Hyperspectral imaging (HSI) technique in remote sensing gauges its performance based on the usefulness of the products. Many of the products consists of the detection and identification of various materials within an imaging element or pixel. Nearly all HSI systems image at a level in which several materials may compromise a single pixel. We have derived an expression for the sub-pixel detection capabilities assuming complete knowledge of background and target spectra and the spectral influence of the atmosphere. This derivation assumes that a pixel is comprised of a set of background and target spectral components which when added together, weighted by the area extent, produce the observed spectrum of the pixel. The analysis is done for two sensor types: one that is photon-noise limited and one that is detector-noise limited. Results of this analysis show that the basic sensor parameters can be separated effectively from phenomenology limitations, thus providing an ability to trade sensor parameters, leading to an understanding of their effect on target detection. This derivation forms a framework for discussing conditions for which backgrounds and the atmospheric interferences are not completely known and for cases in which the target spectrum may or may not be known. These results are compared to analyses directed at empirically understanding sensor trades.

The TRW Imaging Spectrometer III airborne hyperspectral imager was competed in 1996. The spectrometer is a pushbroom sensor that gathers information in 384 contiguous spectral channels covering the 400nm to 2450nm wavelength range. TRWIS III was designed to fly on many different aircraft platforms and to meet critical performance requirements for image quality, co-registration of spectral samples, spectral calibration, noise and radiometric accuracy. Along with its first several seasons of operational demonstrations, the instrument has undergone laboratory performance validation, radiometric calibration, and system upgrades. This paper will describe the current TRWIS III system, the data calibration and correction system, and the instrument's applications to remote sensing.

Recent advances in large format detector arrays and holographic diffraction gratings have made possible the development of imaging spectrographs with high sensitivity and resolution, at relatively low component cost. Several airborne instruments have been built for the visible and near IR spectral band with 10-nm resolution, and SNR of 200:1. Three instruments are compared, an all-reflective spectrography using a convex grating in an Offner configuration, and two off-the-shelf transmission grating spectrographs using volume holograms. The camera is a 1024 X 1024 frame transfer, back-thinned CCD, with four taps for obtaining high frame rates. Performance and scan data is presented and compared to the design for image quality, distortion, and throughput.

We describe a pushbroom imaging spectrometer having a number of attractive features for remote sensing applications, including compact and simple form, good image quality, high efficiency, and very low levels of distortion. These properties are made possible by the unique characteristics of convex gratings manufactured by electron-beam lithography. A laboratory prototype has been built and is under evaluation. If has an f-number of 2.8, covers a spectral band from 400 to 1000 nm with 3 nm spectral resolution and has 750 spatial elements across the entrance slit. Experimental results are shown that demonstrate very low distortion, on the level of 2 percent of a pixel.

Airborne hyperspectral and nadir-viewing laser data can be combined to ascertain shallow water bathymetry. The combination emphasizes the advances and overcomes the disadvantages of each method used alone. For laser systems, both the hardware and software for obtaining off-nadir measurement are complicated and expensive, while for the nadir view the conversion of laser pulse travel time to depth is straightforward. The hyperspectral systems can easily collect data in a full swath, but interpretation for water depth requires careful calibration and correction for transmittance through the atmosphere and water. Relative depths are apparent in displays of several subsets of hyperspectral data, for example, single blue-green wavelengths, endmembers that represent the pure water component of the data, or ratios of deep to shallow water endmembers. A relationship between one of these values and the depth measured by the aligned nadir laser can be determined, and then applied to the rest of the swath to obtain depth in physical units for the entire area covered. We demonstrate this technique using bathymetric charts as a proxy for laser data, and hyperspectral data taken by AVIRIS over Lake Tahoe and Key West.

Hyperspectral imaging (HSI) systems have the potential for tremendous military utility. The ability to use detailed spectral information to characterize objects that may be sub-pixel in size enables efficient surveillance and terrain characterization over wide areas. TRW is actively involved in all aspects of HSI systems form building hyperspectral sensors and real-time processors to the development of processing algorithms. This paper presents performance results on hyperspectral imagery data for target detection. Whereas a recent paper presented results for the detection of known targets, this paper focuses on anomaly detection algorithms for application in those situations involving a high degree of target uncertainty or poor knowledge of atmospheric effects.

