基于SVM的彩色扫描仪特征化

研究了基于支持向量机(SVM)的彩色扫描仪特征化方法,分析了采用SVM回归法实现RGB色空间与CIELab色空间非线性变换的可能性。研究结果表明:L,a,b测试值与预期值的回归相关性达到99%以上,其中CIELab色差平均值、最大值和最小值分别为2.314 3,5.791 7和0.507 3;利用SVM回归法可满足扫描仪特征化的精度要求,体现了SVM在小样本拟合中的精度和时间优势。     

基于多元非线性回归模型的彩色扫描仪特征化

研究了基于多元非线性回归模型的彩色扫描仪特征化方法,分析了多元回归的阶数以及输出色空间的选择对非线性多项式转换模型是否存在影响。研究表明:训练组样本特征优化随着阶数增加而增加,测试组样本在高阶数非线性回归下优化精度下降,多项式方程呈现病态,泛化能力下降;回归模型采用RGB到L*a*b*的转换比采用RGB到XYZ的转换,优化精度略高。结果说明,利用多元非线性回归模型可满足扫描仪特征化的精度要求。     

基于数码打样CMYK到L*a*b*色彩空间转换方法的研究

通过对ECI2002标准色靶数码打样样张进行实验数据测试、分析得到其色块网点呈色符合平面规律,并建立起不同K值下的CMYK到L*a*b*色彩空间转换方程,应用△E*ab色差公式进行精度检验后,最终得出基于平面理论建立起来的CMYK到L*a*b*色彩空间转换方程是实现数码打样设备色空间转换的有效方法,可为数码打样中色彩管理色空间转换部分的研究提供更加广泛的理论依据.     

基于最小二乘支持向量机的扫描仪特征化

目的基于最小二乘支持向量机回归(LSSVR),研究扫描仪图像输入设备的特征化方法。方法以ColorChecker SG标准色卡为目标,通过最小二乘支持向量机建立RGB三通道值到CIE Lab色度值的非线性映射模型,采用基于交叉验证的网格搜索确定模型最优参数,优化LSSVR模型,实现彩色扫描仪的色度特征化。结果所建模型的训练集R-squared为0.996,验证集R-squared为0.998,训练集与验证集的CIEDE2000平均色差分别为1.1463,1.2754。结论 LSSVR模型能够较好地实现彩色扫描仪色度特征化,泛化能力较强,此模型可有效地提高彩色扫描仪特征化的精度且计算处理速度更快。     

色度计算方法及其影响因素研究

基于XYZ,CIE L*a*b*颜色空间的色度值和色差计算方法,分析了波长间隔对三刺激值计算的影响以及不同加权表的差别和应用范围,最后研究了CIEL*a*b*空间中色差与光谱反射率差、色差与三刺激值间的逆转现象,并对其在颜色测量仪器校正方面的应用进行了分析。     

印刷品质量检测颜色转换模型研究

目的研究印刷品质量检测中的颜色转换模型。方法基于色靶的测量数据,采用三维查找表法和多项式回归法,分别建立RGB与CIE L*a*b*颜色空间转换模型,并实验比较了2种算法转换模型的精度。结果三维查找表法转换模型的最大色差为3,四面体插值算法精度稍高于三线性、三棱柱和金字塔插值算法,平均色差为0.64;多项式回归算法的精度随着项数的增多而提高,20项的平均色差为2.58。结论三维查找表法转换模型精度高于多项式回归法,效果理想,能判断出CCD采集到的印刷图像是否存在色偏,且随着RGB颜色空间划分越细小,转换精度会越高。     

基于立方体内缩的反向颜色空间转换算法

目的提出一种基于立方体内缩的反向颜色空间转换算法,实现颜色值从CIELab颜色空间到RGB颜色空间的转换。方法选取4096组建模点和512组测试点,以三维查表插值法为理论基础,设计一种基于立方体内缩的方法逐步搜索符合条件的特征点来代替测试点,计算得到测试点反向转换后的RGB值。结果通过CIELAB 1976,CIE94和CIE2000色差公式对算法的转换精度进行评价,计算测试点的Lab值和反向转换得到RGB值对应的Lab值之间的色差。测试点的平均色差分别为2.07,1.53和0.96,大部分色差的分布范围在0~3之间,算法的转换精度较高。测试点反向转换后的结果较为理想,在绿色调区域和亮度较低的红色调区域的精度可进一步提升。结论实验结果表明此算法可精准、快速、有效地实现颜色值从CIELab颜色空间到RGB颜色空间的转换。     

