On comparing color spaces for fabric defect classification based on local binary patterns

摘要

There is a numerous color spaces with different properties in literature. In order to find the appropriate and relevant color space for the fabric defect classification problem, we propose to investigate the performance and robustness of the Local Binary Pattern (LBP) descriptor in supervised context by using SVM classifier. The experimental results show that the luminance-chrominance spaces are suitable for coding fabric defect with the classification accuracy obtained is 92.1%.