Anomaly detection offers a means by which to identify potentially important objects in a scene without prior knowledge of their spectral signatures. As such, this approach is less sensitive to variations in target class composition, atmospheric and illumination conditions, and sensor gain settings than would be a spectral matched filter or similar algorithm. The best existing anomaly detectors generally fall into one of two categories: those based on local Gaussian statistics, and those based on linear mixing moles. Unmixing-based approaches better represent the real distribution of data in a scene, but are typically derived and applied on a global or scene-wide basis. Locally adaptive approaches allow detection of more subtle anomalies by accommodating the spatial non-homogeneity of background classes in a typical scene, but provide a poorer representation of the true underlying background distribution. The CHAMP algorithm combines the best attributes of both approaches, applying a linear-mixing model approach in a spatially adaptive manner. The algorithm itself, and teste results on simulated and actual hyperspectral image data, are presented in this paper.

Hyperspectral image data sets acquired near Cuprite, Nevada in 1995 with the SWIR full spectrum imager (SFSI) and in 1996 with the Airborne Visible/IR Imaging Spectrometer (AVIRIS) are analyzed with a spectral unmixing procedure and the result compared. The SFSI image has pixels on 1 m by 1.5 m centers, the AVIRIS on 17 m centers; the region imaged by SFSI is a small portion of the full AVIRIS scene. Both have nominal spectral band center spacings of about 10 nm. The image data, converted to radiance units, are atmospherically corrected and converted to surface reflectance. Spectral end members are extracted automatically from the two data sets; those representing mineral species common to both are compared to each other and to reference spectra obtained with a portable IR mineral analyzer. The full sets of end members are used in a constrained linear unmixing of the respective hyperspectral image cubes. The resulting unmixing fraction images derived from the AVIRIS and the SFSI data sets for the minerals alunite, buddingtonite, and kaolinite exhibit strong similarities.

A multisensor airborne campaign is carried out in Switzerland in summer 1997. The campaign did not only involve a suite of different sensors but also extensive ground supporting measurements. Amongst the sensor that acquired data over a predefined set of three standard test sites were the hyperspectral imagers DAIS 7915 and CASI, a wide angle airborne camera (WAAC) and a SAR (E-SAR) system as well as an imaging laserscanner. On the ground, geolocation is performed with differential GPS systems and a number of georeferenced ground control points. An active navigation system for the aircraft is used for accurate flight path determination in order to support single- and multi-pass interferometric flights. The thermal ground references consist of a number of targets in the field to verify the thermal performance of the DAIS. Radiometric validation on the ground involves spectroradiometric measurements of a number of selected reference targets, measurements of global flux and radiant temperature, as well as sunphotometer measurements. Conventional field mapping completes the full documentation of the three test sites. The generation of digital surface models using the stereo approach of the WAAC camera and the laserscanner is a goal to support the georeferencing of the different acquired image data. Finally all data are projected onto a common reference system and can be used for further analysis.

A specialized hyperspectral imager has been developed that preprocesses the spectra from an image before the light reaches the detectors. This 'optical computer' does not allow the flexibility of digital post-processing. However, the processing is done in real time and the system can examine approximately equals 2 X 10⁶ scene pixels/sec. Therefore, outdoors it could search for pollutants, vegetation types, minerals, or man-made objects. On a high-speed production line it could identify defects in sheet products like plastic wrap or film, or on painted or plastic parts.

We report results from a simple signal-to-noise analysis of the computed tomography imaging spectrometer (CTIS). The CTIS is non-scanning high-sped imaging spectrometer capable of simultaneously recording spatial and spectral information about dynamic events. This instrument is based on computed- tomography concepts and operates in the visible. The numerical estimate of the noise equivalent spectral radiance for the CTIS is 1.6 nanowatts per centimeter squared per steradian per micrometer.

IR spectral imagers are being considered for air to ground target detection applications. The targets can be detected not only by the exploitation of target to background color but also by the high spectral band to band correlation of many background. Detection of low contrast targets in high thermal clutter backgrounds can therefore be improved by using IR spectral sensors as opposed to broad or narrow band IR sensor. However, the improvement requires a high quality IR spectral sensor. A previous paper explored the trade-offs of some parameters associated with the requirement for low sensor noise levels - namely, instantaneous field of view, aperture diameter, spectral frame rate, number of detector pixels, area coverage rate, spectral bandwidth, and integration time. This paper explores the trade-offs of some parameters associated with the requirement for the preservation of high spectral band to band correlation - namely, spatial ground resolution, spatial registration of spectral bands, and sensor noise level. The Aerospace Corporation Spatially Enhanced Broadband Array Spectrography System was used to collect high quality spectral imagery of vegetated backgrounds at Redstone Arsenal, Alabama. This imagery is analyzed to determine background spectral band to band correlation and coherence. The degradation of correlation as ground spatial resolution, spatial registration accuracy, and sensor noise level are varied was then studied. The result can be used to help set sensor requirements for ground resolution, spatial registration, and noise level.

