Research Of Fabric Pattern Retrieval System Based On Image Processing

With development of multi-media technology, the demand of people is increasing for dressing. The number of images in everyday life is also growing by leaps and bounds. For large demand of pictures industry, image retrieval has become a very active area of research. How to analyze, store and retrieve these images is an imperative problem. Fabric styles and patterns are various and changing rapidly. There are thousands of patterns stored in sample library of the fabric production companies, and designers create more and more new patterns every day. In traditional manner, fabric patterns need to be retrieved in the sample library by manual classification and numbering. It cannot accurately describe the retrieval request to meet the needs of information query for different customers, and cost a lot of time and labor power. The image retrieval based on computer vision can quickly find the similar existing patterns from sample library according to samples or scanned images, overcome the shortcomings of traditional manual retrieval and improve efficiency.In this paper, according to the fabric pattern image characteristics, fabric pattern retrieval system based on image processing is carried out. Firstly, this paper introduces the fabric image retrieval system background, the selected topic significance and status of this system at home and abroad. Fabric pattern retrieval system based on image processing is discussed with the general framework, key technology and retrieval characteristics and application field. Secondly, fabric pattern retrieval system based on image processing of key technology is introduced in detail, including feature description, similarity measure and evaluation standard of retrieved results. The feature description module contains color, texture and spatial relation feature. Each of extraction features could use a variety of methods. Finally, two kinds of fabric pattern image retrieval algorithms are put forward. One method is fabric pattern retrieval based on global feature(GIST feature). Another is fabric pattern retrieval based on visual bow model. Both methods extract color feature, texture feature and spatial relation feature of fabric pattern image respectively. Extracted features are converted to feature vectors, which is used to calculate the similarity in the sample database. Final results would be shown in the interface according to the similarity order. Difference of two methods is the way to extract spatial relation feature, leading to the difference of retrieval efficiency and results. A large number of experiments show that two algorithms are available in different fields to achieve fabric pattern image retrieval.This paper designed and implemented fabric design retrieval system based on image processing. The system contains display modules, feature extraction modules, feature measure modules. In this paper, each modules is introduced in detail, achieving the display images, based on different requirements of feature extraction and feature vector similarity judgment. Fabric design retrieval system based on image processing adopts the development tools for Visual Studio 2010, making use of MFC providing the simple user-friendly interface. Experiments show that this system is strong and the function runs normally, which could achieve the retrieval for fabric pattern image perfectly.