Maximum a posteriori estimation of spectral reflectance from color image and multipoint spectral measurements

Accurate color image reproduction under arbitrary illumination can be realized if the spectral reflectance functions in a scene are obtained. Although multispectral imaging is one of the promising methods to obtain the reflectance of a scene, it is expected to reduce the number of color channels without significant loss of accuracy. This paper presents what we believe to be a new method for estimating spectral reflectance functions from color image and multipoint spectral measurements based on maximum a posteriori (MAP) estimation. Multipoint spectral measurements are utilized as auxiliary information to improve the accuracy of spectral reflectance estimated from image data. Through simulations, it is confirmed that the proposed method improves the estimation accuracy, particularly when a scene includes subjects that belong to various categories.     

Hyperspectral diffuse reflectance imaging for rapid, noncontact measurement of the optical properties of turbid materials

We present a method and technique of using hyperspectral diffuse reflectance for rapid determination of the optical properties of turbid media. A hyperspectral imaging system in line scanning mode was used to acquire spatial diffuse reflectance profiles from liquid phantoms made up of absorbing dyes and fat emulsion scatterers over the spectral range of 450-1000 nm instantaneously. The hyperspectral reflectance data were analyzed by using a steady-state diffusion approximation model for semi-infinite homogeneous media. A calibration procedure was developed to compensate the nonuniform instrument response of the imaging system, and a curve-fitting algorithm was used to extract absorption and reduced scattering coefficients (mua and mus', respectively) for the phantoms in the wavelength range from 530 to 900 nm. The hyperspectral imaging system gave good measures of mua and mus' for the phantoms with average fitting errors of 12% and 7%, respectively. The hyperspectral imaging technique is fast, noncontact, and easy to use, which makes it especially suitable for measurement of the optical properties of turbid liquid and solid foods.     

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.     

Spatial-spectral modulating snapshot hyperspectral imager

Experimental results are presented for a computed tomography imaging spectrometer (CTIS) with imposed spatial-spectral modulation on the image scene. This modulation structure on the CTIS tomographic dispersion created substantial gains in spectral reconstruction resolution after standard iterative, nonlinear, inversion techniques were used. Modulation limits system ambiguities, so high-frequency spectral and low-frequency spatial scene data could be recovered. The results demonstrate how spatial modulation acts as a high-frequency spectral deconvolver for the snapshot hyperspectral imager technology.     

Instrumental error in chromotomosynthetic hyperspectral imaging

Chromotomosynthetic imaging (CTI) is a method of convolving spatial and spectral information that can be reconstructed into a hyperspectral image cube using the same transforms employed in medical tomosynthesis. A direct vision prism instrument operating in the visible (400-725?nm) with 0.6?mrad instantaneous field of view (IFOV) and 0.6-10?nm spectral resolution has been constructed and characterized. Reconstruction of hyperspectral data cubes requires an estimation of the instrument component properties that define the forward transform. We analyze the systematic instrumental error in collected projection data resulting from prism spectral dispersion, prism alignment, detector array position, and prism rotation angle. The shifting and broadening of both the spectral lineshape function and the spatial point spread function in the reconstructed hyperspectral imagery is compared with experimental results for monochromatic point sources. The shorter wavelength (λ<500 nm) region where the prism has the highest spectral dispersion suffers mostly from degradation of spectral resolution in the presence of systematic error, while longer wavelengths (λ>600 nm) suffer mostly from a shift of the spectral peaks. The quality of the reconstructed hyperspectral imagery is most sensitive to the misalignment of the prism rotation mount. With less than 1° total angular error in the two axes of freedom, spectral resolution was degraded by as much as a factor of 2 in the blue spectral region. For larger errors than this, spectral peaks begin to split into bimodal distributions, and spatial point response functions are reconstructed in rings with radii proportional to wavelength and spatial resolution.     

The characteristics of peptide collision-induced dissociation using a high-performance MALDI-TOF/TOF tandem mass spectrometer

A new matrix-assisted laser desorption/ionization (MALDI) time-of-flight/time-of-flight (TOF/TOF) high-resolution tandem mass spectrometer is described for sequencing peptides. This instrument combines the advantages of high sensitivity for peptide analysis associated with MALDI and comprehensive fragmentation information provided by high-energy collision-induced dissociation (CID). Unlike the postsource decay technique that is widely used with MALDI-TOF instruments and typically combines as many as 10 separate spectra of different mass regions, this instrument allows complete fragment ion spectra to be obtained in a single acquisition at a fixed reflectron voltage. To achieve optimum resolution and focusing over the whole mass range, it may be desirable to acquire and combine three separate sections. Different combinations of MALDI matrix and collision gas determine the amount of internal energy deposited by the MALDI process and the CID process, which provide control over the extent and nature of the fragment ions observed. Examples of peptide sequencing are presented that identify sequence-dependent features and demonstrate the value of modifying the ionization and collision conditions to optimize the spectral information.     

