It has been approximately five years since Kodak introduced t1 notion ofan "Imaging Chain' at anElectronic Imaging Workshop in Tokyo. That metaphor gave rise to a diagram that shwed how conventional silver-halide and electmnic photography could relate to one anot1r. At this Symposium we will take this idea one step further. We will discuss a new idea —the Digital StilL read noise, x-ray, ultra-violet, high speed readout.

A data sheet is presented outlining the performance and characteristics of a Kodak DCS 200mi camera. In addition to providing information on this camera, the format and content of the data sheet could serve as a guide in the organization and display of pertinent information on electronic still cameras in general. Such data sheets are already common in silver halide photography. Pictorial electronic still photographers could benefit greatly from the publication of data sheets which enable them to judge the possibilities and limitations of the hardware and software. The proposed data sheet will contain descriptive specifications and numerical values such as sensor type, active sensor area, photoelement size, fill factor, filters, normal lens focal length, lens MTF, shutter type, shutter speed range, exposure metering modes, image storage, data output, compression, saturation speed, noise equivalent quanta (NEQ) speed, resolution, spatial frequency response (SFR), optoelectronic conversion function (OECF), noise power spectrum, effective bit depth (and spatial frequency dependence), peak equivalent image quality (EIQ) values, spectral response, etc. Where appropriate, the numerical values will be calculated according to the developing ISO standards in electronic photography. The graphical format of the data sheet as well as the information depth will be discussed.

Most real time CCD video cameras are designed to operate on TV format. However, TV format CCD cameras have some basic limitations in many applications due to the interlace scanning, limited horizontal scanning lines per frame and relatively low scanning speed. Non-TV format cameras have now been developed based on progressive scan interline transfer CCDs in order to optimize performance in high speed machine vision, electronic shutter applications, dynamic motion frame capturing, high resolution real time imaging and direct computer interface applications. The simultaneous exposure of all pixels and the structure of the substrate drain mechanism make it possible to operate the high speed electronic shutter with full frame resolution and uniform long integration. The digital processing, built-in frame memory and output format are described herein for asynchronous reset and image capturing, long integration image capturing, and scan conversion. The noise characteristics and the specific CCD characteristics during high speed electronic shutter are studied from a camera design standpoint. A Peltier-cooled CCD camera and an image intensifier-to-camera coupling are studied for real time, high resolution fluorescent microscopy applications. A progressive scan color camera is also described. The CCD color filter, color processing (digital processing), and output format are discussed.

A high speed CCD camera system has been developed for commercial, industrial, and scientific applications. The system incorporates two video channels, each capable of running at rates up to 20 million pixels/second. The output of each channel is digitized to provide true 8-bit output signals. The camera offers high linearity, low noise, wide dynamic range, and low shading. The camera can be used in a range of applications, including medical, industrial inspection, microscopy, and surveillance.

A prototype of portable digital still camera (DSC) used as a computer input device, and a consumer product as well, is developed in the laboratory. The system performs auto focus, auto exposure, and auto white balance, associated with a zoom lens and its controller. The auto exposure algorithm used fuzzy logic and picture segments for exposure decision. A DSP ASIC has been developed for color image processing. The images are compressed with a JPEG IC, and are stored in a PCMCIA memory card. At compression ratio 18 to 1, a 2 Mbyte memory card can store about 36 color images. An energy saving strategy is implemented in a power control module, and results in two hours operation time or 200 pictures by using a normal camcorder battery. A color LCD panel is used for real time display and playback. The SCSI interface driver has been designed to provide PC connection. The system also has NTSC and S-video connection terminals for real time video display.

We have developed a new progressive scan CCD imaging system, which realizes high vertical resolution of 480 TV lines at 60 frames/sec. This system comprises the following devices: the image sensor which has square pixels suitable for VGA, the timing generator which has a new asynchronous electronic shutter function and various external synchronization functions. Since the minimum exposure period of the asynchronous electronic shutter is 1/10000 sec, an object moving at high speed can be captured as a full frame still image. Consequently, this system is suitable for industrial applications such as factory automation.

