Minority Costume Image Retrieval by Fusion of Color Histogram and Edge Orientation Histogram

Abstract

It has very important practical significance to analyze and research minority costume from the perspective of computer vision for minority culture protection and inheritance. As first exploration in minority costume image retrieval, this paper proposed a novel image feature representation method to describe the rich information of minority costume image. Firstly, the color histogram and edge orientation histogram are calculated for divided subblocks of minority costume image. Then, the final feature vector for minority costume image is formed by effective fusion of color histogram and edge orientation histogram. We have evaluated the performances of the proposed algorithm on self-build minority costume image dataset, and the experimental results show that our method can effectively express the integrated feature of minority costume images, including color, texture, shape and spatial information. Compared with some conventional methods, our method has higher and stable retrieval accuracy.

Exploiting Best Practice of Deep CNNs Features for National Costume Image Retrieval

“…We compare our proposed scheme with some recent methods using lowlevel visual features in terms of AP while considering 12 retrieved images. Among them, single features such as LBP, HoG and color histogram (CH) are used, and multi-features such as color histogram fused with edge orientation histogram [26] and multi-features fusing color, texture and shape in multi-regions are also used [33].…”

“…There are some related works about national costume image retrieval. Xumei Shen et al [26] proposed an image feature representation method to describe the rich information of national costume image based on color histogram and edge orientation histogram. The method was verified in their self-built national costume image dataset, which contained six national costume images including those from Bai nationality, Jingpo nationality, Hani nationality, Miao nationality, Bouyei nationality and Va nationality.…”

“…The dataset [21] used in our experiments is built by [26], which contains six classes of national costume images including the Bai nationality, Jingpo nationality, Hani nationality, Miao nationality, Bouyei nationality and Va nationality. Each nationality has 100 images, and the total number of images is 600.…”

“…The CBIR technology has successfully been applied in many areas, from natural image retrieval to special field of image retrieval, including exhibition and museum management, construction and engineering design, remote sensing and geographical resources management, GIS, weather forecast, cultural heritage image retrieval, Chinese Ink Paintings image retrieval and national costume image retrieval, etc. [2,14,25,26,33].…”