Automatic pattern extraction of jacquard fabric based on texture filtering and color clustering

This paper proposes a pattern extraction method based on texture elimination filtering algorithm and density peak clustering algorithm in order to improve the accuracy of jacquard fabric pattern extraction and eliminate the interference of woven fabrics structure on the extraction result. Firstly, the fabric weave texture region was separated from pattern edge structure by the statistics of regional gradient information of pictures. Secondly, the edge structure of patterns obtained by non-maximum suppression was used as the guidance image of rolling guidance filter to achieve fabric texture filtering by multiple iterative computations. Finally, the images were transported to CIELab color space from RGB color space, and density peak clustering algorithm was utilized to fast segment the fabric colors to gain the pattern extraction results. The experimental results show that the proposed method can fast and precisely segment and extract jacquard fabrics.