A motion image acquisition technique that enhances resolution by combining periodic position-shifting of the CCD sensor with motion adaptive processing is described. The technique is applied in a prototype system that employs an HDTV CCD sensor that is shifted alternately by half a pixel pitch in the horizontal and vertical directions in synchrony with the frame frequency. The series of four frame images so obtained is integrated into one image as an output frame image for each frame. The resolution is enhanced by this integration, but this technique is not effective for moving regions because of image lag. Therefore, moving regions are reproduced by using only the latest frame image. Still and moving regions are synthesized using the movement coefficient derived from the frame difference signal over the past four frames. The resolution of the moving region is half that of the still region. However, this difference in resolution cannot be perceived clearly because the resolving power of the human eye is low for a moving region. An evaluation of the prototype system confirmed that this technique enhances the resolution in both the vertical and horizontal directions, and that high-quality images far superior to HDTV can be output.

We are developing an astronomical imaging system which employs a thermoelectrically cooled focal plane consisting of two 'edge-buttable' Loral 2048 X 2048 pixel CCDs. To allow strip scanning, the columns of the CCDs are mutually aligned on a custom Kovar mount. The clocking and bias voltage levels for each CCD are independently adjustable, but both CCDs are operated synchronously. Each chip is read out from one output and measured at 14 bits with commercially available A/D converters at a rate of 250 kpixels/s, permitting scanning across the sky at up to 1000 deg²/hr (about twenty times faster than the equatorial sidereal rate) to a limiting magnitude (S/N equals 3) near V equals 19. The instrument will be used as part of the Lowell Observatory Near-Earth-Object Search (LONEOS) using a 57-cm Schmidt telescope at Lowell Observatory in Flagstaff, Arizona.

At the Copenhagen University, we have developed a compact CCD camera system for single and mosaic CCDs. The camera control and data acquisition is performed by a 486 type PC via a frame buffer located in one ISA-bus slot, communicating to the camera electronics on two optical fibers. The PC can run as well special purpose DOS programs, as in a more general mode under LINUX, a UNIX similar operating system. In the latter mode, standard software packages, such as SAOimage and Gnuplot, are utilized extensively thereby reducing the amount of camera specific software. At the same time the observer feels at ease with the system in an IRAF-like environment. Finally, the LINUX version enables the camera to be remotely controlled.

CCD detector arrays have potential applications in the field of Atomic Emission Spectroscopy (AES). If used with a cross dispersive polychromator, a two dimensional array offers the possibility of providing fast, simultaneous, multielement analyses. This paper describes the characterization of a 1280 X 1024 pixel CCD with antiblooming structures, together with a flexible camera system. The characteristics are considered with respect to AES. The results presented in this paper cover the aspects of, dark currents, spurious events, CTE at low signal levels, linearity, full-well antiblooming levels. Conclusions are made about the suitability of this CCD detector to the above application.

Serial readout of cooled, linear Si-photodiode arrays or CCDs offer a high dynamic range but only at moderate pixel readout rates. Recording spectroscopic information, i.e. 1D images, requires only a few sensitive lines of a 2D CCD array. A camera is presented where the unexposed part of the CCD is used as a buffer to store successive spectra with high frame rates. In this 'streak mode' the time resolution for data acquisition depends only on the line shift time of the CCD and no longer on the slow pixel readout time. A frame rate of 10 kHz for X-ray absorption spectra has been achieved. The CCD is part of an X-ray camera consisting of a scintillating screen lens-coupled to the CCD. The camera provides a high dynamic range of 17 bit, a spatial resolution of 60 micrometers (FWHM) and a high detective quantum efficiency of > 40% for x- ray energies between 4 keV and 25 keV. The camera is used for time- resolved energy-dispersive XAFS (X-ray Absorption Fine Structure) experiments. This technique permits the study of time-dependent variations of electronic properties and the local environment of atoms under induced external perturbation.

Turbulence is known to be the factor limiting resolution in long exposure images taken through the earth's atmosphere. The degradation is such that the images obtained can be described by the classical Seidel aberrations. Adaptive optics can be used to remove these aberrations to the maximum extent possible with corrective optics and a reference object. Full compensation systems are preferred, but are costly and complex. Significant improvement in image quality can be obtained by removing only the lowest order aberration, that is random tilts. These 'partial adaptive systems' have been used to produce high quality images in the visible and infrared. For most systems, determining the orientation of the incoming wavefront has required two separate imagers; one for determining wavefront variations and another for integration of photons in the field of interest. This has resulted in a need to split the incoming light with beamsplitters or choppers. In this paper, a new method of tracking tilts with only one focal plane array has been devised. Experiments with a Charge Injection Device (CID) used in conjunction with tip-tilt mirrors has been shown to be capable of first order motion correction. Limitations of this overall system as well as results and analysis from experiments will be presented.

