Color correction using explicit illumination models, color and registered range

This work focuses on correcting color images illuminated by one or more light sources to improve the results of segmentation and feature extraction. First, we linearize the RGB data to make it conform to the assumptions made by many image processing algorithms. Then, we also eliminate intensity variation due to the illuminant with respect to distance and shading produced by surface orientation changes. We introduce a novel calibration procedure for estimating the finite distance and direction of the light source from points in the scene. We compare segmentations of original color images with that of their color-corrected images to evaluate the effectiveness of the methodology