Title: High-speed imaging device
Abstract: A high-speed image sensor has a plurality of signal converting means (30) for generating electric signals corresponding to an incident light intensity and a plurality of electric signal recording means (33) for recording electric signals output from corresponding signal converting means (30). The electric signal recording means (33) is linearly shaped and has a read-out line (58a) for each of longitudinal sections thereof. The read-out line (58) is used for directly reading out the electric signals out of a light receptive area.
Patent Number: 6,972,795 Issued on 12/06/2005 to Etoh, et al.
| Inventors:
|
Etoh; Takeharu (Mino, JP);
Mutoh; Hideki (Odawara, JP)
|
| Assignee:
|
Hispec Goushi Kaisha (Osaka, JP);
Shimadzu Corporation (Kyoto, JP)
|
| Appl. No.:
|
554882 |
| Filed:
|
September 21, 1999 |
| PCT Filed:
|
September 21, 1999
|
| PCT NO:
|
PCT/JP99/05146
|
| 371 Date:
|
May 22, 2000
|
| 102(e) Date:
|
May 22, 2000
|
| PCT PUB.NO.:
|
WO00/17930 |
| PCT PUB. Date:
|
March 30, 2000 |
Foreign Application Priority Data
| Sep 22, 1998[JP] | 10/268010 |
| Oct 29, 1998[JP] | 10/308648 |
| Current U.S. Class: |
348/311; 348/315 |
| Intern'l Class: |
H04N 003/14 |
| Field of Search: |
348/294,311,303,316,319,320,301
|
References Cited [Referenced By]
U.S. Patent Documents
| 5355165 | Oct., 1994 | Kosonocky et al.
| |
| 5614744 | Mar., 1997 | Merrill.
| |
| 6091091 | Jul., 2000 | Moon.
| |
| 6118483 | Sep., 2000 | Etoh.
| |
| 6157016 | Dec., 2000 | Clark et al.
| |
| 6157408 | Dec., 2000 | Etoh.
| |
| 6674470 | Jan., 2004 | Tanaka et al.
| |
| 2001/0010551 | Aug., 2001 | Dierickx.
| |
| Foreign Patent Documents |
| A2809393 | Nov., 1997 | EP.
| |
| A-5284282 | Oct., 1993 | JP.
| |
| A-11225288 | Aug., 1999 | JP.
| |
Other References
Takeharu Etoh, Kohsei Takehara, "An in-situ storage image sensor 1,000,000 pps
with an elongated CCD strip under each photodetector", SPIE vol. 2869, 1997.
Takeharu Etoh, "An Improved Design of an ISIS for a Video Camera of 1,000,000
pps", SPIE vol. 3642, 1999.
Takeharu Etoh, " Ultra high-speed multiframing camera with an automatic trigger",
SPIE vol. 1757, 1992.
Takehary Etoh, "A CCD Image Sensor of 1Mframes/s for Continuous Image Capturing
103 Frames", ISSCC 2002, 2002.
Takeharu Etoh, "Specifications of highspeed image sensors based on requirements
of multi-scientific fields", SPIE vol. 3173.
Takeharu Etoh, "An ISIS with curved coupled CCD channels for a video camera of
1,000,000 pps", SPIE vol. 3516,1999.
SPIE vol. 1757, Etoh, Takeharu et al. pp. 53-57, 1992.
Etoh, Takeharu, Kinki University. PP. 403-412, 1997.
|
Primary Examiner: Garber; Wendy R.
Assistant Examiner: Hernandez; Nelson D.
Attorney, Agent or Firm: Birch, Stewart, Kolasch & Birch, LLP
Claims
1. A high-speed image sensor, comprising:
a plurality of signal converters for generating electric signals according to
an incident light intensity;
a plurality of longitudinal sections of electric signal storage devices, each
of said longitudinal sections comprising a plurality of linearly shaped electric
signal storage devices for storing electric signals output from corresponding signal converters;
a plurality of drain gates, each provided at the output of one of said longitudinal
sections, each drain gate for discharging electric signals generated by an associated
signal converter to a drain line connected to each said drain gate; and
a read-out circuit connected to said drain line for directly reading-out a read-out
signal from said longitudinal sections.
