A Study on Improvement of Prediction Accuracy of Illumination Estimation by Combining Multiple Estimation Methods

Abstract

Illumination estimation is one of the basic technologies in the field of image recognition . In this paper, we propose a method to improve the estimation accuracy by selecting representative illumination estimation method s and learning the estimated illuminations using linear SVR (Support Vector Regression). We have evaluated the estimation accuracy of the several methods with hyperspectral images rendered at multiple color temperatures, and have confirmed that the estimation accuracy was improved.