Hyperspectral imagers sample the electromagnetic spectrum at greater resolution than more traditional imaging systems, which result in a higher band-to-band correlation and greater amounts of data. With bandwidth limitations on the communications channels and storage space, intelligent system design, band selection, and/or data compression will be very important. The data from a new hyperspectral sensor, SEBASS, which collects data in the thermal IR was characterized for compression. As expected, it was found that the data's spectral characteristics were very dependent on scheme content and the collection time of day. It was found that the band-to-band correlation was greater in this data than either HYDICE or AVIRIS hyperspectral data. Compression ratios of 7:1 lossless and 20:1 with minimal loss were achieved compared to 3:1 lossless and 7:1 lossy for HYDICE and AVIRIS data. This increase in compression is directly attributable to the increase in band-to-band correlation. Unique characteristics of the thermal IR hyperspectral data is also discussed.

Hyperspectral imaging is the latest advent in imaging technology, providing the potential to extract information about the objects in a scene that is unavailable to panchromatic imagers. This increased utility, however, comes at the cost of tremendously increased data. The ultimate utility of hyperspectral imagery is in the information that can be gleaned from the spectral dimension, rather than in the hyperspectral imagery itself. To have the broadest range of applications, extraction of this information must occur in real-time. Attempting to produce and exploit complete cubes of hyperspectral imagery at video rates, however, present unique problems for both the imager and the processor, since data rates are scaled by the number of spectral planes in the cube. MIDIS, the Multi-band Identification and Discrimination Imaging Spectroradiometer, allows both real-time here are the major design innovations associated with producing high-speed, high-sensitivity hyperspectral imagers operating in the SWIR and LWIR, and of the electronics capable of handling data rates up to 160 megapixels per second, continuously. Discussion of real-time algorithms capable of exploiting the spectral dimension of the imagery is also included. Beyond design and performance issues associated with producing and processing hyperspectral imagery at such high speeds, this paper also discusses applications of real-time hyperspectral imaging technology. Example imagery includes such problems as detecting counterfeit money, inspecting surfaces, and countering CCD.

We compare the results produced by the NRL ORASIS algorithm with those produced by ENVI's Pixel Purity Index. Both procedures attempt to find appropriate estimations of the constituent endmembers.

Changing illumination and atmospheric conditions hamper the automated analysis of spectral imagery. Applied Analysis Inc. developed an environmental correction module as part of its subpixel classifier software. This module derives atmospheric and sun angle correction factors directly from an image without the use of predictive models. Subpixel occurrences of dark and bright surface features are used to characterize atmospheric radiance, atmospheric attenuation and sensor transfer functions. A significant component of each pixel used to derive this information can be unwanted surface reflectance from sun glint, sky illumination, or other solar-illuminated terrain materials. These spectral contributions distort the accurate assessment of atmospheric radiance, atmospheric attenuation and sensor transfer functions. By working at a subpixel level, the subpixel classifier software is able to more accurately derive these factors, resolution in improved environmental correction.

A variety of hyperspectral image pixel unmixing methods have been developed and are reported in the literature. This paper addresses the use of SOLVER, a constrained optimization technique, implemented as a feature in the Microsoft Excel software package. The method is illustrated on example data from the NEFDS spectral library. Hyperspectral imagery, i.e., imagery with more than a hundred spectral bands, has been shown to be particularly useful for identifying the material constituents of the are imaged. Since each pixel is a spectral signature, comparing that signature with a library of signatures for known materials allows each pixel's material to be identified as the one with the closest match. Since many measures of matching are used in the community, it is attractive that SOLVER allows the specification of any chosen objective function, including nonlinear expressions. This material identification process becomes an unmixing process when the pixel on the ground includes multiple materials; then the pixel is 'mixed' and no one library signature will match. Rather, a sum of library signatures, with appropriate coefficients of proportionality, that matches the pixel's signature must be determined. In this paper hypothetical pixel signatures are constructed from signatures selected from the NEFDS spectral signature library. These hypothetical signatures are then shown to respond well to SOLVER unmixing for diverse cases.