2. The high-speed image sensor of claim 1, further comprising connectors for
directly connecting said signal converters with the read-out lines without passing
through said electric signal recorders.
3. The high-speed image sensor of claim 1, wherein each said electric signal
recorder is a charge coupled device type electric signal recorder.
4. The high-speed image sensor of claim 1, wherein each said electric signal
recorder is a MOS type electric signal recorder.
5. The high-speed image sensor of claim 1, wherein each of said signal converters
is divided into a plurality of portions insulated from each other.
6. The high-speed image sensor of claim 4, wherein each of said signal converters
is divided into a plurality of portions insulated from each other and wherein amplifiers
for amplifying the electric signals are interposed between said plurality of divided
portions and said electric signal recorders.
7. A high-speed image sensor, comprising
a plurality of signal converters for generating electric signals according to
an intensity of electromagnetic waves or particle streams; and
a plurality of longitudinal sections of linear shaped electric signal storage
devices, each of said longitudinal sections comprising a plurality of said electric
signal storage devices for storing electric signals output from corresponding signal converters;
a plurality of drain gates, each said drain gate connected to the output of one
longitudinal section, for discharging electric signals generated by said signal converters;
a drain line connected to said drain gates, wherein a read-out signal is directly
read-out from said longitudinal sections using said drain line.
8. The high-speed image sensor of claim 7, wherein each said electric signal
recorder is a charge coupled device type electric signal recorder.
9. The high-speed image sensor of claim 7, wherein each said electric signal
recorder is a MOS type electric signal recorder.
10. The high-speed image sensor of claim 7, wherein each of said signal converters
is divided into a plurality of portions insulated from each other.
11. The high-speed image sensor of claim 9, wherein each of said signal converters
is divided into a plurality of portions insulated from each other and wherein amplifiers
for amplifying the electric signals are interposed between said plurality of divided
portions and said electric signal recorders.
12. The high-speed image sensor of claim 7, further comprising a cuttable band-shaped
space which continuously extends from one side to another side of the light receptive area.
13. A high-speed image sensor comprising a plurality of signal converters for
generating electric signals according to an incident light intensity and a plurality
of electric signal recorders for storing electric signals output from corresponding
signal converters,
wherein said signal converters are disposed in all of, or every other, square
or rectangular frames on a light receptive area; and
wherein a center line of each said electric signal recorder, in a direction from
one position where electric signals are input from a signal converter to another
position where electric signals are input from an adjacent signal converter, is
inclined with respect to a line connecting two positions where electric signals
are input from two of said signal converters, adjacent to each other in an extension
direction of said electric signal recorders, to corresponding electric signal recorders.
14. The high-speed image sensor of claim 13, wherein each said electric signal
recorder is a charge coupled device type electric signal recorder.
15. The high-speed image sensor of claim 13, wherein each said electric signal
recorder is a MOS type electric signal recorder.
16. The high-speed image sensor of claim 13, wherein each of said signal converters
is divided into a plurality of portions insulated from each other.
17. The high-speed image sensor of claim 15, wherein each of said signal converters
is divided into a plurality of portions insulated from each other and wherein amplifiers
for amplifying the electric signals are interposed between said plurality of divided
portions and said electric signal recorders.
18. The high-speed image sensor of claim 13, further comprising a cuttable band-shaped
space which continuously extends from one side to another side of the light receptive area.
19. An image sensing apparatus comprising said high-speed image sensor claimed
in claim 1.
20. An image sensing apparatus comprising said high-speed image sensor claimed
in claim 7.
21. An image sensing apparatus comprising said high-speed image sensor claimed
in claim 13.
22. A high-speed image sensor, comprising:
a plurality of signal converters for generating electric signals according to
an incident light intensity;
a plurality of longitudinal sections of electric signal storage devices, each
of said longitudinal sections comprising a plurality of electric signal storage
devices for storing electric signals output from corresponding signal converters;
a plurality of drain lines, each provided at the output of one of said longitudinal
sections, for discharging electric signals generated by said plurality of signal
converters; and
a read-out circuit connected to said drain lines for directly reading-out a read-out
signal from said longitudinal sections.