Eastman Kodak company Motion Analysis System division has invested many years in developing technology used in our new 24-bit color accurate digital motion analyzer. this paper describes the method for producing accurate 24-bit color at 1000 fps. additionally, applications will be discussed using new tools embedded in our motion analyzer.

An imaging system for monitoring traffic on multilane highways is discussed. The system, named Safe-T-Cam, is capable of operating 24 hours per day in all but extreme weather conditions and can capture still images of vehicles traveling up to 160 km/hr. Systems operating at different remote locations are networked to allow transmission of images and data to a control center. A remote site facility comprises a vehicle detection and classification module (VCDM), an image acquisition module (IAM) and a license plate recognition module (LPRM). The remote site is connected to the central site by an ISDN communications network. The remote site system is discussed in this paper. The VCDM consists of a video camera, a specialized exposure control unit to maintain consistent image characteristics, and a 'real-time' image processing system that processes 50 images per second. The VCDM can detect and classify vehicles (e.g. cars from trucks). The vehicle class is used to determine what data should be recorded. The VCDM uses a vehicle tracking technique to allow optimum triggering of the high resolution camera of the IAM. The IAM camera combines the features necessary to operate consistently in the harsh environment encountered when imaging a vehicle 'head-on' in both day and night conditions. The image clarity obtained is ideally suited for automatic location and recognition of the vehicle license plate. This paper discusses the camera geometry, sensor characteristics and the image processing methods which permit consistent vehicle segmentation from a cluttered background allowing object oriented pattern recognition to be used for vehicle classification. The image capture of high resolution images and the image characteristics required for the LPRMs automatic reading of vehicle license plates, is also discussed. The results of field tests presented demonstrate that the vision based Safe-T-Cam system, currently installed on open highways, is capable of producing automatic classification of vehicle class and recording of vehicle numberplates with a success rate around 90 percent in a period of 24 hours.

A new 6000 element line scan camera is described in this paper which is capable of very high speed operation. Unlike previous line scan cameras with such high resolution, this camera is capable of data rates up to 30 Mpixel/sec, provides video digitized to 8 bits, is a single unit, and is very compact. The speed and resolution of this camera provides vision system designers new flexibility to inspect with a single camera where multiple cameras were previously required, achieving significant system integration cost and time savings. This paper describes the performance and features of this camera. Results show that despite the large size of the image sensor, performance does not degrade significantly with speed.

Electronic cameras using a single CCD detector acquire scene color by subsampling in three, color planes and subsequently interpolating the information to reconstruct three, full-resolution color planes. The nature and size of the interpolation errors are a function of the algorithm used. When interpolation errors are propagated through the rest of the imaging chain, it becomes evident that synergistic effects among image processing operations must be considered when selecting and tuning an interpolation algorithm. This presentation demonstrates and comments on these image processing interactions.

The green, red, and blue color filters used for CCD sensors generally have different responses. It is often necessary to balance these three colors for displaying a high-quality image on the monitor. The color filter arrays on sensors have different architectures. A CCD with standard G R G B pattern is considered for the present discussion. A simple method of separating the colors using CDS/H that is a part of KASPs (Analog Signal Processors made by Kodak) and using the gain control, which is also a part of KASPs for color balance, is presented. The colors are separated from the video output of sensor by using three KASPs, one each for green, red, and blue colors and by using alternate sample pulses for green and 1 in 4 pulses for red and blue. The separated colors gain is adjusted either automatically or manually and sent to the monitor for direct display in the analog mode or through an A/D converter digitally to the memory. This method of color balancing demands high-quality ASPs. Kodak has designed four different chips with varying levels of power consumption and speed for analog signal processing of video output of CCD sensors. The analog ASICs have been characterized for noise, clock feedthrough, acquisition time, linearity, variable gain, line rate clamp, black muxing, affect of temperature variations on chip performance, and droop. The ASP chips have met their design specifications.

MOS photodetector control technology that enables nonlinear charge collection at the photosite. This innovation permits breakthrough performance of CCD image sensors by overcoming previous limitations in the measurement of absorption characteristics of film and other nonlinear forms.

