Approximating the CIECAM02 color appearance model by means of neural networks  

作　　者：柴冰华 廖宁放 赵达尊 

机构地区：[1]National Lab of Color Science and Engineering

出　　处：《Chinese Optics Letters》2004年第11期637-639,共3页中国光学快报（英文版）

基　　金：ThisworkwassupportedbytheNationalNaturalSci-enceFoundationofChina(No.60278022)theNat-uralScienceFoundationofBeijing(No.4032016).

摘　　要：An artificial neural network used to realize the approximating problem of the color appearance model (CAM) CIECAM02 in color management is demonstrated. GretagMacbeth ColorChecker Charts, which now axe widely used in calibration of digital camera, are chosen as samples to implement the forward and reverse color appearance models. When the predictive results are evaluated, for forward model, the output color appearance space is converted to the uniform color space based on CAM and is evaluated, while for reverse model, because the prediction precision is insufficient, we try to convert the color appearance space, which is the cylinder space, to the cube space similar to the red, green, and blue (RGB) space, and the results show that the precision is obviously improved.An artificial neural network used to realize the approximating problem of the color appearance model (CAM) CIECAM02 in color management is demonstrated. GretagMacbeth ColorChecker Charts, which now axe widely used in calibration of digital camera, are chosen as samples to implement the forward and reverse color appearance models. When the predictive results are evaluated, for forward model, the output color appearance space is converted to the uniform color space based on CAM and is evaluated, while for reverse model, because the prediction precision is insufficient, we try to convert the color appearance space, which is the cylinder space, to the cube space similar to the red, green, and blue (RGB) space, and the results show that the precision is obviously improved.

关 键 词：Color computer graphics COLORIMETRY Mathematical models Neural networks 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]