The information-efficient spectral imaging sensor (ISIS) seeks to improve system performance by processing hyperspectral information in the optical hardware. Its output may be a gray scale image in which one attempts to maximize the contrast between a given target and the background. Alternatively, its output may be a small number of images, rather than a full data cube, that retain the essential information required in the application. The principal advantages of ISIS is that it offers the discrimination of hyperspectral imaging while achieving the signal-to-noise ratio of multispectral imaging. The paper focuses on construction of the filter vectors that are needed to program ISIS. The instrument produces an image which is essentially a dot product of the scene and the filter vector. Both single vector and multiple vector approaches are considered. Also, we discuss some subtle points related to optimizing the filter vectors.

Hyperspectral data potentially contain more information than multispectral data because of higher dimensionality. Information extraction algorithm performance is strongly related to the quantitative precision with which the desired classes are defined, a characteristic which increase rapidly with dimensionality. Due to the limited number of training samples used in defining classes, the information extraction of hyperspectral data may not perform as well as needed. In this paper, schemes for statistics enhancement are investigated for alleviating this problem. Previous works including the EM algorithm and the Leave-One-Out covariance estimator are discussed. The HALF covariance estimator is proposed for two-class problems by using the symmetry property of the normal distribution. A spectral-spatial labeling scheme is proposed to increase the training sample sizes automatically. We also seek to combine previous works with the proposed methods so as to take full advantage of statistics enhancement. Using these techniques, improvement in classification accuracy has been observed.

This paper describes a study effort whose objective is to investigate the impact of hyperspectral compression on the utility of the compressed and subsequently reconstructed data for nonliteral exploitation. The goal is to investigate and quantify the extent of degradation introduced by compression that can be tolerated for various exploitation applications in order to establish acceptable compression bit rates. Two nonliteral exploitation functions were performed on the original and compressed-reconstructed image cubes produced by two hyperspectral compression algorithms at four compression bit rates on two scenes. The results showed that, in general, no appreciable degradation in exploitation performance occurred between the compressed- reconstructed and original hyperspectral data sets using these two compression algorithms. The highly encouraging results indicate that compression technology may be a viable means to significantly alleviate transmission data rate limits.

There is a need to assess hyperspectral image processing algorithms in a way that does not require applying the algorithm to a large set of spectral scenes. The statistical nature of hyperspectral scenes can be modeled as a set of means and covariances. In this model, each mean-covariance pair describes some physical component of the scene. Modeling the scene in this fashion allows non-gaussian nature of scene to be explored, with the assumption that the scene statistics are linear sums of gaussians. Once this component model of a scene is constructed, filter performance can be assessed quickly by applying the filter to the ensemble of means of covariances. Furthermore, filter performance can be predicted for scenes not yet collected, as scene models may be artificially generated from statistics of physical components. As a validation of the model we generate plots of target probability of detection versus probability of false alarm for natural scenes and models based on those scenes.

Traditional hierarchical clustering algorithms require the calculation of a dissimilarity matrix which is mapped to a binary tree or 'dendogram' based upon some predetermined criterion. Although 'optimally efficient' algorithms requiring O(N²) time and O(N) storage are known for several clustering methods, with few exceptions these algorithms are relatively inefficient in practice as many pairwise distance are measured which are not necessary for generation of the binary tree. We describe here a novel 'almost single link' algorithm which is efficient both theoretically and in practice, and which can be extended to provide fast algorithms for centroid, medium and single link clustering of large data sets. Generalization to other related clustering methods is expected to be straightforward. Our algorithm also suggests a fairly efficient method for generating minimal spanning trees. In performing the segmentation we employ a particular representation of the binary tree which simplifies the task of manual investigation of the hierarchy. A customized graphical user interface including a 2D scatter plot, a visual display of the dendogram, and a false color image with overlayered clusters makes the clustering procedure a highly interactive one. By suggesting, for each of the clustering methods, possible criteria which might be useful for extracting relevant clusters from the tree information, we are able to fully automate the cluster selection procedure and thereby further reduce the effort required to segment an image. The algorithms described have been transcribed into C code and combined into a single package, the 'hierarchical agglomerative clusterer', which has been applied to the analysis of hyperspectral image data of various forest and desert scenes acquired by the HYDICE sensor. The analyses were performed on a 266 Mhz Pentium PC platform running Windows NT 4.0. Typical segmentation times for the fastest algorithm ranged form 17 seconds for a 15232-pixel image to 2833 seconds for a 209840-pixel image, each pixel representing a 210-band spectrum. These initial studies suggest that the HAC package will provide a sound framework for making detailed comparisons of the effects of different clustering algorithms or dissimilarity measures. Its overall speed makes it a promising tool not only for hyperspectral image processing applications but for multivariate data analysis as a whole.