23. A high-speed image sensor, comprising:
a plurality of signal converters for generating electric signals according to
an intensity of electromagnetic waves or particle streams; and
a plurality of longitudinal sections of electric signal storage devices, each
of said longitudinal sections comprising a plurality of said electric signal storage
devices for storing electric signals output from corresponding signal converters;
a drain line connected to the output of each said longitudinal section for discharging
electric signals generated by said signal converters,
wherein a read-out signal is directly read-out from said longitudinal sections
using said drain line.
Description
This application is the national phase under 35 U.S.C. § 371 of PCT International
Application No. PCT/JP99/05146 which has an International filing date of Sep. 21,
1999, which designated the United States of America.
TECHNICAL FIELD
The present invention relates to a high-speed image sensor and an image sensing
apparatus fitted to continuous image sensing of high-speed phenomenon.
BACKGROUND ART
In a solid image sensing apparatus for the purpose of high-speed image sensing,
the most critical factor deciding a image sensing speed is a time required to transfer
image information generated at a photoelectric transferring apparatus such as a
photodiode to a recording apparatus. Thus, most high-speed video cameras including
cameras invented by the present inventor employ method of accelerating the speed
of transferring image information from an image sensor to a recording apparatus
by concurrently reading out the information using many read-out lines (e.g., see
"4500 Frames/se. High-Speed Video Camera", by Takeharu ETOH, published at pp. 543-545
of Vol.46, No.5 (1992) of Journal of Television Society).
There is provided such a method for minimizing the time for transferring information
to the recording apparatus that the recording apparatuses are disposed adjacent
each of the photodiodes provided on the light receptive area of the image sensor,
without reading out image information from the image sensor. In this case, the
image information is transferred from all of the photodiodes at once to the recording
apparatus, so that the information transfer is performed concurrently by a total
number (usually, a few 10,000 to a few 1,000,000) of these photodiodes provided
on the light receptive area. Therefore, a transfer speed at which image information
is transferred from each photodiode to the adjacent recording apparatus provides
an image sensing speed for each frame of an image. Further, its reciprocal number
provides the number of frames per second (frame rate). The present inventor calls
such a pixel peripheral recording type image sensor an ISIS (In-Situ Storage Image
Sensor). With the ISIS, continuous image sensing is possible at an image sensing
speed of equal to or more than 100 million frames/sec.
The image sensors are roughly categorized into an analog type for processing
image information in such an analog format as charge or voltage and a digital type
for processing it in a digital format.
Ideally, one AD converter is provided for each pixel to digitize image information
from the beginning so that the subsequent processes may all be performed at a digital
circuit. Further, an ultra-miniaturized AD converter that can be incorporated into
each pixel being possible logically. However, miniaturizing the AD converter causes
a drastic deterioration in the conversion accuracy. A repetitive sampling method
etc. may be available for enhancing the conversion accuracy but may not fit ultrahigh-speed
image sensing because it involves a lot of time-consuming repetitions.
Nevertheless, a digital type high-speed image sensor may be possible
to manufacture in the future, taking into consideration that the above-mentioned
pixel peripheral recording type imaging sensor has a large pixel size enough to
reserve a space for incorporating therein an AD converter measuring a few tens
of microns in length and also that a video camera with a frame rate of about one
million frames/sec. can sufficiently be operated at a relatively low frequency
of a few mega-Hertz.
The analog type image sensors may be categorized into two groups. The first type
of an image sensor (hereinafter refereed to as MOS type or XY address type sensor)
sends an image signal to a recording apparatus near a photoelectric converter via
an electric wire through a MOS type switch. The second type of an image sensor
(hereinafter refereed to as CCD type or transfer type sensor) uses a charge-coupled
device type electric-signal transfer path (CCD transfer path) as a means for transferring
an image signal to a recording apparatus in such a manner as to use this CCD transfer
path as the recording apparatus.
