CCDs feature low amplifier noise, high sensitivity and wide dynamic range. Therefore, CCDs have been used in such diverse applications as general purpose imaging, industrial inspection, medical diagnostics and astrophysics. For each application, a suitable image area, optical characteristics, electrical characteristics and even appropriate package must be required. Among these, back illuminated CCDs have enhanced sensitivity for UV light, soft x-rays and electrons. Hamamatsu Photonics KK has developed FFT, MPP mode back illuminated CCDs optimized for spectroscopy. For spectroscopic applications, CCDs should have responsivity from the UV to near-IR as well as having low dark current and wide dynamic range in the binning mode. Stability of the spectral response curve is also important to achieve high precision measurements.

With the advent of ground-based 8-meter telescopes and large format detector arrays, the mid-IR (MIR) astronomy is expected to produce images with higher angular resolution and improved sensitivity than currently available. TO obtain diffraction limited images, it is necessary to compensate for atmospheric seeing. We proposed a scheme in which, by detecting the near-IR (NIR) speckle positions of a reference star, for which we have higher sensitivity than in the MIR, we superpose the MIR images taken simultaneously by shifting its position to cancel the effects of seeing. We call this two-wavelength shift-and-add (TWSAA) method. In this paper, we describe an observation carried out in the fall of 1995 to test the concept. A NIR camera was built with 256 by 256 InSb array detector. With a dichroic splitter, a pair of images of a single start at two wavelengths were formed on two halves of the array. We obtained 126 pairs of images each integrated for 0.1 seconds simultaneously at the K- and L-bands. With the TWSAA method using the K-band images as the reference, the peak signal to noise ratio of the L-band integrated image was enhanced by a factor of 5.7 compared to simple addition.

A large area, charge-couple-device (CCD) based fiberoptic taper array detector (APS-1) has been installed at the insertion-device beam line of the Structural Biology Center at the ANL Advanced Photon Source. The detector is used in protein crystallography diffraction experiments, where the objective is to measure the position and intensity of x-ray Bragg peaks in diffraction images. Large imaging area, very high spatial resolution, high x-ray sensitivity, good detective quantum efficiency, low noise, wide dynamic range, excellent stability and short readout time are all fundamental requirements in this application. The APS-1 detector converts the 2D x-ray patterns to a visible light images by a thin layer of x-ray sensitive phosphor. The phosphor coating is directly deposited on the large ends of nine fiberoptic tapers arranged in a 3 by 3 array. Nine, thermoelectrically cooled 1024 by 1024 pixel CCDs image the patterns, demagnified by the tapers. After geometrical and uniformity corrections, the nine areas give a continuous image of the detector face with virtually no gaps between the individual tapers. The 18 parallel analog signal- processing channels and analog-to-digital converters assure short readout time and low readout noise.

A novel projection display, based on a new phenomenon of voltage-induced color-selective absorption with surface plasmons, is able to generate bright image on a large screen with high efficiency. In addition, neither color filter nor phosphor is needed to generate the color. With an incident beam of +/- 7 degrees half cone angle, the color purity of this device can be as good as the CRT displays, and the contrast ratio can reach over 200:1. System analysis shows that, for a color sequential single panel 1.8 inches surface plasmon projector, the luminous efficiency can reach 3 Lumens/Watt, and for a 1.8 inches surface plasmon 'smart card' projector, the luminous efficiency can reach over 10 Lumens/Watt.

The sensitivity of a solid state detector array varies as a result of changes in transmission, diffusion effects and scattering. This variation will occur on a scale length less than that of a single pixel and its knowledge is of importance for improved device design and in applications such as event centroiding in photon counting systems. This paper reports on a measurement of sensitivity variation on a sub-pixel scale for a two-phase front-illuminated CCD. The measurement was made using a scanning reflection microscope. Variation in sensitivity between the phases within a pixel and on much smaller spatial scale are clearly observed as well as crosstalk.

