人工神经网络在色彩变换中的应用
Application of Neural Network in Color Conversion
-
摘要: 给出了一种针对数字彩色硬拷贝的输出设备,建立了由CIE色度值到设备控制值色彩变换的神经网络方法。基于蒙版方程的颜色传递机理,由色度密度变量实现色彩变换。结果表明,该色彩变换对各种色调均有较明显的校正作用,且保持了高饱和色的饱和度。Abstract: A neural network method for the color conversion from CIE colorimetric values to RGB control signals for color digital hardcopy devices is presented.Based on the color transfer mechanism of the making equations,the color conversion can be performed by introducing the colorimetric density variables.A BP neural network is utilized to achieve the three-dimensional transformation from the synthetic to analytical colorimetric densities.And then the analytical colorimetric densities are translated to RGB values by three one-dimensional computations.Experiments were carried out on several color devices.Compared with the conventional 3D_LUT color conversion,the network method had a close mean and much more lower maximal errors.