A new quantitative evaluation metric for color correction algorithm

Abstract

This paper presents a new quantitative evaluation metric called the saturated pixel detection analysis (SPDA) to evaluate the effectiveness of color correction algorithms widely used to address the color constancy problem. Unlike most existing evaluation metrics, an innovative approach is introduced into the proposed SPDA analysis to identify the undesirable saturated image pixels. This is achieved by comparing the pixel distribution of a given image represented in the CIE Lab color space before and after the color correction process. Extensive experiments show that the results of SPDA metric are consistent with those obtained from the visual inspection results. This suggests that the SPDA metric is a more reliable quantitative metric used to evaluate the performance of the color correction algorithms as compared with other existing quantitative evaluation metrics.