Generating synthetic hyperspectral data cubes for the thermal is useful because it is very expensive to design and fly such sensors. Using synthetic da is useful in performing trade-off studies, e.g. spectral and radiometric performance requirements, and also in testing new algorithms, e.g. temperature emissivity separation. We have developed a method to simulate complex thermal scenes under variable solar illumination, different materials and including gas plumes. The IDL-based scene simulation tool consists of public domain tools for: 3D geometry generation, raytracing, spectral and thermal property libraries, atmospheric transmission and emission modeling. The data can then be used in standard hyperspectral processing programs or processed by special purpose programs. We show how this model can be used to generate synthetic data cubes for dispersive and Fourier transform IR spectrometers. The data in turn is then used to evaluate algorithms to separate temperature and emissivity.

We present a new algorithm for chromotomographic image restoration. The main stage of the algorithm employs the iterative method of projections onto convex sets, utilizing a new constraint operator. The constraint takes advantage of hyperspectral data redundancy and information compacting ability of singular value decomposition to reduce noise and artifacts. Results of experiments on both in-house and AVIRIS data demonstrate that the algorithm converges rapidly and delivers high image fidelity.

We describe a tunable, 1.3 to 5 micrometers wavelength reflectance measurement system using an optical parametric oscillator (OPO) as the light source. The OPO source consists of a 1 micrometers Nd:YAG laser which is frequently shifted to 1.3-5 micrometers wavelengths using a periodically poled lithium niobate nonlinear optical crystal. The system design, calibration, and measurement of the directional-hemispherical reflectance factor and the bi-directional reflectance distribution function of different target materials are presented.

The European Space Agency (ESA) has identified the necessity to initiate a study that concentrates on the definition of an airborne imaging spectrometer which could represent a precursor to the spaceborne PRISM. The study included the definition of an Airborne PRISM Experiment (APEX) that will contribute to the preparation, calibration, validation, simulation, and application development for the PRISM mission. The APEX instrument is defined as an airborne pushbroom imager with 1000 pixels across track and 200 user selectable spectral bands over the wavelength range from 450-2500 nm. The complete APEX system will include an aircraft, navigation data, laboratory and in-flight calibration as well as a data archiving and processing facility. The definition of the specifications of the APEX instrument is based on a sensor model taking into account various parameters of the expected operation range of the instrument. The approach used defines the radiometric properties of expected scene radiances including SNR and NE(Delta) (rho) . The APEX system is presented and in compliance with the PRISM instrument, conclusions on the simulation possibilities are derived and discussed.

Imaging spectroradiometric sensor performance has been analyzed in order to develop the available space of those parameters available to the designer. This has been done in order to facilitate the choice of reasonable alternatives from which a smaller subset can be chosen to perform the very detailed analyses required to achieve an 'optimum' design. Dispersive grating spectrometers and Fourier Transform spectrometers have been considered as representative of the general class of dispersive and interferometric technologies.

We present a survey of spectral imaging for biological and medical applications. Brief philosophical and historical considerations are followed by an overview of the reasons for the modalities of achieving the fertile confluence of spectroscopy and imaging. Methods of wavelength selection at both the excitation and detection ends of an imaging system are listed and critically evaluated. A number of biological and medical applications of spectral imaging are discussed, highlighting microscopy and including our own work. We emphasize that the outlook for this research area critically depends on the further development of all component technologies, from reagents and optics to electronics and software.

