Nonlinear estimation of spectral reflectance based on Gaussian mixture distribution for color image reproduction

Nonlinear estimation method of spectral reflectance from camera responses is proposed. The proposed method minimizes the mean square error of spectral reflectance when the reflectance can be regarded as a random sequence of Gaussian mixture distribution. In computer simulations, 168 samples of spectral reflectance from a color chart are estimated from their image signals obtained by three- and six-band cameras. It is confirmed that the proposed method improves the accuracy in comparison with the conventional Wiener estimation method.     

Correcting cross-media instrument metamerism for reflectance estimation in multispectral imaging

In multispectral imaging, the color accuracy of spectral reflectance estimation degrades significantly if the medium of test samples is different from that of calibration samples. This occurs mainly for two reasons, i.e., the different characteristics of spectral reflectances and the different measurement principles between an imaging system and a spectrophotometer. In this paper, this problem is referred to as cross-media instrument metamerism. We propose to correct it by using calibration samples from a standard color chart and a limited number of tuning samples with a target medium as a priori knowledge. The reflectance transform is computed by using both calibration and tuning samples, and the metamerism transform is calculated by modeling the correlation of camera responses between neighboring imaging channels. Experimental results show that the proposed method produces satisfactory spectral and colorimetric accuracy in reflectance estimation. The method could be deployed in practical applications when the available samples of certain media are inadequate for accurate reflectance estimation in a multispectral imaging system.     

基于最优化维纳估计算法的多光谱图像重构

目的提高光谱反射率重构精度,更好实现图像颜色再现。方法选择4种不同的训练样本,比较不同的改进的维纳估计算法重构光谱反射率精度。分别把Color Checker Color Rendition Chart(RC-24色)和Color Checker SG(140色)以及Panton色卡、孟塞尔色卡(Munsell-1269)4种不同颜色数量的哑光色卡作为训练样本,用伪逆算法、维纳估计算法、自适应维纳估计算法和最优化维纳估计算法分别对多光谱图像进行重构,并进行精度对比。从均方根误差、色差等方面评价这些光谱重构方法。结果最优化维纳估计算法重构均方根误差为0.000 41,色差为0.7874,均小于其他几种算法。结论最优化维纳估计算法能更好实现颜色再现。     

Optimal selection of representative colors for spectral reflectance reconstruction in a multispectral imaging system

In a multispectral color imaging system, the spectral reflectance of the object being imaged always needs to be accurately reconstructed by employing the training samples on specific color charts. Considering that the workload is heavy when all those color samples are used in practical applications, it is important to select only a limited number of the most representative samples. This is possible as the color charts are usually designed to cover the range of commonly imaged colors, and the color samples are redundant for spectral image reconstruction. We propose an eigenvector-based method and a virtual-imaging-based method for representative color selection by minimizing the total reflectance root-mean-squares errors. The effectiveness of the proposed methods is confirmed by experimental results when compared with existing techniques.     

System design for accurately estimating the spectral reflectance of art paintings

Accurately estimating the spectral reflectance of art paintings from low-dimensional multichannel images requires that both image-acquisition hardware with appropriate spectral characteristics and appropriate estimation software be applied to the captured multichannel image. In this study, a system that incorporates both factors is designed and developed on the basis of the minimum-mean-square error criterion. The accuracy of spectral estimation by use of this system is evaluated, and the system's high performance is demonstrated.     

Weighted compression of spectral color information

Spectral color information is used nowadays in many different applications. Accurate spectral images are usually very large files, but a proper compression method can reduce needed storage space remarkably with a minimum loss of information. In this paper we introduce a principal component analysis (PCA) -based compression method of spectral color information. In this approach spectral data is weighted with a proper weight function before forming the correlation matrix and calculating the eigenvector basis. First we give a general framework for how to use weight functions in compression of relevant color information. Then we compare the weighted compression method with the traditional PCA compression method by compressing and reconstructing the Munsell data set consisting of 1,269 reflectance spectra and the Pantone data set consisting of 922 reflectance spectra. Two different weight functions are proposed and tested. We show that weighting clearly improves retention of color information in the PCA-based compression process.     

