目的 为实现调配的专色与目标色尽可能一致，对光谱匹配算法生成的初始配方进行修正，从而使均方根误差和色差均较小。方法 选取潘通色为目标色，对比分析光谱修正和三刺激值修正算法对色差和谱差的影响，选取实际生产专色为目标色，进一步将配色效果较优的算法应用到基于白卡纸和光柱镭射纸的配色中。结果 基于Kubelka-Munk单常数理论的光谱匹配算法进行颜色预测，白卡纸配色中初始配方、光谱修正和三刺激值修正的色差的均值分别是4.1834，3.6202，1.0300;谱差的均值分别是0.0804，0.0364，0.0303。光谱修正效果不理想，三刺激值修正的色差都小于3，符合生产要求范围。将三刺激值修正算法应用到白卡纸和光柱镭射纸专色中，白卡纸配色中初始配方、三刺激值修正的色差平均值分别是2.7857和1.0850，谱差的均值分别是0.0596和0.0323。8组修正配方的色差皆降至3以下。基于光柱镭射纸配色的初始配方、三刺激值修正的色差平均值分别是3.2861和1.2971，谱差的均值分别是0.0450和0.0165。结论 在印刷专色配色中，通过光谱匹配算法预测并结合三刺激值修正，可有效降低预测色和目标色之间的色差和谱差。

In order to achieve the consistency of the allocated spot color and the target color as much as possible, the initial formula generated by the spectral matching algorithm is modified to achieve a small root mean square error and color difference. Firstly, Pantone color is selected as the target color. The influence of spectral correction and tristimulus value correction algorithm on color difference and spectral difference is compared and analyzed. The actual production spot color is selected as the target color. Furthermore, the algorithm with better color matching effect is applied to the color matching based on white cardboard and laser paper. Color prediction is carried out in spectral matching algorithm based on Kubelka-Munk single constant theory. The mean values of the initial formula, spectral correction and tristimulus correction are 4.1834, 3.6202 and 1.0300 respectively. The mean values of spectral differences were 0.0804, 0.0364 and 0.0303 respectively. The result of spectrum correction is not ideal. The color difference of three stimulus value correction is less than 3, which meets the production requirements. The results show that the average chromatic aberration of initial formula and tristimulus correction is 2.7857 and 1.0850 respectively. The mean values of spectral differences are 0.0596 and 0.0323 respectively. The chromatic aberration of the eight modified formulas are all reduced to less than 3. Based on the original formula of laser paper color matching, the average value of color difference corrected by tristimulus value is 3.2861 and 1.2971 respectively. The mean values of spectral differences are 0.0450 and 0.0165 respectively. The spectral matching algorithm predicts and combines the tristimulus value correction in the color matching of the printed spot color, which can effectively reduce the color difference and spectral difference.