The method for perceptual color correction
Keywords:
color correction, CIEdE2000, color correction transformations, perceptual color correction, color difference formula.Abstract
Introduction: Color correction transformations are widely used in the graphics processing pipeline, for example in sensor characterization, white balance, artistic processing, gamut mappings of displays, and so on. However, the current methods of coordinate transformation between different color spaces do not sufficiently take into account the human visual properties, which leads to color reproduction errors. Purpose: To develop a method for the estimation of color correction parameters taking into account human visual perception properties. Results: The proposed method consists of two stages. At the first stage the estimation of color correction transformation parameters is carried out by the least squares method. At the second stage the obtained estimation is specified by means of the numerical minimization of the target function constructed on the perceptual criterion (based on the CIE DE 2000 formula for calculation of color differences). Experimental study has shown that this method allows to significantly increase the accuracy of color reproduction. The average error is reduced by 10 to 50% depending on the color correction model used. In addition, the number of colors with significant for human perception color reproduction errors is reduced by 10 to 15%. The method can use different models of color correction transformations, which makes it universal and gives wide possibilities of implementation in the pipeline of graphic information processing. Practical relevance: The developed method for the estimation of color correction parameters taking into account human visual perception properties can be applied for the synthesis of TV systems in tasks, where accurate color rendering is important for the user, for example, in digital photography, telemedicine, etc.