At presents, the technique of multi-feature combination of trademark image retrieval is an increasingly popular branch of CBIR. In this paper, it's proposed that multi-feature combination based on color and shape is applied to trademark image retrieval. Firstly, RGB color space is converted to HSV color space, which complied with human visual perception better. Then color moments, which can describe the spatial distribution of color, was extracted. Thirdly, a color image is transformed to a gray image, on which low orders Hu moments and eccentricity were extracted. Finally, being normalized by means of Gaussian, the technique of multi-feature combination is applied to retrieve trademark images. Good results have been obtained in the following experiments which prove that the multi-feature combination way is better than other single-feature ways.
展开▼
