Color Correction for Multi-view Images Using Relative Luminance and Chrominance Mapping Curves

In order to capture 3D scenes, a multi-view camera consisting of two or more cameras is widely used; however, color consistency among views is not guaranteed in many situations. In this paper, we design relative mapping curves with consideration of the properties of luminance and chrominance components to improve the consistency. The input images are categorized into source and reference views. We convert their color domain to the YUV color space, and estimate coefficients in the mapping curves by analyzing correspondences between the two views. After that, we generate lookup tables and convert the color distributions of the source views. From the experimental results, we confirm that our proposed method improves the visual quality of multi-view images and reduces Euclidean distances in the CIELab color space among views.