面向高保真印刷的特征光谱提取与匹配方法研究

快速、准确的进行光谱重建是高保真印刷复制研究的难点。针对多个基色参与复制造成颜色在光谱表征和油墨配比之间转换模型复杂的问题，提出了一种用于高保真印刷的特征光谱提取和匹配方法。该方法利用光谱的导数变换对印刷基色油墨的特征光谱波段进行了筛选和提取，在此基础上提出了特征光谱的多阈值编码算法;针对高保真印刷中的分区判断问题，利用导数平均特征光谱表示各分区特征，并通过目标光谱与各分区的导数平均特征光谱的光谱匹配算法，解决了高保真印刷中的分区判断问题。实验结果表明，提取的特征光谱波段能够显著代表各个基色的光谱特征，基于特征光谱的色彩转换模型精度高于基于全波段的色彩转换模型;光谱匹配算法在判断分区上达到了很高的精度，极大的提高了效率，具有较高的实用性。

Research on Characteristic Spectrum Extracting and Matching for High-Fidelity Reproduction

Reconstructing the spectrum rapidly and accurately is the key to the research on high-fidelity reproduction. A characteristic spectrum extracting and matching method for high-fidelity printing is proposed aiming at the problem of complex conversion between spectrum and ink combination caused by multi-color. The method filters and extracts feature bands of primary ink through derivative spectrum, and a characteristic spectrum multi-threshold coding method is proposed. Considering the problem of subarea judgment in hi-fi printing, an average derivative spectrum is taken as characteristic spectrum of each subarea and a spectrum matching method between target spectrum and average derivative spectrum of sub-spaces is proposed. The results show that the feature bands extracted can represent spectral characteristic of primary color significantly and the precision of color conversion model based on feature bands is higher than the model based on full bands. The spectrum matching method can achieve a high accuracy in sub-space judgments and greatly improve the efficiency of color convention. The spectrum extracting and matching method has the high practicability.