Research And Realization Of Image Retrieval Technology Of Minority Costume Based On Content

The minority costume is the important symbol of the ethnic group identification and the precious wealth of the Chinese nation. However, influenced by economic globalization and social modernization, various minority costume cultural traditions have been rapidly disappearing, which has prompted people to value the heritage and protection of minority costume culture. Each ethnic group has its own apparel style with distinct ethnic characteristics(Costume of same nation has a uniform color, fabric texture and pattern), influenced by different culture, traditions, and geographical feature, which make them more advantageous than ordinary natural images in image processing. Therefore, the main purpose of this paper is to analyze the visual features of minority costume image, which will make contribution to the protection and inheritance of minority costume culture. The main research contents are as follows:(1) We construct a minority costume image dataset. The dataset itself is a digital way to protect the minority costume culture. At the same time, the dataset provides data sources for our experiments on retrieval algorithms and future researches on minority costume image retrieval.(2) We study content based image retrieval technology theoretically and then we use classical image feature extraction algorithms to experiment on minority costume image retrieval. Through comparison experiment, we can analysis the visual features of minority costume image. The experimental results show that the color histogram and the histogram of gradient obtain better performance than others in minority costume image retrieval.(3) In view of the two visual feature extraction algorithms, color histogram and the histogram of gradient, obtain better performance than others in minority costume image retrieval, considering that the histogram of gradient has a high dimension and its object is gray image, this paper presents a novel image feature representation method——fusion of color histogram and colored edge orientation histogram. Experimental results show that our method can describe the comprehensive visual feature of minority costume image with lower dimension, including color, texture, shape and spatial information. Through comparison experiment, we also find that our method obtains a higher retrieval accuracy.(4) We design and implement a web-based image retrieval system for Yunnan minority costume image using the proposed feature extraction method for minority costume image——fusion of color histogram and color edge orientation histogram. This web-based system, namely Yunnan minority costume image retrieval system, is realized by using ASP. Net technology under Microsoft Windows. It is a platform for sharing the minority costume image resources and it provides a modern means for the dissemination of minority costume culture.