A 1D camera was developed to provide a small high-dynamic-range, high- resolution imager for use in a variety of optical processors. The camera is built around a high-performance 16-port 1024-pixel by 1-line charge- coupled device (CCD) array. The array and associated drive circuitry are integrated into a compact camera head. The camera contains a three-board set consisting of the array and its output buffers, a clock driver board, and a timing control board. Because of the parallel output structure, the readout time can be less than 10microsecond(s) , giving a maximum data output rate of over 100 million pixels per second. The camera has a dynamic range of over 55 dB of optical intensity. The camera electronically buffers the CCD outputs to drive 75-ohm coaxial cables to an external array of analog multiplexers and an analog to digital converter (ADC). At the highest transfer rate of over 100,000 lines per second, 16 ADCs could be used to read all the channels simultaneously. The camera also provides transistor-transistor logic (TTL) clocking waveforms, which are converted to metal oxide semiconductor (MOS) voltage levels to control and drive the CCD array. In order to allow a sufficient amount of flexibility, the camera provides different adjustable parameters and interface options for the user. Interface options include selection of an internal or external clock source, a shutter signal, and an internal/external trigger source.

CCD imagers introduce artifacts into the images they reproduce with a variety of mechanisms related to sampling by a regular array of finite- size pixels. Two classes of artifacts are aliasing and modulation transfer function (MTF) limitations. This work analyzes the origin of both effects and concludes that aliasing is solely an artifact of the sampling process and that MTF limitations are due to the properties of the pixels. The dithering technique is intended to decrease these limitations by repeatedly sampling the same scene. We show in this paper that dithering is effective in removing aliasing artifacts but cannot avoid MTF limitations inherent in the image. In situations where dithering is effective, we analyze some errors likely to be encountered in implementing the technique.

We have developed a compact HDTV camera tube that combines high sensitivity with high resolution and is compact enough for hand-held cameras. This new camera tube employs an 8-micrometers -thick HARP (High-gain Avalanche Rushing amorphous Photoconductor) target. Unlike other photoconductors, this target is unique in that its sensitivity can be increased to very high levels to cope with darker illumination. In addition, we have achieved high resolution over the entire picture through the use of a new all-electrostatic focusing and deflection system in the 18-mm-diameter tube that facilitates the use of a narrow electron beam while assuring the optimal suppression of aberration. a prototype hand-held HDTV camera equipped with three of these camera tubes has attained a maximum sensitivity of 2,000 lux at f/25 (normal gain) and a limiting resolution of more than 1,400 TV lines.

In recent years polymer films based on bacteriorhodopsin (BR) have attracted a lot of attention in the area of optical imaging systems. The high photosensitivity of these films allows the processing of low-power optical signals (several mW/cm² CW gas laser irradiation). Spatial resolution does not fall below 5000 lines/mm, photoresponse time is 50 microsecond(s) and images can be recorded and erased over million cycles. Polymer film with BR combine a dynamic recording with optical image processing. The characteristics of anisotropically-saturating nonlinearity of polymer films with BR allow a suppression of the background with greater intensity than usable signal intensity of be performed. Low saturation intensity of the polymer films with BR allows the operation of the polarization of low-intensity signals to be realized. Nonlinear photoresponse of the high photosensitivity BR genetic variant Asp96-Glu is studied in this work too. We hope that the polymer films based on BR and its analogs will find potential use precisely in the medical low- light-level imaging systems.

The steady development of megapixel detector arrays with decreasing pixel size has improved the performance of present imaging systems. These high spatial resolution detectors have been incorporated into a variety of scientific experiments. The sensitivity of the diode arrays has allowed significant progress in instrumentation development and application. However, full application of these detectors to low light level measurements has been hampered by the lack of image intensifiers which can fully exploit the available spatial resolution of the diode arrays. Current architecture of image intensifiers allows significant room for improvement. We involved in a project to design, develop and characterize an 18 mm GEN II image intensifier with improved spatial resolution. Recent advances in microchannel plate production and fiber optic architecture have been exploited to produce a series of image intensifiers. A production run of a series of tubes with reduced cathode to MCP spacing, reduced microchannel diameter and pitch, reduced MCP to phosphor screen spacing and an integral fiber optic taper has been carried out. This intensifier output will be visually examined and coupled to a megapixel array for digital characterization. The goal is to produce a significantly higher limiting spatial resolution to allow for improved measurements in scientific, commercial and military applications. First results from this production run will be discussed and compared to physical performance models.