This paper addresses the critical challenge of illumination estimation in image processing, focusing on robotics, where robust environmental perception under various lighting conditions is essential. Existing methods such as RGB histograms or GIST descriptors are sensitive to illumination changes and often fail in complex environments. This study presents a novel method to estimate the illuminant and light direction in images by utilizing the Wasserstein distance based on optimal transport theory. Through experiments on various images such as indoor scenes, black and white photographs, and night images, we demonstrate that it detects the main light source and estimates its direction with better performance than existing statistical methods in complex lighting environments. We suggest potential applications such as light source localization, image quality assessment, and object detection enhancement, and propose future research to improve accuracy by integrating adaptive thresholding and gradient analysis, and to provide a scalable solution to real-world lighting problems.