Improvement of color reproduction in color digital holography by using spectral estimation technique

We propose a color digital holography by using spectral estimation technique to improve the color reproduction of objects. In conventional color digital holography, there is insufficient spectral information in holograms, and the color of the reconstructed images depend on only reflectances at three discrete wavelengths used in the recording of holograms. Therefore the color-composite image of the three reconstructed images is not accurate in color reproduction. However, in our proposed method, the spectral estimation technique was applied, which has been reported in multispectral imaging. According to the spectral estimation technique, the continuous spectrum of object can be estimated and the color reproduction is improved. The effectiveness of the proposed method was confirmed by a numerical simulation and an experiment, and, in the results, the average color differences are decreased from 35.81 to 7.88 and from 43.60 to 25.28, respectively.     

Using reflectance models for color scanner calibration

We examine the use of linear spectral reflectance models for calibrating a color scanner to generate device-independent CIE XYZ values from scanner vectors. Polynomial regression approaches to color scanner calibration use parameterized functions to approximate the calibration mapping over a set of training colors. These approaches can perform poorly if the parameterized functions do not accurately model the structure of the desired calibration mapping. Several studies have shown that linear reflectance models accurately characterize a wide range of materials. By viewing color scanner calibration as reflectance estimation, we can incorporate linear reflectance models into the calibration process. We show that in most cases linear models do not constrain the calibration problem sufficiently to allow exact recovery of X, Y, Z from a scanner vector obtained with three filters. By examining a series of methods that exploit information about reflectance functions, however, we show that reflectance information can be used to improve the accuracy of calibration over that of standard methods applied to the same set of inputs.     

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。     

Spectral characterization of a color scanner based on optimized adaptive estimation

A scanner characterization method is proposed to estimate spectral reflectance from scanner responses by using an optimized adaptive estimation method. In contrast to our previous study [J. Opt. Soc. Am. A21, 1125 (2004)], this method considers the weighting of training samples. It is demonstrated that the color accuracy of this method is only slightly affected by the number of training samples and can provide more accurate reflectance estimation.     

Remote sensing of ocean color: a methodology for dealing with broad spectral bands and significant out-of-band response

A methodology for delineating the influence of finite spectral bandwidths and significant out-of-band response of sensors for remote sensing of ocean color is developed and applied to the Sea-viewing Wide-Field-of-view Sensor (SeaWiFS). The basis of the method is the application of the sensor's spectral-response functions to the individual components of the top-of-the-atmosphere (TOA) radiance rather than the TOA radiance itself. For engineering purposes, this approach allows one to assess easily (and quantitatively) the potential of a particular sensor design for meeting the system-sensor plus algorithms-performance requirements. In the case of the SeaWiFS, two significant conclusions are reached. First, it is found that the out-of-band effects on the water-leaving radiance component of the TOA radiance are of the order of a few percent compared with a sensor with narrow spectral response. This implies that verification that the SeaWiFS system-sensor plus algorithms-meets the goal of providing the water-leaving radiance in the blue in clear ocean water to within 5% will require measurements of the water-leaving radiance over the entire visible spectrum as opposed to just narrow-band (10-20-nm) measurements in the blue. Second, it is found that the atmospheric correction of the SeaWiFS can be degraded by the influence of water-vapor absorption in the shoulders of the atmospheric-correction bands in the near infrared. This absorption causes an apparent spectral variation of the aerosol component between these two bands that will be uncharacteristic of the actual aerosol present, leading to an error in correction. This effect is dependent on the water-vapor content of the atmosphere. At typical water-vapor concentrations the error is larger for aerosols with a weak spectral variation in reflectance than for those that display a strong spectral variation. If the water-vapor content is known, a simple procedure is provided to remove the degradation of the atmospheric correction. Uncertainty in the water-vapor content will limit the accuracy of the SeaWiFS correction algorithm.     

Spectral color constancy using a maximum entropy approach

This paper proposes a solution to the spectral color constancy problem. The method is based on a statistical model for the surface reflectance spectrum and applies a maximum entropy constraint. Unlike prior methods based on linear models, the solution process does not require a set of basis functions to be defined, nor does it require a database of spectra to be specified in advance. Experiments on simulated and real data show that spectral estimation using the maximum entropy approach is feasible and performs similarly to existing spectral methods in spite of the lower level of a priori information required.     

Kernel-based spectral color image segmentation

In this work, we propose a new algorithm for spectral color image segmentation based on the use of a kernel matrix. A cost function for spectral kernel clustering is introduced to measure the correlation between clusters. An efficient multiscale method is presented for accelerating spectral color image segmentation. The multiscale strategy uses the lattice geometry of images to construct an image pyramid whose hierarchy provides a framework for rapidly estimating eigenvectors of normalized kernel matrices. To prevent the boundaries from deteriorating, the image size on the top level of the pyramid is generally required to be around 75 x 75, where the eigenvectors of normalized kernel matrices would be approximately solved by the Nystr?m method. Within this hierarchical structure, the coarse solution is increasingly propagated to finer levels and is refined using subspace iteration. In addition, to make full use of the abundant color information contained in spectral color images, we propose using spectrum extension to incorporate the geometric features of spectra into similarity measures. Experimental results have shown that the proposed method can perform significantly well in spectral color image segmentation as well as speed up the approximation of the eigenvectors of normalized kernel matrices.     

