Relating reflectance spectra space to Munsell color appearance space

The goal is to construct a simple model relating the conceptually defined Munsell color space to a physical representation of the relationship among the reflectance spectra obtained from the color chips comprising the Munsell color atlas. In the model both the Munsell conceptual system and the transformed reflectance spectra are shown to be well represented in Euclidean space, and the two spaces are related by a simple linear transformation. A practical implication is that the method allows one to compare the location of an empirical reflectance spectrum with the aiming point in the conceptual structure.     

Optimal sampling of color spectra

We explain a method for calculating the optimal sampling interval of color spectra. The 1269 measured Munsell matt reflectance spectra set is used as the test set. The effect of light sources on the required sampling interval with different types of spectra is studied. It is shown that a 20 nm interval is enough for the smooth Munsell set alone, but 10 nm is not enough for the same set matched with a fluorescent light source. However, 5 nm is shown to be enough in most situations.     

样本特征对光谱图像重构影响的研究

目的以光谱图像作为检测样本讨论不同训练样本数量、分布对光谱图像重构的影响。方法选择ColorCheckerSG(140色)和ColorCheckerColorRenditionChart(24色)以及Munsell(1269色)等3种色卡作为训练样本,对其光谱反射率进行主成分分析,利用提取的主成分对光谱图像进行重构。结果采用ColorChecker Color Rendition Chart(24色)色卡的7个主成分重构光谱图像对图像的再现精度最高,其色差比其他2种色卡小,且最大色差小于3。结论在同一重构条件下,光谱图像的重构精度并不随着训练样本数量增多以及分布范围增大而提高,3种训练样本对红紫色的重构精度相对较低。     

Nonnegative color spectrum analysis filters from principal component analysis characteristic spectra

Nonnegative color analysis filters are obtained by using an invertible linear transformation of characteristic spectra, which are orthogonal vectors from a principal component analysis (PCA) of a representative ensemble of color spectra. These filters maintain the optimal compression properties of the PCA scheme. Linearly constrained nonlinear programming is used to find a transformation that minimizes the noise sensitivity of the filter set. The method is illustrated by computing analysis and synthesis filters for an ensemble of measured Munsell color spectra.     

不同算法模型对光谱重构精度的影响

目的研究光谱颜色复制中原稿图像的光谱信息,并对目标色的光谱反射率进行重构,探究影响重构光谱精度的因素。方法通过选取Munsell Color Matt(1269色块)和Color Checker Classic(24色块)2种色卡作为光谱反射率数据样本,建立不同的主成分分析线性重构模型,选取不同的基向量个数分别重构光谱,并对其精度进行评价,取Classic色卡模拟多光谱图像中重建光谱反射率的目标色,研究比较光谱重构模型和基向量数目对重构精度的影响。结果实验表明,降维模型1最终恢复的数据在RMSE,GFC和色差上均优于模型2,随着基向量数目的增加,2种降维模型差距在减小,当基向量数目达到13以后,2种模型基本没差异。结论文中提到光谱重建模型1和7个基向量是重构光谱图像的最佳方案。     

基于双层动态筛选训练样本的光谱重建算法

目的研究光谱重建过程中训练样本筛选方法对光谱重建精度的影响。方法利用违逆的方法对测试样本Munsell样本和Color Checker SG样本进行光谱重构,训练样本分别选择未经筛选的Munsell样本集、经过动态聚类筛选的和经过文中提出的双重动态筛选的Munsell样本集,然后比较3种样本筛选方法得到的光谱重构精度。结果实验结果表明,经过双层动态筛选的训练样本重构精度无论是均方根误差(RMSE)、拟合优度(GFC)还是不同光源下(A,D50和F2)的色差,明显高于动态聚类分析的样本和未经筛选的样本。结论提出了一种新的样本筛选方法,该筛选方法效果良好,具有一定的先进性。     

