<title>New HSL and HSV color spaces and applications</title>

Abstract

The color classification applied for large documents requires simple color transformations that enables the implementation of high speed algorithms with reasonable performance. The use of HSV and HSL color spaces in such applications is limited due to their perceptual non- linearity, even if these spaces are attractive due to the simplicity of their definition and transformation. This paper proposes new relationships of the HSL and HSV color spaces based on the replacement of the lightness and brightness definitions, with new relationships based on more perceptual lightness and brightness. The new HSL and HSV color spaces are visualized in three dimensions and the shapes are compared with the old regulate spaces. These spaces remain device dependent as the conventional HSV and HSL spaces, but the distribution of color is more suitable for some application like color clusterization. The new defined color spaces are asymmetric. It is shown how the new HSV and HSL color spaces simplify and improve the accuracy of a clusterization process. The results of clusterization process are compared in the processed image area as well as in the 3D histogram diagram. The color spaces are used for clusterization of colormaps, with application to automatic color classification in large documents (maps, blueprint documents, etc.). Results of clusterization process for blueprint documents are provided for the new introduced HSL and HSV color spaces.