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。     

Rendering complex scenes for psychophysics using RADIANCE: how accurate can you get?

Rendering packages are used by visual psychophysicists to produce complex stimuli for their experiments, tacitly assuming that the simulation results accurately reflect the light-surface interactions of a real scene. RADIANCE is a physically based, freely available, and commonly used rendering software. We validated the calculation accuracy of this package by comparing simulation results with measurements from real scenes. RADIANCE recovers color gradients well but the results are shifted in color space. Currently, there is no better simulation alternative for achieving physical accuracy than by combining a spectral rendering method with RADIANCE.     

基于人工神经网络的光谱反射率重建

目的研究基于BP神经网络法和FNN神经网络法重构图像光谱反射率的精度。方法以SG标准色卡作为训练样本,分别使用BP和FNN神经网络法,对测试样本DC标准色卡的光谱反射率进行预测,并利用CIEL*a*b*色差公式、均方根误差(ERMS)和光谱匹配精度(GFC)对结果进行评价。结果 BP和FNN神经网络重构的光谱反射率平均色差(ΔEab)分别为2.997和3.071,平均均方根误差(ERMS)分别为0.056和0.049,平均光谱匹配精度(GFC)分别为0.987和0.991。结论 2种神经网络方法重构的光谱反射率具有相当优越的色度和光谱精度。相比于FNN神经网络,BP神经网络更加适合于光谱图像的获取领域。     

Scene illuminant classification: brighter is better

Knowledge of the scene illuminant spectral power distribution is useful for many imaging applications, such as color image reproduction and automatic algorithms for image database applications. In many applications accurate spectral characterization of the illuminant is impossible because the input device acquires only three spectral samples. In such applications it is sensible to set a more limited objective of classifying the illuminant as belonging to one of several likely types. We describe a data set of natural images with measured illuminants for testing illuminant classification algorithms. One simple type of algorithm is described and evaluated by using the new data set. The empirical measurements show that illuminant information is more reliable in bright regions than in dark regions. Theoretical predictions of the algorithm's classification performance with respect to scene illuminant blackbody color temperature are tested and confirmed by using the natural-image data set.     

A method to estimate correlated color temperatures of illuminantsusing a color video camera

An indirect method of estimating light source color or the correlated color temperature of illuminants is proposed. It incorporates a color video camera with fixed white balance, a set of known color chips, and a desktop microcomputer. The correlated color temperature of the light source illuminating the color chips is estimated. The mathematical method used is the estimation of functions to be integrated from several values of the definite integrations. These values, which are measurable, are given as the output of the color video camera under different conditions. By means of the method of least squares, estimation accuracies of from 0.1 to 1.4% of the correlated color temperature are achieved. The accuracy achieved depends on the spectral distribution of the illuminants, except for illuminants like fluorescent lamps     

Spectral characterization of a color scanner by adaptive estimation

A new method of spectral characterization for color scanners by the use of adaptive estimation is proposed. It deals with estimation of high-dimensional reflectance vectors from low-dimensional scanner response vectors when the scanner departs from linearity. We first investigate the spectral linearity of the scanner, and then estimate the spectral reflectance adaptively based on the local statistics of a set of neighboring training samples. As the proposed characterization method does not utilize the mathematically recovered spectral responsivity, its inherent inaccuracy is not critical to the spectral characterization. Experimental results showed significant advantage of adaptive estimation when compared with other methods.     

Spectral reflectance estimation from camera responses by support vector regression and a composite model

Regression methods are widely used to estimate the spectral reflectance of object surfaces from camera responses. These methods are under the same problem setting as that to build an estimation function for each sampled wavelength separately, which means that the accuracy of the spectral estimation will be reduced when the training set is small. To improve the spectral estimation accuracy, we propose a novel estimating approach based on the support vector regression method. The proposed approach utilizes a composite modeling scheme, which formulates the RGB values and the sampled wavelength together as the input term to make the most use of the information from the training samples. Experimental results show that the proposed method can improve the recovery accuracy when the training set is small.