Removal of analyte-irrelevant variations in near-infrared tissue spectra

This paper describes mathematical techniques to correct for analyte-irrelevant optical variability in tissue spectra by combining multiple preprocessing techniques to address variability in spectral properties of tissue overlying and within the muscle. A mathematical preprocessing method called principal component analysis (PCA) loading correction is discussed for removal of inter-subject, analyte-irrelevant variations in muscle scattering from continuous-wave diffuse reflectance near-infrared (NIR) spectra. The correction is completed by orthogonalizing spectra to a set of loading vectors of the principal components obtained from principal component analysis of spectra with the same analyte value, across different subjects in the calibration set. Once the loading vectors are obtained, no knowledge of analyte values is required for future spectral correction. The method was tested on tissue-like, three-layer phantoms using partial least squares (PLS) regression to predict the absorber concentration in the phantom muscle layer from the NIR spectra. Two other mathematical methods, short-distance correction to remove spectral interference from skin and fat layers and standard normal variate scaling, were also applied and/or combined with the proposed method prior to the PLS analysis. Each of the preprocessing methods improved model prediction and/or reduced model complexity. The combination of the three preprocessing methods provided the most accurate prediction results. We also performed a preliminary validation on in vivo human tissue spectra.     

Analytic solution for separating spectra into illumination and surface reflectance components

The measured light spectrum is the result of an illuminant interacting with a surface. The illuminant spectral power distribution multiplies the surface spectral reflectance function to form a color signal--the light spectrum that gives rise to our perception. Disambiguation of the two factors, illuminant and surface, is difficult without prior knowledge. Previously [IEEE Trans. Pattern Anal. Mach. Intell.12, 966 (1990); J. Opt. Soc. Am. A21, 1825 (2004)], one approach to this problem applied a finite-dimensional basis function model to recover the separate illuminant and surface reflectance components that make up the color signal, using principal component bases for lights and for reflectances. We introduce the idea of making use of finite-dimensional models of logarithms of spectra for this problem. Recognizing that multiplications turn into additions in such a formulation, we can replace the original iterative method with a direct, analytic algorithm with no iteration, resulting in a speedup of several orders of magnitude. Moreover, in the new, logarithm-based approach, it is straightforward to further design new basis functions, for both illuminant and reflectance simultaneously, such that the initial basis function coefficients derived from the input color signal are optimally mapped onto separate coefficients that produce spectra that more closely approximate the illuminant and the surface reflectance for any given dimensionality. This is accomplished by using an extra bias correction step that maps the analytically determined basis function coefficients onto the optimal coefficient set, separately for lights and surfaces, for the training set. The analytic equation plus the bias correction is then used for unknown input color signals.     

Multivariate curve resolution of wavelet and Fourier compressed spectra

The multivariate curve resolution method SIMPLe to use Interactive Self-Modeling Mixture Analysis (SIMPLISMA) was applied to Fourier and wavelet compressed ion-mobility spectra. The spectra obtained from the SIMPLISMA model were transformed back to their original representation, that is, uncompressed format. SIMPULSMA was able to model the same pure variables for the partial wavelet transform, although for the Fourier and complete wavelet transforms, satisfactory pure variables and models were not obtained. Data were acquired from two samples and two different ion mobility spectrometry (IMS) sensors. The first sample was thermally desorbed sodium gamma-hydroxybutyrate (GHB), and the second sample was a liquid mixture of dicyclohexylamine (DCHA) and diethylmethylphosphonate (DEMP). The spectra were compressed to 6.3% of their original size. SIMPLISMA was applied to the compressed data in the Fourier and wavelet domains. An alternative method of normalizing SIMPLISMA spectra was devised that removes variation in scale between SIMPLISMA results obtained from uncompressed and compressed data. SIMPLISMA was able to accurately extract the spectral features and concentration profiles directly from daublet compressed IMS data at a compression ratio of 93.7% with root-mean-square errors of reconstruction < 3%. The daublet wavelet filters were selected, because they worked well when compared to coiflet and symmlet. The effects of the daublet filter width and compression ratio were evaluated with respect to reconstruction errors of the data sets and SIMPLISMA spectra. For these experiments, the daublet 14 filter performed well for the two data sets.     