不同光源下样品颜色三刺激值测试的比较分析

针对光学伪装勘察和检测的特殊性,在380～800nm波长范围里,比较了不同颜色样品分别在太阳光和D65光源下三刺激值的差别,采用CIE1976L*a*b*空间分析了2种光源下颜色样品的色差。结果表明,太阳光作为光学伪装勘察和检测的标准光源更加客观、准确。     

基于平面方程理论实现L*a*b*到CMYK色空间的转换与输出

目的 研究 L*a*b*到 CMYK 颜色空间转换建模的思路及方法。方法 在平面方程理论的基础上, 按照K取特征值的思路, 在 Matlab中采用拟合等方法, 求出四色平面方程的相关系数, 建立色空间转换的数学模型。结果 实现了L*a*b*到CMYK的颜色转换, 并根据模型输出了图像, 通过计算色差验证了模型精度达到要求。结论 该模型具有结构相对简单、 运算速度较快的特点, 对于颜色空间转换的研究有一定的参考和借鉴意义。     

基于平面方程理论实现L*a*b*到CMYK转换与输出

目的研究L*a*b*到CMYK颜色空间转换建模的思路及方法。方法在平面方程理论的基础上,按照K取特征值的思路,在Matlab中采用拟合等方法,求出四色平面方程的相关系数,建立色空间转换的数学模型。结果实现了L*a*b*到CMYK的颜色转换,并根据模型输出了图像,通过计算色差验证了模型精度达到要求。结论该模型具有结构相对简单、运算速度较快的特点,对于颜色空间转换的研究有一定的参考和借鉴意义。     

3CCD camera's capability for measuring color differences: experiment in the nearly neutral region

A method to assess the discrimination capability of a camera to measure small color differences is proposed. The method helps to fix the working conditions of the camera and analyzes the reliability of the measurements through comparison with a reference instrument. Attention is paid to the camera's performance in the nearly neutral region of color space. The color differences are calculated using the Commission Internationale de l'Eclairage L*a*b* (CIELAB) deltaEab* and CIE 2000 color-difference formula metrics. The Sony DX-9100P 3CCD camera results are very close to those obtained by the Photo Research PR-715 spectroradiometer. Their absolute discrepancy is lower than the suprathreshold of visual discrimination (0.887 CIELAB unit).     

Performance of recent advanced color-difference formulas using the standardized residual sum of squares index

The standardized residual sum of squares (STRESS) index was used to reevaluate four experimental datasets employed during the development of CIEDE2000, the current CIE recommended color-difference formula. This index enables statistical inferences not achievable by other metrics used commonly for performance evaluation. It was found that CIEDE2000 was statistically superior at a 95% confidence level to either CIE94, the previous recommended equation by the CIE, or the simple Euclidean distance in CIELAB, DeltaE*ab. Recent formulas based on the CIECAM02 color-appearance space and chroma-compressed variants of CIELAB were also evaluated and found to have only slightly reduced performance compared with CIEDE2000. These formulas have the advantage of simplicity and easier interpretation when used for quantifying color accuracy. Finally, each experimental dataset was evaluated separately rather than weight averaged as used during the development of CIEDE2000. Significant differences were found between datasets, suggesting that combining datasets may obscure important differences and that the practice of parameter optimization during formula development using combined data is likely suboptimal.     

Performance of a color-difference formula based on OSA-UCS space using small-medium color differences

An investigation of the color metrics and the complexity of the CIEDE2000 formula shows that CIELAB space is inadequate to represent small-medium color differences. The OSA-UCS (Uniform Color Space) Committee has shown that no space with uniform scale for large color differences exists. Therefore the practical way for color-difference specification is a color-difference formula in a nonuniform space. First, the BFD (Bradford University) ellipses are considered in the OSA-UCS space, and their very high regularity suggests a new and very simple color-difference formula at constant luminance. Then the COM (combined) data set used for the development of the CIEDE2000 formula is considered in the OSA-UCS space, and the color-difference formula is extended to sample pairs with a different luminance factor. The value of the performance factor PF/3 for the proposed OSA-UCS-based formula shows that the formula performs like the more complex CIEDE2000 formula for small-medium color differences.     