Based on delay-line read-out methods of micro-channelplate (MCP) stacks we develop imaging system for single particle and photon spectroscopy. A complete system consists of an open MCP-detector with helical wire anode, specially designed front-end electronics and a stand alone PC-based TDC-system. We achieve a position resolution better than 0.1 mm and excellent linearity for open dimensions up to 100 mm, multi-hit operation, and detection rates up to 20 kiloEvents/sec in an event-listing mode or over 1 MegaCount/sec in a histogram mode. Both modes allow 2D position and time-of-flight (TOF) spectroscopy with approximately 1 nanosec TOF resolution. Furthermore, we currently test a delay-line anode on printed circuit that operates with image charge pick-up from a high-resistive collecting anode. With an image charge detection method this 3D-imaging technique can be applied to commercial sealed MCP single-photon detectors. While a simple high-resistive collection anode is placed inside the tube, a position sensitive pick-up electrode can be mounted next to it outside the vacuum wall.

Nearly all current imaging spectroscopy data are obtained by scanning airborne systems. The stability of such systems is always worse than that of spaceborne platforms. Thus, geometric distortions occur due to variations of the flightpath as well as of the attitude of the plane. These distortions cannot be corrected simply by ground control point based traditional georeferencing procedures since the movements cannot be approximated satisfactorily by polynomial transformations of the image. A pixel by pixel calculation has to be performed instead, to account for the position and attitude of the plane during the scanning process. A georeferencing procedure is described which is based on a parametric approach and theoretically allows sub- pixel accuracy even in steep terrain. The current work resulted in a new algorithm and application for parametric geocoding. A ground control point based procedure has been developed to recalibrate the offsets of the attitude data since they usually are given as relative angles. It exactly reconstructs the scanning geometry for each image pixel using position, attitude, and terrain elevation data. The procedure is tested on AVIRIS and on DAIS data and compared to digital topographic data. The geocoding results are of reliable accuracies of down to 1-2 pixels for both data sets.

The multi-sensor multi-resolution technique (MMT) was used to unmix simulated ASTER data. The simulation was performed using airborne spectrometer data of the open lignite mine Zwendkau in the Central German lignite mining district. The unmixing of low resolution ASTER thermal IR images with the reflective bands allowed for significant improvement of the spatial resolution. The radiometric accuracy was estimated using reference images and extracted pixel spectra. In comparison to other techniques, the MMT preserves the radiometric information in the TIR. Therefore, the spectral information can be used for a mineralogical analysis of the dumped material.

In this work we quantify the separability between specific materials and the natural background by applying receiver operating curve (ROC) analysis to the residual errors from a linear unmixing. We apply the ROC analysis to quantify performance of the multi-spectral thermal imager (MTI). We describe the MTI imager and simulate its data by filtering HYDICE hyperspectral imagery both spatially and spectrally and by introducing atmospheric effects corresponding to the MIT satellite altitude. We compare and contrast the individual effects on performance of spectral resolution, spatial resolution, atmospheric corrections, and varying atmospheric conditions.

A concept for processing of hyperspectral data is described which would make hyperspectral data from an operational system routinely available to customers. Customers not be required to be expert in spectral science. Customers would be offered data in a form readily usable by traditional image processing and Geographical Information Systems, with flexibility for application to their particular interests. This concept consists of an automated processing environment an a rigorous chain of algorithms to generate a variety of products, orderable by end users. The proposed processing chain can support many users, generate products within a few hours, provide repeatable information content, and enable users to focus their expertise on their area of interest and not on spectral analysis. Implementation of this concept would lead to a national standard for spectral data and products.

Many remote sensing applications rely on imaging spectrometry. Here we use imaging spectrometry for thermal and multispectral signatures measured from a satellite platform enhanced with a combination of accurate calibrations and on-board data for correcting atmospheric distortions. Our approach is supported by physics-based end- to-end modeling and analysis, which permits a cost-effective balance between various hardware and software aspects.

We describe fluorescence spectral-imaging results with the microscope computed-tomography imaging spectrometer ((mu) CTIS). This imaging spectrometer is capable of recording spatial and spectral data simultaneously. Consequently, the (mu) CTIS can be used to image dynamic phenomena involving multiple, spectrally overlapping fluorescence probes. The result presented in this paper consists of proof-of-concept imaging result using two static targets. The first is composed of 6-micrometers fluorescing microspheres and the second consists of rat sinusoid epithelial cells loaded with 0.5-micrometers fluorescing microspheres. Image data were collected in integration times of 16 msec, comparable to video frame rate integration times. The emission spectra were sampled at 10-nm intervals between 420 nm and 710 nm. The smallest spatial sampling interval presented in this paper is 1.7 micrometers .