色貌模型的人工神经网络方法的研究

色貌模型(CAM)主要解决不同观察条件、不同背景和不同环境下的颜色真实再现问题。采用人工神经网络(ANN)的方法来实现目前最新的色貌模型CIECAM02的预测,包括正向预测(从色度参数到色貌属性参数)和逆向预测(从色貌属性参数到色度参数),应用自然色系统(NCS)中的部分色样作为神经网络的训练和测试样本。由于正向输出色貌属性参数空间不是均匀的,对于网络预测精度用特殊方法评估,而对于逆向模型则可直接利用LAB色差公式评价。测试的结果表明:用神经网络对CIECAM02模型的预测达到了较高的精度。

Color appearance model based on artificial neural network

A method of predicting the newest color appearance model(CIECAM02) results using artificial neural network(ANN) is presented, including forward model(from colorimetric attributes to color appearance attributes) and reverse model(from color appearance to colorimetric attributes). A part of color charts of natural color system(NCS) are selected as training and testing samples for the networks. Since the space of color appearance attributes is not uniform, a special method is used to evaluate the prediction precision in forward network, while for reverse operation the color difference formula in LAB space can be used directly. The testing results show that the CIECAM02 model prediction using BP neural network may have satisfied precision.