改进乌鸦算法优化多阈值图像分割

目的 针对传统乌鸦算法随机搜索的盲目性和易陷入局部最优的缺点,提出一种改进乌鸦算法,用于多阈值图像分割.方法 采用精英分享策略,弥补乌鸦位置更新的盲目性;引入Levy飞行机制,避免算法陷入局部最优;随迭代次数调整变尺度系数,限制搜索步长,加快算法收敛;以Kapur熵为适应函数,利用改进乌鸦算法对不同类型图像进行多阈值分割,并与传统乌鸦、布谷鸟等4种算法的分割结果进行对比分析.结果 改进乌鸦算法对Lena,Flower,Fruits和Boat图分割后的结构相似性分别为0.7703,0.7761,0.7276和0.7921;标准偏差分别为0.0295,0.0385,0.0344和0.0173,实验数据表明,改进算法较其他算法有着更好的分割效果.结论 文中算法有效地改进了传统乌鸦算法的盲目性和易陷入局部最优的缺点,能够准确地分割复杂图像,在多阈值图像分割领域具有一定的参考价值.     

基于加权欧氏距离的光谱重构训练样本选择

目的研究光谱反射率重建过程中训练样本的选取,以提高光谱反射率的重建精度。方法根据检验样本与训练样本间的欧氏距离判断两者的相似度,并将样本向量各维度分量"标准化"到均值、方差相等,使得各维度分别满足标准正态分布,同时将其方差的倒数作为权重赋予训练样本。实验以Munsell色卡为总的训练样本集,经Mohammadi方法、Cao方法以及文中方法选择出的样本作为最终的训练样本,用Color Rendition Chart 24色卡为检验样本,采用伪逆法分别对选择出来的训练样本进行光谱反射率重建。结果通过Matlab软件仿真实验,文中方法的平均色差为0.7918△E_(ab)~*,最大色差为1.7148△E_(ab)~*,平均均方根误差为0.0060,最大光谱均方根误差为0.0127。结论基于加权欧氏距离的训练样本选择可以有效地提高光谱的重建精度,能更好地实现颜色再现。     

Evaluation and unification of some methods for estimating reflectance spectra from RGB images

The problem of estimating spectral reflectances from the responses of a digital camera has received considerable attention recently. This problem can be cast as a regularized regression problem or as a statistical inversion problem. We discuss some previously suggested estimation methods based on critically undersampled RGB measurements and describe some relations between them. We concentrate mainly on those models that are using a priori information in the form of high-resolution measurements. We use the "kernel machine" framework in our evaluations and concentrate on the use of multiple illuminations and on the investigation of the performance of global and locally adapted estimation methods. We also introduce a nonlinear transformation of reflectance values to ensure that the estimated reflection spectra fulfill physically motivated boundary conditions. The reported experimental results are derived from measured and simulated camera responses from the Munsell Matte, NCS, and Pantone data sets.     

基于不同色块数量的光谱重构对比

分别对24色块、140色块、1 269色块3种不同颜色数量的哑光色卡光谱反射率进行主成分分析(PCA),利用分析所提取的最大6个特征向量重构光谱,并使用均方根误差值(RMSE)和CIE1976色差值对结果进行评价。实验结果表明:同等条件重构后,140色卡在累计贡献率和色差方面均优于另外2种数量的色卡,均方根误差也仅次于1 269色卡,表明不同色块数量的训练样本会影响光谱重构精度;重构后的光谱反射率均在对应黄色、绿色、蓝色等的中间波段,再现效果最好,蓝紫色相对较弱,橙色-红色表现最差。     

Enhanced Quality Control in Pharmaceutical Applications by Combining Raman Spectroscopy and Machine Learning Techniques

