segmentation algorithm is proposed by the minimum risk Bayes decision theory, which can get the final accurate results of segmentation by limited human assistance, that is, users indicate the relationship of some different regions in textile image by mouse. The algorithm firstly quantizes the textile image and then clusters by Bayes decision with prior segmentation information. Experiment result shows that the proposed algorithm is feasible and effective.