Television – Camera – system and detail – Solid-state image sensor
Reexamination Certificate
1998-06-12
2003-08-12
Garber, Wendy R. (Department: 2712)
Television
Camera, system and detail
Solid-state image sensor
C348S294000
Reexamination Certificate
active
06606121
ABSTRACT:
BACKGROUND OF INVENTION
The present invention relates to a method of operating image sensors according to the “local autoadaptive method” of recording highly dynamic image scenes.
Electronic image sensors are known and are manufactured in a wide variety of technologies. Such sensors are designed so that a number of picture elements (pixels) are arranged in a suitable manner, usually as a single line or as a matrix in columns and lines. An image projected onto the image sensor is converted by the pixels into an electric signal proportional to the amount of incident light at the pixel locus. The respective proportionality constant is known as the conversion sensitivity (sensitivity) of the sensor. In a very widespread type of image sensor, the “integrating image sensor,” the pixels generate the output signal by integrating photoelectric current′-generated charge carriers over a certain interval of time (integration time) to a capacitor. The electric signal is read out by means of suitable control signals (clock signal or read signal) applied to the pixel and readout paths leading away from the pixel and is sent with suitable means to image analyzing units or image utilizing units, such as a recording device.
An important feature of an image sensor is its dynamics, defined by the minimum and maximum image brightness at the pixel locus leading to a signal that can be utilized appropriately at the output of the image sensor. The dynamics are limited at the lower end by the electric noise of the output signal: a pixel signal smaller than the electric noise cannot be utilized. The dynamics are limited at the upper end by the saturation of the components used for signal conversion and signal transmission.
Another known problem with image sensors is that their dynamics are not sufficient to completely image the brightness contrasts occurring in many applications to the output signal. Therefore, either dark image parts are swallowed by the noise and/or lighter image parts are in saturation, which can also lead to additional interference such as smear or blooming. Several methods are known for eliminating the problem of limited dynamics.
U.S. Pat. No. 5,168,532 is cited as representative of a large number of patents and publications wherein the effective dynamics of an image sensor system are increased by the “dual readout” principle. For this purpose, the image sensor system is provided with an option of varying the known sensitivity-for example, by selecting the integration time of an integrating image sensor or, even more simply, with the help of an iris diaphragm. Then two images are output by the image sensor, one at a low sensitivity and one at a high sensitivity. These two images are then combined by a suitable method to form a complete image. The method “multiple readout” which has also been patented and published in a wide variety of variants (e.g., U.S. Pat. No. 5,638,118), supplemented by the “dual readout” method only inasmuch as instead of only two images, a larger number of images with different sensitivities are recorded, stored and combined. Both methods have the serious disadvantage of a complicated sensor system, which, in addition to the image sensor, also contains means for storing and processing the image data, such as a frame grabber.
A second group of methods of increasing sensor dynamics is compression of the image signal in signal generation in the pixel. With the conventional compression method of logarithmic compression, the light-dependent photoelectric current′ is converted to a logarithmically dependent signal voltage by utilizing the logarithmic voltage-current characteristic of diodes or MOS transistors in operation below the threshold voltage, as published by N. Ricquier and B. Dierickx, for example, in “Active Pixel CMOS Image Sensor with On-Chip Non-Uniformity Correction,” IEEE Workshop on CCDs and Advanced Image Sensors, California, Apr. 20-22, 1995. These and all other compression sensors lose image details due to dynamic compression. In addition, fixed interference (so-called fixed pattern noise or FPN) which occurs in the pixels and the signal paths due to local fluctuations in component parameters such as transistor threshold voltages or capacitances is amplified exponentially, which must in turn be corrected by expensive measures. Other methods (for example, U.S. Pat. No. 5,572,074 or “Analog VLSI Phototransduction by Continuous-Time, Adaptive, Logarithmic Photoreceptor Circuits” by T. Delbruck and C. A. Mead in
computation and Neural Systems Program, Technical Report, CNS Memorandum No
. 30, 1994, pages 1-23, California Institute of Technology, Pasadena, Calif. 91125) control the pixel sensitivity locally through the output signal produced by the pixel itself with the help of a complicated pixel circuit; however, this method again effectively corresponds to the compression method with the same disadvantages resulting from it.
Locally adaptive methods are better methods of operating image sensors with high dynamics. A locally adaptive method is characterized in that the sensitivity of the image sensor is not adjusted for all pixels at the same time (globally) but instead is adjusted for smaller subgroups, preferably individually for each pixel (locally). According to the nature of the image sensors, it is possible to connect at different points in the signal path.
Methods that reduce, pixel by pixel, the light striking the pixel are described in U.S. Pat. No. 5,410,705 (attenuation by polarizers) and U.S. Pat. No. 5,218,485, for example. All these methods presuppose a more complex optical structure and an expensive external system for controlling the attenuation elements.
Simpler systems with higher dynamics are achieved with such locally adaptive methods which control the integration time on a pixel by pixel basis. A conventional method with locally adaptive integration time uses individual pixel reset, IPR, for example, as reported by O. Yadid-Pecht, B. Pain, C. Staller, C. Clark and E. Fossum in “CMOS Active Pixel Sensor Star Tracker with Regional Electronic Shutter,” in
IEEE Journal of Solid-State Circuits
, vol. 32, no. 2, Feb. 1997 and in “Wide Dynamic Range APS Star Tracker,” in
Solid State Sensor Arrays and CCD Cameras
, San Jose, Calif.,
Proceedings of the SPIE
, vol. 2654 (1996) pp. 82-93, and by S. Chen, R. Ginosar in “Adaptive Sensitivity CCD Image Sensor,”
Proceedings of the SPIE
, vol. 2415 (1995) pp. 303-309. With the sensors described here, the pixel circuits have been modified so that their integration capacitor can be reset individually in each pixel at any time. Due to this fact, an individual integration time can be achieved for each pixel in certain limits, and therefore the pixel sensitivity can be adapted to the light sensitivity striking the respective pixel locus. The methods described here are characterized by a low additional expense in the pixel circuit with high dynamics at the same time.
With the locally adaptive image sensor (LAS) developed at the Institute for Semiconductor Electronics at the University of Siegen, the integration time of each pixel can be programmed individually into the pixel in the form of an analog voltage before the actual integration phase (see T. Lulé, H. Fischer, S. Benthien, H. Keller, M. Sommer, J. Schulte, P. Rieve, M. Böohm, “Image Sensor with Per-Pixel Programmable Sensitivity in TFA Technology”; H. Reichl, A. Heuberger, Micro System Technologies '96, VDE Verlag, Berlin, Offenbach, pages 675 ff., 1996).
A disadvantage of the IPR and LAS methods, however, is the considerable expense in terms of supplementary circuits which are needed to generate, pixel by pixel, the exposure times required for the next integration cycle and the resulting driving clock pulses from the image read out last. The additional disadvantage of the IPR methods is the problem that all pixels must be reset in the desired sequence during the integration phase, which leads to collisions and non-deterministic clock pulse overcoupling on the sensitive pixel electronics. In addition, all the adaptiv
Böhm Markus
Lulé Tarek
Bach Klaus J.
Garber Wendy R.
Wilson Jacqueline
LandOfFree
Local auto-adaptive optic sensor does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Local auto-adaptive optic sensor, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Local auto-adaptive optic sensor will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3104927