This paper presents the CCD transfer concept, a characterization technique that has been successfully employed by the scientific CCD community for over two decades. The method has been shown to be an indispensable tool in characterizing, optimizing and calibrating CCDs and high performance digital camera systems in absolute terms. We introduce the CCD transfer technique, in its entirety, by presenting a complete set of transfer curves and support equations. Commercial digital camera groups, currently changing from relative to absolute performance standards, will find the CCD transfer approach beneficial.

An inexpensive measurement system for characterizing digital still cameras is described. The method adopted involves performing mathematical analyses directly on the images of test charts obtained from the DSC. These charts comprise of squarewave patterns with selected spatial frequencies, the phases of which with respect to the CCD sensors cover most or all of the phase domain. By working on the final images generated by DSCs, we can include the effects of the digital signal processing on the performance of the DSC. Slightly rotated optical patterns of squarewaves with different spatial frequencies were projected on the CCD to determine the phase-dependent CTF. These phase-dependent data were then statistically analyzed to obtain average CTF values that better describe the characteristics of the DSC. The low cost system developed in our laboratory is very suitable for evaluating and designing digital signal processing algorithms when developing digital still cameras.

In this paper we report the degree of reliability of image sequences taken by off-the-shelf TV cameras for modeling camera rotation and reconstructing 3D structure using computer vision techniques. This is done in spite of the fact that computer vision systems usually use imaging devices that are specifically designed for the human vision. Our scenario consists of a static scene and a mobile camera moving through the scene. The scene is any long axial building dominated by features along the three principal orientations and with at least one wall containing prominent repetitive planar features such as doors, windows bricks etc. The camera is an ordinary commercial camcorder moving along the axial axis of the scene and is allowed to rotate freely within the range +/- 10 degrees in all directions. This makes it possible that the camera be held by a walking unprofessional cameraman with normal gait, or to be mounted on a mobile robot. The system has been tested successfully on sequence of images of a variety of structured, but fairly cluttered scenes taken by different walking cameramen. The potential application areas of the system include medicine, robotics and photogrammetry.

A family of analog CMOS velocity sensors is described which measures the velocity of a moving edge by computing its time of travel between adjacent pixels. These sensors are compact, largely invariant to illumination over a wide range, sensitive to edges with very low contrast, and responsive to several orders of magnitude of velocity. Two successful 1D velocity sensors are described in detail; preliminary data on a new 2D velocity sensor is shown. Work in progress to extend these sensors to processing of 2D optical flow is outlined.

An 800k-pixel active pixel device has been developed for use in a digital still camera application. A mode of operation has been developed to cancel the problems of fixed patterned noise and dark current inherent to this type of CMOS imaging array. A time multiplexed read-out mechanism allows the device to operate at 5 Mpix/s and still settle to within 0.1 percent. The device, with a full well of approximately 100K electrons, is capable of delivering a S/N of 66dB, and a sensitivity of approximately 50mV/lux at 50mS exposure. The sensor is covered with a color mosaic to allow an accurate recovery of a color image. This sensor makes new in-roads for CMOS imaging into areas traditionally occupied by CCDs.

To develop a high-definition image sensor having ultra-high sensitivity far superior to CCDs. A new compact image sensor consisting of a matrix field emitter array (FEA) and a high- gain avalanche rushing amorphous photoconductor (HARP) target is studied. Most research on devices with FEAs has so far focused on flat displays and high-frequency devices. Although the concept of a flat image sensor with FEAs has been proposed before, image sensors with FEAs have yet to be fabricated. In this paper, the fundamental driving characteristics of a new flat image sensor comprised of an X-Y matrix FEA set face to face with a HARP target have been investigated in a vacuum chamber for the first time. Experiments confirmed that the device successfully picks up images. High-sensitivity is achieved with the HARP target, and the temporal fluctuation and spatial variation of the emission currents obtained from the FEA have almost no effect on the signal output currents. Furthermore, the resolution characteristics are affected by the mesh voltage and by divergence of the emission from the field emitter tips.

We have developed a new contact-type color image sensor for personal use. Three color light emitted diodes are used, for illuminating source of document. The photodiode MOS image sensor chips are used for the photodetector element, which has superior linearity characteristic. It scans an A-size document in 25 seconds at 300 dots per inch. This color image sensor is compact and very low power, so its is most suitable to sheetfed color scanner.

