A comparative study on the difference of color space conversion based on table look-up method and neural network

Abstract

Objective: To study the conversion model of RGB color space to CIE1976 L*a*b* color space with higher accuracy, which provides some value for the fields of computer color matching, color detection, and color reproduction. Methods: The method of three-dimensional look-up table method and back propagation (BP) neural network are proposed, and the effect of the model built under the two methods is evaluated under the calculation of CIE1976 L*a*b* and CIE2000 color difference by NCS color card data. Results: Under the calculation of CIE1976 L*a*b* color difference, the average color difference under the four interpolation methods of the three-dimensional look-up table is within 3, and the average color difference of the BP neural network algorithm is 1.8720. Under the calculation of CIE2000 color difference, the average color difference of the four interpolation methods of the three-dimensional look-up table drops within 1, and the average color difference of the BP neural network also shows a downward trend, and the specific value is 1.3449. Conclusions: According to the result obtained by the research method, the color difference of the tetrahedral interpolation method is the smallest among the four interpolation methods of the three-dimensional look-up table method under both color difference formulas. Whether it is the three-dimensional look-up table method or the BP neural network, the model obtained by the CIE2000 color difference formula is the best. In general, for the two methods, the BP neural network method is more convenient and faster, and the color difference effect is also desirable.