The present inventors have provided a MOS type image sensor (see "Ultrahigh-speed
multi-framing camera with an automatic trigger, Ultrahigh- and High-speed Photography
and Photonics" by T. Etoh and K. Takehara, SPIE vol. 1757, pp. 53-59, 1992).
Although the MOS type sensor has been regarded as being inferior to the
CCD type in noise level, a recently available MOS type sensor can obtain an improved
SN ratio by incorporating an about 30-magnification amplifier into each pixel.
This MOS type sensor is called a CMOS-APS (Active Pixel Sensor).
When a CMOS type device is employed as the pixel peripheral recording type image
sensor, an amplifier should be ideally provided for each photodiode so that image
information may be amplified before being transferred to a recording section. The
currently available ultra-small amplifier cannot amplify a signal to a large extent
at a high frequency. Like in the case of the digital type sensor, however, it may
be highly possible to incorporate a 10-magnification-order amplifier at a stage
before transfer to a recording section, taking into account a large pixel size
and a relatively low frequency of a few mega-Hertz.
In the case of a MOS type image sensor, the larger the capacitance of a read-out
line, the larger would be the random noise. However, in the case of a pixel peripheral
recording type sensor having a few to a few tens of microns of its image information
transfer distance, the random noise can be reduced to 1/1000 through 1/100 as compared
to the case with a typical sensor of reading out the image information outside
the light receptive area. Also, the cross-sectional area of a metal wire for use
in information transfer can be substantially reduced by micro-machining. In this
case, the random error may be smaller at the time of information transfer to a
recording element in the periphery of pixels and an amplifier may be used for reading
out a signal after completion of image sensing.
Time-wise invariable fixed-pattern noise (FPN) constitutes one problem
in the MOS type image sensors. In particular, a CMOS-APS has such FPN due to irregularities
in performance of the amplifier in each pixel. However, the FPN is not much of
a problem in science and technological measurement. Namely, since it generally
employs digital image processing after image sensing, the FPN can be canceled at
the digital image processing.
In view of the above, it is highly possible to manufacture a CMOS type pixel
peripheral
recording type image sensor with current technologies.
One CCD type pixel peripheral recording type image sensor, shown in FIG. 23,
has been invented by Kosonocky et al. (see U.S. patent application No. 5,355,165).
This image sensor is capable of continuous photography at an even interval of 500,000
frames/sec. and capable of continuously capturing 30 image frames.
In a typical image sensor, a unit region consisting of one photodiode and one
accompanying CCD transfer path is called one pixel. Alternately, one unit of image
information stored in one CCD element is called one pixel. Kosonocky et al. refer
to a unit region containing one photodiode 1, multi-element CCD transfer
paths 5 and 6, a gate 3 necessary for operation, etc. which
are all shown enclosed in a dotted line in FIG. 23 as a macro pixel 4. This
is done so in order to discriminate from each other a macro pixel 4 and
the above-mentioned each CCD element or each unit of image information accumulated
therein. Hereinafter, one macro pixel is called one pixel.
First, charge generated at the photodiode 1 is transferred rightward
in the horizontal CCD transfer path 5 by five steps, so that five elements
of the horizontal CCD transfer path are all filled with charges. At the same time,
the charges are transferred into the five parallel vertical CCD transfer paths
6 disposed below the horizontal CCD transfer path 5 in the figure.
By repeating this operation, continuous images of 30 (5×6) frames can be stored.
One more operation of this vertical transfer permits five image signals stored
in the lowest line of the five parallel vertical CCD transfer paths 6 to
be shifted to a horizontal CCD transfer path 5
b in pixels directly
below. At the next step of transferring the charges from the photodiode 1
to the horizontal CCD transfer path 5, the image signals shifted from the
upper pixel 4 to the horizontal CCD 5
b are consecutively transported
rightward and disposed into a drain 7 provided at the left top corner for
each pixel in the figure (in this case, a drain 7
c in the right bottom
corner pixel 4) and then discharged out of the image sensor. Accordingly,
30 frames of the most recent image information are overwritten continuously.
