Evaluation And Optimization Of Lightness Prediction Models

With the development of science and technology and the improvement of people's living standards, the field of color science continue to expand. Color Science efforts to solve the problem that how to accurately predict human visual perception of color. Lightness as one of the three attributes of color, is an important scientific research. The main research of this paper is evaluation and optimization of the lightness prediction models. Including an objective assessment of the predictive accuracy of lightness prediction models, find the optimal prediction model; build a new algorithm to discoloration, the optimal lightness prediction model used in discoloration and build lightness dataset of self-luminous colors.Firstly, evaluat predictive accuracy of three lightness prediction formula proposed by Wyszecki, Nayatani and Fairchild and four lightness prediction model in color appearance model include Nayatani, Hunt, CIECAM97s and CIECAM02. After objective data analysis found that the prediction formula has higher accuracy than the lightness color appearance model. The most accurately predict model is the third model which Fairchild proposed, but it has bad prediction in the light tone area, especially in light yellow, magenta and blue tone.In order to combine our study to practice, this article will put porward a new discoloration algorithm applied to image. The best lightness predictive model as the new lightness algorithm. Commonly used method is abandon color information directly, only leaving the physical lightness value, this paper will use new lightness value calculate by third lightness model which Fairchild proposed instead of the leaving lightness value. Found that the new methid is unsatisfactory on light tone and over-saturated colors, but it works well on shadow and middle tone.The subject also study the establishment lightness datesets of a self-luminous color on account of the increasing proportion of self-luminous colors in daily life. Through VAC visual lightness matching method to obtain achromatic color lightness value which matching chromatic color in lightness, processing experiments data, remove the gross error, get self-luminous color brightness datasets.Analyze the datesets found that the strongest H-K effect color is blue, the weakest color is yellow in self-luminous colors gamut.Depth study in lightness prediction of chromatic color, play a role in promoting development of lightness prediction model and color appearance model.