Simple spectral stray light correction method for array spectroradiometers

A simple, practical method has been developed to correct a spectroradiometer's response for measurement errors arising from the instrument's spectral stray light. By characterizing the instrument's response to a set of monochromatic laser sources that cover the instrument's spectral range, one obtains a spectral stray light signal distribution matrix that quantifies the magnitude of the spectral stray light signal within the instrument. By use of these data, a spectral stray light correction matrix is derived and the instrument's response can be corrected with a simple matrix multiplication. The method has been implemented and validated with a commercial CCD-array spectrograph. Spectral stray light errors after the correction was applied were reduced by 1-2 orders of magnitude to a level of approximately 10(-5) for a broadband source measurement, equivalent to less than one count of the 15-bit-resolution instrument. This method is fast enough to be integrated into an instrument's software to perform real-time corrections with minimal effect on acquisition speed. Using instruments that have been corrected for spectral stray light, we expect significant reductions in overall measurement uncertainties in many applications in which spectrometers are commonly used, including radiometry, colorimetry, photometry, and biotechnology.     

Prelaunch characterization of the Ozone Monitoring Instrument transfer function in the spectral domain

A method and an experimental measurement setup to accurately characterize the instrument transfer function in the spectral domain for hyperspectral spectrometers in the ultraviolet-visible wavelength range are described. The application to the on-ground calibration of the Ozone Monitoring Instrument (OMI) on board the Earth Observing System Aura satellite is presented and discussed. With this method and setup, based on an echelle grating, a sampling of the instrument transfer function in the spectral domain can be selected and is not limited by the spectral resolution and sampling of the spectrometer that is being characterized. The importance of accurately knowing the OMI instrument transfer functions in the spectral domain for in-flight differential optical absorption spectroscopy retrievals and wavelength calibration is discussed. The analysis of the OMI measurement data is presented and shows that the instrument transfer functions in the spectral domain as a function of wavelength and viewing angle can be determined with high accuracy.     

Some practical issues in anomaly detection and exploitation of regions of interest in hyperspectral images

We address method of detection of anomalies in hyperspectral images that consists in performing the detection when the spectral signatures of the targets are unknown. We show that, in real hyperspectral images, use of the full spectral resolution may not be necessary for detection but that the correlation properties of spectral fluctuations have to be taken into account in the design of the detection algorithm. Anomaly detectors are useful for detecting regions of interest (ROIs), but, as they are prone to false alarms, one must analyze the ROIs obtained further to decide whether they correspond to real targets. We propose a method of exploitation of these ROIs that consists in generating a single image in which the contrast of the ROI is optimized.     

高光谱数据融合研究

多传感器的遥感数据融合在城市规划、土地利用、矿产探测、军事侦察等领域有着非常广泛的应用前景。本文高光谱图像光谱分辨率高、空间分辨率低、多光谱图像空间分辨率高、光谱分辨率高的特点,阐述了其数据融合。特别介绍了CRISP锐化算法。     

基于CEM的高光谱图像小目标检测算法

针对高光谱图像中小目标检测问题,提出了一种基于约束能量最小化(Constrained Energy Minimization,CEM)的目标检测算法.该算法首先对原始图像进行背景信息抑制从而抑制背景地物、突出低概率的小目标,用迭代误差分析的自动端元提取算法找出目标的端元光谱,然后把目标端元光谱代入CEM滤波器得到该目标的检测结果图.用高光谱数据进行了实验研究,并与CEM滤波器进行了比较.结果表明,其检测性能与直接采用CEM方法的检测性能相当,但是相对于CEM方法,该算法不需要目标的先验光谱信息,更具有实用性.     

Neutron spectrometry for radiation protection purposes

Determination of the dose equivalent is required for radiation protection purposes, however such a determination is quite difficult for neutron radiation. In order to perform accurate dosimetric determinations, it is advantageous to acquire information about the neutron fluence spectrum in the workplace as well as the reference radiations used to calibrate dosimetric instruments. This information can then be used to select the appropriate dosimetric instrument, the optimum calibration condition or to establish correction factors that account for the differences in calibration and workplace conditions. For quite some time, neutron spectrometry has been used for these purposes. A brief review of the applications of spectrometers in radiation protection and some recommendations for further development are given here.     

Polychromatic transmissometer for in situ measurements of suspended particles and gelbstoff in water

The beam attenuation coefficient is an optical parameter that sensitively depends on suspended and dissolved substances in water. Its measurement is not only of interest for an understanding of the radiative transfer in a water column. With appropriate algorithms for data interpretation, it also allows a fast determination of absorbing and scattering matter as time-series measurements or depth profiles that cannot easily be obtained with other methods. An instrument has been developed for measuring spectral attenuation coefficients over a wavelength range from 340 to 785 nm. The optical path length can be set between 0 and 400 mm. This allows application in a wide range of turbidity in coastal and inland (case 2 and case 3) waters and a calibration of the instrument during in-situ measurements. This makes the instrument suitable for long-term applications in which signals from conventional instruments would degrade owing to the biofouling of optical windows. From the data, the amount and the size distribution of suspended particles and the specific absorption of dissolved organic matter are derived in real time. Algorithms based on Monte Carlo methods are available for a classification of transparent particles and phytoplankton.     

