AUTHOR=Wang Fei , Wang Wei , Wu Dan , Gao Guowang 

TITLE=Color Constancy via Multi-Scale Region-Weighed Network Guided by Semantics

JOURNAL=Frontiers in Neurorobotics

VOLUME=Volume 16 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/fnbot.2022.841426

DOI=10.3389/fnbot.2022.841426

ISSN=1662-5218

ABSTRACT=In the task of color constancy, estimating the illumination of the scene is the most important. However, due to the unknown light source and the influence of external imaging e nvironment, the estimated illumination is prone to color ambiguity. In this paper, a learning based multi-scale region weighed network guided by semantics(MSRWNS) is proposed to estimate the illuminated color of the light source in a scene. Learn from the human brain’s processing of color constancy, we use image semantics and scale information to guide the process of illumination estimation. First, put the image and it’s semantic into the network, and the region weights of the image at different scales is obtained; then through the weight pooling layer(WPL), the illumination estimation on each scale is obtained; The global illumination is calculated by weighting on each scale. The results of extensive experiments on Color Checker and NUS 8-Camera datasets show that the proposed approach is superior to the state-of-the-art methods both in efficiency and effectiveness.