Color image retrieval based on textural and chromatic features

Image content-based retrieval has been widely explored. Image contents like colors, shapes, textures and patterns are extracted to represent the image feature for indexing and retrieval in an image database. Researchers have been focusing on the efficiency of indexing and searching algorithms. We propose to use the LHS color coordinate system, which is based on the Maxwell triangle of the RGB space. The luminance component is used for textural analysis. The hue and saturation components are used for chromatic analysis. Given an image, both features are used to retrieve similar images. Experiments are conducted. It is shown that image retrieval using the HS space has better precision and recall than the image retrieval using the RGB space. Moreover, it also shows that retrieval through the chromatic features has better performance than retrieval through the textural features