This function of continuously overwrite-image sensing is extremely useful for
synchronizing phenomenon occurrence and image sensing timing which is important
in high-speed image sensing. That is, with this function, by stopping continuous
overwrite-image sensing immediately after detection of phenomenon occurrence in
an object, it is possible to reproduce image information at the time point of the
phenomenon occurrence dating back to the past from the current time point. Although
it is extremely difficult to start image sensing immediately before an ultrahigh-speed
phenomenon in an object by anticipating it, the phenomenon occurrence can be detected
readily, so that the image sensing can be stopped immediately thereafter.
In the image sensor shown in FIG. 23, however, there is a problem that the perpendicular
change in charge transfer direction changes rapidly between horizontal and vertical
directions in an alternating manner. Namely, when the perpendicular change in charge
transfer direction takes place rapidly, incomplete transfer (residual charge) cannot
be avoided, thus resulting in a deteriorated image quality. Also, an electrode
layout at a portion where the transfer direction changes is complicated, resulting
in noise occurrence. Further, the complicated electrode layout causes an increase
in a size of each element on the CCD transfer path, so that the recording capacity
(number of continuous image frames) is decreased unless the size of the pixel 4
is increased. If the pixel 4 is increased in size, on the other hand, the
total number of the pixels for the same light receiving area is decreased, thus
deteriorating the resolution.
FIGS. 24, 25A, and 25B show a pixel peripheral recording type
image sensor invented by the present inventor (see Japanese Laid-Open Patent Publication
No. Hei-9-55889, ETOH et al.: "Performance and Future of ISIS (Pixel Peripheral
Recording Type Image sensor)", and ETOH et al.: Papers of Overall Symposium 1997
Lecture on High-Speed Photography and Photonics, pp 403-412, 1997).
In this image sensor, charge generated by a photodiode 8 is transferred
through an input gate 9 to a. CCD transfer path 10, which has regularly
provided meander portions 10
a obliquely crossing between two photodiodes
8. In the CCD transfer path 10, the charge is transported vertically
in one direction as long as six pixels and then discharged to a drain 15
through a drain gate 14. Consequently, the most recent information is continuously
overwritten in each portion of the CCD transfer path 10 between the input
gates 9 and the drain gates 14.
This image sensor involves one-directional transport of charge transfer and
does not involve rapid changes in transfer direction, thus enabling complete transport
of charges. Further, the electrode layout can be simplified, thus reducing the
size of CCD elements and increasing the number of continuously overwriting image
frames or the total number of pixels.
In this image sensor, however, its pixel layout is not square but shifts a little
rightward. Since science and technological measurement involves a variety of digital
image processing on captured images after image sensing, it is easy to convert
a non-square layout into a square layout. However, since a smaller number of steps
for image processing would produce less deterioration in image quality, it is preferable
that the pixel layout be square.
Since the pixel peripheral recording type image sensor incorporates many recording
elements in each pixel, the pixel size is a few times that with a usual image sensor,
thus resulting in an extremely large sized light receiving area even with a device
having a minimum required number of pixels for reproduction of images (about 256×256
pixels). This image sensor, which is generally called a large format sensor, has
a small yield.
For example, the above-mentioned image sensor shown in FIG. 24 has, for each
pixel, eight vertical CCD elements and six parallel horizontal CCD transfer paths
10. Therefore, in order to achieve a resolution of 256×256 pixels with
this image sensor, about three million CCD elements (i.e., (256×8)×(256×6)=2048×1536)
are required. Even with current technology, it is difficult to make a CCD type
image sensor having three million elements without defect.
The CCD type image sensor generally has very small noise when charges are transferred.
A defect in a CCD transfer path inhibits information stored in the pixels upstream
side to the defect from being read out, thus resulting in a line-shaped defect
and a large decrease in yield. In the CCD type image sensor, therefore, yield provides
a largest problem in commercially producing the large format sensors. The MOS type
image sensor, on the other hand, has larger noise in reading out information from
a read-out line but manufacturing technology thereof is easier than that of the
CCD type image sensor. Also, its defects appear as a dot-shaped defect per pixel.