In this work, we applied machine learning techniques to Raman spectra for the characterization and classification of manufactured pharmaceutical products. Our measurements were taken with commercial equipment, for accurate assessment of variations with respect to one calibrated control sample. Unlike the typical use of Raman spectroscopy in pharmaceutical applications, in our approach the principal components of the Raman spectrum are used concurrently as attributes in machine learning algorithms. This permits an efficient comparison and classification of the spectra measured from the samples under study. This also allows for accurate quality control as all relevant spectral components are considered simultaneously. We demonstrate our approach with respect to the specific case of acetaminophen, which is one of the most widely used analgesics in the market. In the experiments, commercial samples from thirteen different laboratories were analyzed and compared against a control sample. The raw data were analyzed based on an arithmetic difference between the nominal active substance and the measured values in each commercial sample. The principal component analysis was applied to the data for quantitative verification (i.e., without considering the actual concentration of the active substance) of the difference in the calibrated sample. Our results show that by following this approach adulterations in pharmaceutical compositions can be clearly identified and accurately quantified.     

Separating the fluorescence and reflectance components of coral spectra

The optical signal leaving the surface of fluorescing corals is a combination of elastic and inelastic scatter. A new experimental method was developed to separate the fluorescence and the reflectance components that involves measurements with and without a long-pass cutoff filter that eliminates the fluorescence contribution to the signal. The required measurements were performed underwater to demonstrate the applicability of the method for in situ applications. Computations with prototype rather than individually measured fluorescence emission spectra do not significantly compromise the accuracy of the results. A model was developed for calculating the interaction of the reflectance and the fluorescence components with new incident illumination conditions. The model calculations were supported by field experiment. We show that the contribution of fluorescence to some coral's spectra in various illumination conditions justifies consideration in optical models. The results are applicable to modeling the spectra of fluorescing corals under any irradiance spectrum and interpreting remote-sensing data in the relevant wavelength range. Further research is necessary to examine the significance of fluorescence near coral reefs at various scales.     

Introduction and application of secured principal component regression for analysis of uncalibrated spectral features in optical spectroscopy and chemical sensing

In this study, a novel chemometric algorithm for improved evaluation of analytical data is presented and applied to three spectroscopic data sets obtained by different analytical methods. This so-called secured principal component regression (sPCR) was developed for detecting and correcting uncalibrated spectral features newly emerging in spectra after finalizing the PCR calibration, which may result in major concentration errors. Hence, detection and correction of uncalibrated features is essential. Furthermore, detected uncalibrated features provide qualitative information for sensing and process monitoring applications indicating problems in the process flow. After conventional PCR calibration, sPCR analyzes measurement data in two steps: The first step investigates whether the obtained data set is consistent with the calibration model or not. If spectroscopic features are found that cannot be modeled by the principal components, they are extracted from the measurement spectrum. This corrected spectrum is then evaluated by conventional PCR. In the Experimental Section, sPCR was successfully applied to three data sets obtained by different spectroscopic measurements in order to corroborate general applicability of the proposed concept. For each data set, one of several substances was excluded from the calibration acting in the sPCR assessment as uncalibrated absorber. The test sets consisted of disturbed and undisturbed samples. A total of 109 out of 110 test samples were correctly classified as disturbed or undisturbed by an uncalibrated absorber. It was confirmed that the extracted disturbance spectra are in accordance with the spectra of the uncalibrated analytes. The concentration results obtained with sPCR were found to be equivalent to conventional PCR results in the case of undisturbed samples and more precise for disturbed samples.     

Directional dependence of spectra of fiber Bragg gratings due to excess loss

The reflectance spectra of chirped fiber Bragg gratings can depend substantially on the direction from which the measurement is taken. The measured difference between forward and backward reflectance spectra measured in a linearly chirped grating was shown to be due to the measured excess loss. Simulation using the popular transfer-matrix model demonstrated that the observed asymmetric behavior could be obtained only when excess loss has an asymmetric spectral shape about the local Bragg wavelengths. Application of cladding mode excess losses to the result of a transfer-matrix model accounted for the experimental observation.     

