A novel plaid fabric image retrieval method based on hand- crafted features

Abstract

Fabric image retrieval can obtain technological parameters of existing similar fabric images to guide production, saving lots of labor and material resources in the proofing process. This paper presents a novel image retrieval method for plaid fabrics based on hand-crafted features. First, local texture descriptors are extracted by the local binary pattern on the separated images which are processed by Fourier transform. Global texture descriptors are extracted by Scale-Invariant Feature Transform (SIFT) and Vector of Locally Aggregated Descriptors (VLAD). Second, color moments with image partitioning are extracted to characterize spatial color information of plaid fabric images. Third, the similarities of the three features are calculated and fused by the weight assignment to realize the plaid fabric image retrieval. To verify the proposed method, over 30,000 plaid fabric samples are collected from the factory to build the image database as the benchmark. Experiments show that precision and recall at rank five reach 70.6% and 62.6%, respectively, and mAP reaches 0.690. Results prove that the proposed method is feasible and effective, which can provide references for the worker and help the fabric manufacturing factory save lots of labor and material resources.