Collection Of High Precision Corresponding Colors Data Set

Color appearance model is a mathematical transformation model that simulating the complex visual response of human eyes.It can be used to solve the problem of measuring the color appearance of objects in complex environment,and has become the focus of modern color science.The kernel of color appearance model is color adaptation model,which is also applied to color reproduction.At present,the prediction error of four international CIE recognized excellent color adaptation transformation formula is between 4-5 CMC(1:1),And the accuracy of color adaptation transformation formula is difficult to improve,the reason is that the accuracy of the corresponding color data set which the color adaptation transformation formula derived from is about 4,which limits the accuracy of the color adaptation transformation formula.Therefore,the acquisition of high precision corresponding color data sets becomes a key to the development of color appearance models.In this project,the corresponding color data sets are collected by using the reflective manuscript and the monitor,and the memory matching method is used as the experimental method.In order to reduce the memory error in the process of experiment,the training and testing of the memory precision of the color.block were carried out before the corresponding color experiment.In addition,The GUI in MATLAB is used to control experimental procedure throught voice and provide L*,a*,b*block adjustment function,and the color gamut of GUI in each round of experiments can be adjusted according to the results of the previous rounds of experiments,which can reduce the influence slider-adjusting on the experimental precision.Since the color difference between the candidate color blocks that used in the first two rounds,the three or four and the four rounds are different,the corresponding color data sets are sorted into three corresponding color data sets.The analysis results show that the standard deviation of the data set collected in this experiment is less than that of historical similar data sets,which shows that memory training and testing have achieved good results.But there are still some improvements in the experimental design,which need to be further improved in the future research.