基于两级颜色通道的图像高保真分色研究

项目名称： 基于两级颜色通道的图像高保真分色研究

项目编号： No.61301231

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 无线电电子学、电信技术

项目作者： 朱明

作者单位： 河南工程学院

项目金额： 24万元

中文摘要： 本项目创新地提出了基于两级颜色通道的图像高保真分色方法。新方法一方面通过扩充基色数量来扩大分色图像的色域范围，另一方面通过增加对应基色的浅色级通道来提高图像阶调层次的复制质量。新方法试图将图像分色由传统单级模式转变为两级模式，以提高颜色复制的保真度。新方法拟采用的研究方案是：①首先对两种传统图像分色方法进行改造使其适用于多通道图像分色，即使用基于色域分区和胞元划分改造的纽介堡印刷呈色模型和基于分区特征样本集的经验模型来完成从色度值到基色值的第一级分色转换；②然后通过大量实验找到基色通道和相应浅色通道之间的匹配关系，并确定各浅色和相应基色通道之间的取值阈限，构建从基色通道到所有颜色通道（含浅色）的第二级分色模型。本项目研究成果不仅可直接应用于传统高保真印刷复制行业，在彩色喷墨印刷和纺织品喷墨印花领域也有着重要的应用前景，本项目对我国自主研发数字印刷和数字印花的输出控制系统具有重要意义。

中文关键词： 高保真分色；光谱分色；光谱降维；空间色域映射；引导滤波

英文摘要： The high-fidelity color separation method for color image based on Two-Level color channels is proposed creatively in this project. The new method can expand the printing color gamut by increasing the number of primary colors on one hand; on the other hand the new method can greatly improve the tone-reproduction quality of color image by adding the light color channels of the corresponding primary color. The proposed research program of the new method is as follows: ①Two types of conventional image color separation method will be transformed in order to make them suitable for the multi-channels color separation of image. The Cellular Neugebauer model based on gumat subarea transforming and the empirical model based on the characteristic color sample collection for color subareas will be used to establish the first-lever color separation conversion from chromatic value to primary color value. ②The match function between the primary color channel and the corresponding light color channel will be founded through a lot of experiments. The boundary value between each light color channel and the corresponding primary color channel will be confirmed. So the second-level color separation model (from primary color channels to all channels that contain light color channels) will be established. The research findings of th

英文关键词： High-fidelity color separation；Spectral color separation；Spectral dimensionality reduction；Spatial gamut mapping；Guided filter