We present the results of a development of a photon counting intensified CCD where single MCP events due to photon or particle detections are centroided to CCD subpixel accuracy. In this mode, we have ben able to resolve individual MCP 6 micron pores indicating the ultimate resolution of this device is limited by the microchannel pore spacing and not the readout technique. Using an EEV camera with a 512 by 512 CCD and a frame rate of 60 Hz, we achieve a global counting rate of 50,000 cts. per sec. The key to achieving rates this high for a 3 by 3 centroiding system is the identification of the pixel address of the event location in hardware using a dedicated convolution DSP and comparators which pass only valid events for the centroid calculation to downstream processors. We will also discuss the centroiding algorithm development and optimization and the corrections that must be applied for quantized and truncated sampling of the event charge distribution. At very high event rates the device may also be operated in analog mode with resolution.

A compact imaging system is proposed for electronic document input. In this configuration, a linear image sensor formed on a glass substrate is coupled with an optical fiber array plate. There are multiple apertures in each photodiode of this sensor. Light from a linear light emitting diode array is transmitted through the glass, the apertures and the fiber successively and illuminates a document. The scattered light comes back in the same fiber and is detected by the sensitive part of the photodiode. Experiment with a 200dpi sensor shows a good spatial resolution. With a white paper illuminated at 100(mu) W/cm², a signal yield of 0.75 pC is obtained with 1msec storage time. The depth of field is 70 micrometers . A prototype hand-held scanner has been developed by housing this imaging system in a 10mm-wide pen-shaped speed of 12cm/sec.

A 1M-pixel progressive-scan interline-transfer CCD image sensor has been developed.It features high resolution, real- time speed and square pixels. A real-time speed was obtained by a pair of H-CCDs, located at the upper and lower of the image area and driven at 43MHz. Charges of the upper and lower halves of the image area are output separately by using the upper and lower H-CCDs, respectively. This operation enables a 1024 by 1024 progressive scan at 60 frame/sec. The device can be operated in a 512 by 512 progressive scan at 120 frame/sec by combining the charges in each of two vertically-adjacent pixels and two horizontally-adjacent pixels. Furthermore, a 1024 by 1024 interlace scan at 120 field/sec is also acceptable for either frame-integration or field-integration. The device can also be operated in a 1024 by 1024 progressive scan at 30 frame/sec by using only one of the H-CCD.

Most scientific CCD imagers are fabricated on 30-50 (Omega) - cm epitaxial silicon. When illuminated form the front side of the device they generally have low quantum efficiency in the blue region of the visible spectrum because of strong absorption in the polycrystalline silicon gates as well as poor quantum efficiency in the far red and near infrared region of the spectrum because of the shallow depletion depth of the low-resistivity silicon. To enhance the blue response of scientific CCDs they are often thinned and illuminated from the back side. While blue response is greatly enhanced by this process, it is expensive and it introduces additional problems for the red end of the spectrum. A typical thinned CCD is 15 to 25 micrometers thick, and at wavelengths beyond about 800 nm the absorption depth becomes comparable to the thickness of the device, leading to interference fringes from reflected light. Because these interference fringes are of high order, the spatial pattern of the fringes is extremely sensitive to small changes in the optical illumination of the detector. Calibration and removal of the effects of the fringes is one of the primary limitations on the performance of astronomical images taken at wavelengths of 800 nm or more. In this paper we present results from the characterization of a CCD which promises to address many of the problems of typical thinned CCDs. The CCD reported on here was fabricated at Lawrence Berkeley National Laboratory (LBNL) on a 10-12 K$OMega-cm n-type silicon substrate.THe CCD is a 200 by 200 15-micrometers square pixel array, and due to the very high resistivity of the starting material, the entire 300 micrometers substrate is depleted. Full depletion works because of the gettering technology developed at LBNL which keeps leakage current down. Both front-side illuminated and backside illuminated devices have been tested. We have measured quantum efficiency, read-noise, full-well, charge-transfer efficiency, and leakage current. We have also observed the effects of clocking waveform shapes on spurious charge generation. While these new CCDs promise to be a major advance in CD technology, they too have limitations such as charge spreading and cosmic-ray effects. These limitations have been characterized and are presented. Examples of astronomical observations obtained with the backside CCD on the 1-meter reflector at Lick Observatory are presented.