Optimum sensors for color constancy in scenes illuminated by daylight

The apparent color of an object within a scene depends on the spectrum of the light illuminating the object. However, recording an object's color independent of the illuminant spectrum is important in many machine vision applications. In this paper the performance of a blackbody-model-based color constancy algorithm that requires four sensors with different spectral responses is investigated under daylight illumination. In this investigation sensor noise was modeled as gaussian noise, and the responses were quantized using different numbers of bits. A projection-based algorithm whose output is invariant to illuminant is investigated to improve the results that are obtained. The performance of both of these algorithms is then improved by optimizing the spectral sensitivities of the four sensors using freely available CIE standard daylight spectra and a set of lightness-normalized Munsell reflectance data. With the optimized sensors the performance of both algorithms is shown to be comparable to the human visual system. However, results obtained with measured daylight spectra show that the standard daylights may not be sufficiently representative of measured daylight for optimization with the standard daylight to lead to a reliable set of optimum sensor characteristics.     

Prediction of ethylene content in melt-state random and block polypropylene by near-infrared spectroscopy and chemometrics: influence of a change in sample temperature and its compensation method

This paper reports on the influence of a change in sample temperature, and a method for its compensation, for the prediction of ethylene (C2) content in melt-state random polypropylene (RPP) and block polypropylene (BPP) by near-infrared (NIR) spectroscopy and chemometrics. Near-infrared (NIR) spectra of RPP in the melt and solid states were measured by a Fourier transform near-infrared (FT-NIR) on-line monitoring system and an FT-NIR laboratory system. There are some significant differences between the solid and melt-state RPP spectra. Moreover, we investigated the predicted values of the C2 content from the RPP or BPP spectra measured at 190 degrees C and 250 degrees C using the calibration model for the C2 content developed using the RPP or BPP spectra measured at 230 degrees C. The errors in the predicted values of the C2 content depend on the pretreatment methods for each calibration model. It was found that multiplicative signal correction (MSC) is very effective in compensating for the influence of the change of temperature for the RPP or BPP samples on the predicted C2 content. From the suggestion of principal component analysis (PCA) and difference spectrum analysis, we propose a new compensation method for the temperature change that uses the difference spectra between two spectra sets measured at different temperatures. We achieved good results using the difference spectra between the RPP/BPP spectra sets measured at 190 degrees C and 250 degrees C after correction and the calibration model developed with the spectra measured at 230 degrees C. The comparison between the method using MSC and the proposed method showed that the predicted error in the latter is slightly better than those in the former.     

Correcting attenuated total reflection-Fourier transform infrared spectra for water vapor and carbon dioxide

Fourier transform infrared (FT-IR) spectroscopy is a valuable technique for characterization of biological samples, providing a detailed fingerprint of the major chemical constituents. However, water vapor and CO(2) in the beam path often cause interferences in the spectra, which can hamper the data analysis and interpretation of results. In this paper we present a new method for removal of the spectral contributions due to atmospheric water and CO(2) from attenuated total reflection (ATR)-FT-IR spectra. In the IR spectrum, four separate wavenumber regions were defined, each containing an absorption band from either water vapor or CO(2). From two calibration data sets, gas model spectra were estimated in each of the four spectral regions, and these model spectra were applied for correction of gas absorptions in two independent test sets (spectra of aqueous solutions and a yeast biofilm (C. albicans) growing on an ATR crystal, respectively). The amounts of the atmospheric gases as expressed by the model spectra were estimated by regression, using second-derivative transformed spectra, and the estimated gas spectra could subsequently be subtracted from the sample spectra. For spectra of the growing yeast biofilm, the gas correction revealed otherwise hidden variations of relevance for modeling the growth dynamics. As the presented method improved the interpretation of the principle component analysis (PCA) models, it has proven to be a valuable tool for filtering atmospheric variation in ATR-FT-IR spectra.