基于AOTF的高光谱偏振成像系统设计

为获得目标和场景的偏振信息,提高目标识别和地物探测的准确度,设计了基于液晶相位可变延迟器(LCVR)和声光可调滤波器(AOTF)的高光谱偏振成像探测系统。首先介绍了偏振成像探测的原理,然后给出了探测系统的结构,并对其光学系统的各元件进行了合理的参数分配,最后利用编制的信息处理软件对系统采集的高光谱偏振图像进行了处理,处理结果能够反映出目标的光谱特性和偏振特性,对于复杂背景下的目标探测与识别具有重要的应用价值。     

Learning and Predicting Photonic Responses of Plasmonic Nanoparticle Assemblies via Dual Variational Autoencoders

The application of machine learning is demonstrated for rapid and accurate extraction of plasmonic particles cluster geometries from hyperspectral image data via a dual variational autoencoder (dual-VAE). In this approach, the information is shared between the latent spaces of two VAEs acting on the particle shape data and spectral data, respectively, but enforcing a common encoding on the shape-spectra pairs. It is shown that this approach can establish the relationship between the geometric characteristics of nanoparticles and their far-field photonic responses, demonstrating that hyperspectral darkfield microscopy can be used to accurately predict the geometry (number of particles, arrangement) of a multiparticle assemblies below the diffraction limit in an automated fashion with high fidelity (for monomers (0.96), dimers (0.86), and trimers (0.58). This approach of building structure-property relationships via shared encoding is universal and should have applications to a broader range of materials science and physics problems in imaging of both molecular and nanomaterial systems.     

Spectrum analysis of motion parallax in a 3D cluttered scene and application to egomotion

Previous methods for estimating observer motion in a rigid 3D scene assume that image velocities can be measured at isolated points. When the observer is moving through a cluttered 3D scene such as a forest, however, pointwise measurements of image velocity are more challenging to obtain because multiple depths, and hence multiple velocities, are present in most local image regions. We introduce a method for estimating egomotion that avoids pointwise image velocity estimation as a first step. In its place, the direction of motion parallax in local image regions is estimated, using a spectrum-based method, and these directions are then combined to directly estimate 3D observer motion. There are two advantages to this approach. First, the method can be applied to a wide range of 3D cluttered scenes, including those for which pointwise image velocities cannot be measured because only normal velocity information is available. Second, the egomotion estimates can be used as a posterior constraint on estimating pointwise image velocities, since known egomotion parameters constrain the candidate image velocities at each point to a one-dimensional rather than a two-dimensional space.     

Multivariate hyperspectral Raman imaging using compressive detection

A multivariate hyperspectral imaging (MHI) instrument has been designed and constructed to achieve greatly increased Raman imaging speeds by utilizing a compressive spectral detection strategy. The instrument may be viewed as a generalized spectrometer, which can function either as a conventional monochromator or in a wide variety of other hyperspectral modalities. The MHI utilizes a spatial light modulator (SLM) to produce programmable optical filters to rapidly detect and map particular sample components. A sequence of Hadamard-transform or random filter functions may be used to regenerate full Raman spectra. Compressive detection is achieved either using multivariate signal processing filter functions or the actual component spectra. Compressive detection is shown to be capable of achieving sampling speeds exceeding 1 ms per image pixel and the collection of chemical images in less than a minute.     

基于光谱特征和纹理特征协同学习的高光谱图像数据分类

高光谱遥感图像中包含有大量的高维数据,传统的有监督学习算法在对这些数据进行分类时要求获取足够多的有标记样本用于分类器的训练.然而,对高光谱图像中大量的复杂地物像元所属类别进行准确标注通常需要耗费极大的人力.在本文中,我们提出了一种基于半监督学习的光谱和纹理特征协同学习(STF-CT)--法,利用协同学习机制将高光谱图像光谱特征和空间纹理特征这两种不同的特征结合起来,用于小训练样本集下的高光谱图像数据分类问题.STF-CT算法充分利用了高光谱图像的光谱和纹理特征这两个独立视图,构建起一种有效的半监督分类方法,用于提升分类器在小训练样本集情况下的分类精度.实验结果表明该算法在小训练样本集下的高光谱地物分类问题上具有很好的效果.     

Influence of band selection and target estimation error on the performance of the matched filter in hyperspectral imaging

The matched filter is a widely used detector in hyperspectral detection applications because of its simplicity and its efficiency in practical situations. We propose to estimate its performance with respect to the number of spectral bands. These spectral bands are selected thanks to a genetic algorithm in order to optimize the contrast between the target and the background in the detection plane. Our band selection method can be used to optimize not only the position but also the linewidth of the spectral bands. The optimized contrast always increases with the number of selected bands. However, in practical situations, the target spectral signature has to be estimated from the image. We show that in the presence of estimation error, the maximum number of bands may not always be the best choice in terms of detection performance.