This paper will describe in some detail tow new large area CCD image sensors designed specifically to be used either as a single imager or assembled in mosaics of CCDs. The devices have 2048 X 4096, 15 micrometers pixels; the difference being the orientation of the serial register. Performance data are presented on both front- and back-illuminated parts. In addition, a new production camera test system will also be described which is being used to screen test the Advanced Camera CCDs for the wide field and high resolution channels.

The design of scientific image sensors has not progressed as rapidly as developments in the technology could have permitted. This has been a consequence of an incorrect perception by manufacturers that the scientific market was insufficiently large to justify the required investment which, in turn, has lead to the use by many scientific customers of 'do-it-yourself' design and foundry solutions to their imaging needs. This paper will introduce the basic features of the new generation image sensor and will describe the design of these devices.

DALSA's new CA-D4-1024 camera is the state-of-the-art in high-resolution, high speed, high fidelity, CCD area array image capture camera technology. Its features marry the best of both high frame rate cameras and slow-scan scientific cameras. These include ultra-low dark current at room temperature operation, low noise, high signal capacity, high fill-factor, electronic shuttering, low smear, anti- blooming, 1000 X 1000 resolution, progressive scan, user controlled frame rates up to 40 f.p.s., in-camera digitization, compact camera size, low camera power dissipation,and both 8 to 12 bit video digitization options.

A modular, multi-channel digital camera system has been designed.It provides a platform for the investigation of high-speed, high-resolution CCD sensors that are suited for dynamic medical x-ray imaging. The system supports frame rates from 30 frames per second (fps) for images with 1024 X 1024 pixels and 6-8 fps for images with 2048 X 2048 pixels. The complete system comprises camera head and correction circuits and yields a consecutive output stream of data. The camera electronics permits the testing of sensors with up to four parallel readout channels. It offers a variable bandwidth of up to 25 MHz per channel with a digitization resolution of 12 bits. The modulatory and programmability allow the adaptation of the camera system to a wide range of interesting applications. The camera electronics consists of a majority of sensor independent and a relatively small section of sensor dependent components. A detailed description of the camera, methods for performance analysis, measurement results, images, and a noise analysis are presented and discussed. As an example, linearity, MTF and noise of the camera system equipped with the FTF2020, a 2k X 2k CCD sensor from Phillips Professional Imaging have ben measured. The noise contains fixed pattern and random components. The developed camera architecture is capable of fulfilling the high requirements of dynamic, high-resolution medical imaging.

A dynamic range expansion method for charge modulation device (CMD) imagers has been proposed and an experimental circuit was evaluated. As active pixel sensors, CMD imagers offer significant advantages in terms of non-destructive readout and high-speed operation. Furthermore, the CMD imager used in this study has built-in dual vertical shift registers which can start their scanning independently. We applied our newly developed source reset method to the imager, which enables each pixel charge to be reset by external circuits. By using these functions, we have obtained our dynamic range expansion method. THe source reset method enables each pixel of a CMD imager to be reset selectively. We use it for a regional electrical shutter. For instance, if there is an overexposed pixel, its accumulated charge is reset from outside the imager, so that the pixel charge is prevented from becoming saturated. The non-destructive readout and high-speed operation enable the CMD imager to be read multiple times without destruction of accumulated charges. For instance, if the imager is driven twice in one field period, each pixel's accumulations can be checked and reset selectively from external of the device at the first scanning, then, all the signals are obtained at the second scanning and output through a small process circuit. This dynamic range expansion method for CMD imagers needs only a simple feedback circuit but does not need complex equipment externally nor adding reset circuits on a chip. An experimental circuit with this method was successfully evaluated here.

