Patterned fabric image retrieval using color and space features

Image retrieval has been an active research topic in the last decade. As one of the promising approaches, color histogram based image retrieval has been attracted by many researchers. However, it is sensitive to noisy interference and lost any spatial information. To overcome drawbacks, the algorithm combined weighted color histogram with image segmentation is proposed for patterned fabric images in this paper. Firstly, the patterned fabric images are preprocessed using image enhancement. Secondly, color and spatial information features are respectively extracted by weighted color histogram and image segmentation method, which are transformed into feature vectors. Finally, the similarity between objective image and image database is calculated with Euclidean distance. The retrieval results are displayed in decreasing order of similarity. Experimental results, including comparisons with color histogram algorithm and weighted color histogram algorithm, demonstrate the effectiveness of proposal for patterned fabric images. © 2015 Binary Information Press & Textile Bioengineering and Informatics Society.