Lace fabric image retrieval based on multi-scale and rotation invariant LBP

Lijin Zhang, Xueliang Liu, Zhihong Lu, Fei Liu, Richang Hong

Abstract: It is a challenging problem of lace fabric image retrieval because of the variances of texture pattern sizes, viewing angles and texture densities among the lace fabric images. In this paper, we address the lace fabric image retrieval problem and propose an effective Multi-scale and Rotation Invariant Local Binary Pattern (MRI-LBP) feature, based on the thorough analysis on the characteristics of this kind of image. Specifically, we extract the MRI-LBP feature in three steps. First, the original fabric image is scaled with multiple ratios. Then, we extract the rotation invariant LBP feature using the combination of multiple radiuses under each scale, respectively. Finally, the rotation invariant LBP features of all the scales are concatenated as the final MRI-LBP feature, which is used in the lace fabric image retrieval task. The experimental results show that the proposed MRI-LBP method significantly outperforms other features.