Many welding processes have been successively developed, used, and then left aside in favor of more recent ones. The current processes are welding by induction, at high or medium frequency, and welding by conduction. The procedure used during the application described herein is the high frequency soldering by induction which allows high production speed, contactless heating, and regularity and good quality of the weld, and therefore a significant advantage: the economy of electrical energy. The speed limit of the soldering is principally imposed by a good drive of the metal sheet in the profiling machine. Presently, the regulation or control parameters measured during the process are, either the voltage at the output of the H.F. power generator or the temperature measured with a bicolor pyrometer situated above the melted borders. The use of temperature as the control parameters, allows to stabilize faster the process. We propose herein a new method and approach which consists in controlling the temperature during the process with a thermal image acquisition system. Our system is composed of a CCD sensor, an acquisition card, an association of optic filters in order to capture some thermographic images. Then, applying and developing some low and high level image processings, we extract some spatial and thermal parameters which allow us to control in real time and with a low cost the welding process.

We describe recent results from a new type of CCD imager which is capable of transferring charge in all four directions and is called an orthogonal-transfer CCD or OTCCD. This device has applications in adaptive imaging where the image motion is fast in relation to the frame time. We have built a 512 X 512 pixel frame-transfer device in which the imaging section has OT pixels and the frame store has conventional three-phase pixels. We have demonstrated at the Michigan-Dartmouth-MIT (MDM) observatory on Kitt Peak that significant improvements in seeing can be obtained by partially correcting for atmosphere-induced phase distortions using this device. By imaging a bright guide star on the frame store and operating it as a fast- framing tracker we are able to measure the lowest-order distortion. We then use this information to clock the OT section to maintain registration between the charge packets and the shifting star images. The preliminary results show that the OTCCD can remove approximately 0.5 inches in quadrature with the remaining sources of broadening at MDM. Pockets are more numerous in this device than expected, and their effects are exacerbated by the multiple transfers during image tracking. Some evidence has been found that links the pockets to the use of aluminum as the fourth gate level.

Delta-doped CCDs have achieved stable quantum efficiency, at the theoretical limit imposed by reflection from the Si surface in the near UV and visible. In this approach, an epitaxial silicon layer is grown on a fully-processed CCD using molecular beam epitaxy. During the silicon growth on the CCD, 30 percent of a monolayer of boron atoms are deposited nominally within a single atomic layer, resulting in the effective elimination of the backside potential well. In this paper, we will briefly discuss delta-doped CCDs and their application of low-energy electron detection. We show that modification of the surface this way can greatly improve sensitivity to low-energy detection. We show that modification of the surface this way can greatly improve sensitivity to low-energy electrons. Measurements comparing the response of delta-doped CCDs with untreated CCDs were made in the 50 eV-1.5 keV energy range.For electrons with energies below 300 eV, the signal from untreated CCDs was below the detection limit for our apparatus, and data are presented only for the response of delta-doped CCDs at these energies. The effects of multiple electron hole pair production and backscattering on the observed signals are discussed.

The color cross-talk simulation analysis was carried out. The color cross-talk was depended on the p-well potential profile. The color cross-talk has a minimum value around 2.3 micrometers junction depth of p-well. The color cross-talk can be improved by the suitable well structure.

In this paper, we present results of the investigation of the design and operation of CMOS active pixel sensors for detection of ultra-low light levels. We present a detailed noise model of APS pixel and signal chain. Utilizing the noise model, we have developed APS pixel designs that can achieve ultra-low noise and high responsivity. We present results from two test chips, that indicate (1) that less than 5 electrons of read noise is possible with CMOS APS by reducing the size of the pixel transistors, and (2) that high responsivity can be achieved when the fill-factor of the photodiode is reduced.

First imaging results are obtained with a new CCD mosaic prototype. The CCDs are aligned using an etched socket alignment technique. Three different measurements of the alignment are made using star images, test pattern images, and microscope analysis. The CCDs have an angular misalignment of less than 30 ppm. the composite device is flat to within +/- micrometers , with rows/columns oriented to within 20 ppm. The use of an existing technology with built in precision reduces many of the difficulties and expenses typically encountered with mosaic detector construction. A new camera being built for the UBC liquid mirror telescope is also described.