关于颜色空间转换的RBF网络动态子空间自动划分辨识方法研究

以RGB与CIEL*a*b*颜色空间转换为例,采用径向基函数(RBF)神经网络,研究了颜色值在不同颜色空间之间的转换。利用基本采样数据集建立了颜色空间转换RBF网络模型,并通过增加样本数据,采用动态规划颜色子空间的方法,提高了模型转换精度。研究结果显示,该方法的转换速度和精度都优于基于动态子空间自动划分的BP神经网络颜色空间转换方法。     

基于密度的凹版专色叠印色光谱预测模型的研究及应用

目的 建立一种基于密度与光谱反射率关系的专色叠印色光谱预测模型。方法 首先建立前景色、后景色密度与其光谱反射率关系;基于此建立叠印色光谱反射率的计算方法;其次对模型进行优化,完成专色叠印色光谱预测模型的建立;再通过凹版印刷实打样30张专色叠印色样张,基于Malab利用色差以及拟合决定系数R²对预测模型进行验证;最后在相同工艺条件下实打样专色叠印色样张进行预测,通过色差完成对模型的精度检验。结果 验证表明,不同专色叠印色在同一网点面积率下与其对应的前景色、后景色密度之间存在较强相关关系,其拟合决定系数R²均大于0.9;模型精度检验中,73.7%的模型预测值与样张实测值色差为0~2.5个CIE1976色差单位,大部分色块色差为0~2个CIE1976色差单位,占总频率的60.54%,整体平均色差为1.92个CIE1976色差单位。结论 本文所建模型具有较高的预测精度,能基本满足对颜色预测的需求。     

基于胞元式 RBF 神经网络的高保真分色模型研究

采用胞元式RBF神经网络模型对七色印刷输出系统构建了分色模型。首先,借鉴颜色空间分区理论将7个主色在整个颜色空间中划分为了6个颜色区域,在每个分区中选取了CIE L*a*b*明度值L上等间隔均匀采样的网点面积率,用于设计建模所需的训练样本,然后对每个分区划分胞元,并且为每个小胞元建立了基于RBF神经网络的分色模型。对于任意给定的要复制的目标色,利用提出的胞元搜索算法确定其所在的胞元位置后,使用相应的神经网络模型进行分色预测。实验结果表明,该分色算法能够达到较高的分色精度,可以满足高质量彩色复制的要求。     

Maximum likelihood difference scaling of image quality in compression-degraded images

Lossy image compression techniques allow arbitrarily high compression rates but at the price of poor image quality. We applied maximum likelihood difference scaling to evaluate image quality of nine images, each compressed via vector quantization to ten different levels, within two different color spaces, RGB and CIE 1976 L*a*b*. In L*a*b* space, images could be compressed on average by 32% more than in RGB space, with little additional loss in quality. Further compression led to marked perceptual changes. Our approach permits a rapid, direct measurement of the consequences of image compression for human observers.     

Chromaticity-discrimination thresholds with aperture and object colors: experimental results and predictions of some recent color-difference formulas

Threshold-discrimination ellipses were obtained by three normal observers, at five color centers, by the use of two experimental devices, which provided light-surface colors, and the same method and experimental conditions. The ellipses obtained for each center in both devices were quite similar when the sample distributions were appropriate and slightly smaller for object color than for aperture color. The ellipses predicted by three recent color-difference formulas based on the CIE uniform color space (CIELAB) were compared with those experimentally obtained by us as well as by other researchers who analyzed a greater number of color centers. The color-difference formula proposed by the CIE Technical Committee 1-29 [Color Res. Appl. 18, 137 (1993)] provides the best prediction of the semiaxis relationship for all the experimental datasets used, confirming their good performance in previous works [Appl. Opt. 33, 8069 (1993)], although the differences with respect to the experimental results are higher than those attributable to the interobserver variability.