Such dot-shaped defect can be compensated by interpolation using image information
obtained from pixels surrounding the pixel of dot-shaped defect.
Ultrahigh-speed image sensing requires very strong illumination and
high-speed transfer causing generation of heat in the image sensor. These may contribute
to deterioration of a vital sample or occurrence of thermal noise. The image sensing
time of ultrahigh-speed image sensing is usually less than one second, so that
the time consuming factors is image sensing condition such as image angle setting,
focus setting and sensitivity regulation. With an electronic video camera, on the
other hand, an image frame rate of about 30 frames/sec. is enough to reproduce
continuously moving images and also to set image sensing conditions. Therefore,
when setting the image sensing conditions, it is necessary to perform intermittent
image sensing to prevent heat due to the illumination.
In the case that continuous image sensing at a rate of one million frames/sec.
is performed, it is necessary only to give intermittent illumination of 30 times
per second and having duration time of 1/1,000,000 with a stroboscope which has
a peak light intensity equal thereto when image sensing. Further, by executing
image sensing and transfer of the image information obtained thereby 30 times per
second in synchronization with the intermittent illumination, motion images can
be reproduce with a monitor display. This motion image is as smooth as a general
television image, thus enabling trouble-free setting of the image sensing conditions.
Further, in this case, a total time including a total illumination in setting of
image sensing condition and operation time of the image sensor is drastically reduced
to 30/1,000,000 of that for continuous image sensing. Thus, even if one hour (3,600
seconds) is required for setting of the image sensing conditions, a total illumination
time is only 3,600×30/1,000,000 second, i.e. 0.1 second.
In a CCD type image sensor, in order to reproduce an image, the image information
stored in the CCD elements including unnecessary information must consequently
be read out of the image sensor once. The large format sensor requires high-speed
transfer to read out all the image information therefrom, thus causing heating.
Therefore, to perform intermittent image sensing at the time of setting the image
setting condition with the pixel peripheral recording type image sensor using CCD
transfer paths as its recording means, the image information needs to be read out
of the image sensor rapidly.
In science and technological measurement, image measurement involves use of,
in
addition to visible light, ultraviolet ray, infrared ray, X-ray, gamma ray and
other electromagnetic waves, electron stream, neutron stream, ion stream, and other
particle streams.
DISCLOSURE OF INVENTION
In view of the above, it is an object of the present invention to provide a high-speed
image sensor comprising: a plurality of signals converting means for generating
electric signals in response to an intensity of an incident light, incoming ultraviolet,
infrared, X-ray or other electromagnetic waves, electron stream, ion stream, or
other particle streams; and a plurality of electric-signal recording means for
recording the electric signals output from each of the plurality of signal converting
means, wherein the above-mentioned electric-signal recording means is line shaped
and provided with a read-out line for each of longitudinal section sections thereof,
the read-out line being used for directly reading out the electric signals out
of a light surface. "Line shaped" here means that each of the electric-signal recording
means corresponding to each of the signal converting means is straight-line shaped
but has a bend or curve partially.
The image sensor according to the present invention, even if a defect exists
within a section of the electric-signal recording means, is capable of directly
reading out of the light receptive area electric signals generated at a pixel present
above that defect, so that defects in a reproduced image caused by that defect
can be limited to those dot-shaped ones. The dot-shaped defects can be corrected
using the image information of the surrounding pixels. This leads to improvements
in production yield.
Preferably, such a recording means is provided that is directly connected
from each of the signal converting means to the read-out line without passing through
the electric-signal recording means.
With such a construction, at the time of setting image sensing conditions by
an intermittent illumination, instead of reading out all image information in the
electric-signal recording means, instantaneous image information only upon the
intermittent illumination can be read out to monitor the image sensing state under
almost the same conditions as those at the time of ultrahigh-speed image sensing
performed after the setting of the image sensing apparatus, while preventing the
occurrence of thermal noise in the case where the electric-signal recording means
is a charge-coupled type electric-signal transfer path and the deterioration of
vital samples.
It is another object of the present invention to provide a high-speed image sensor
comprising: a plurality of signal converting means for generating electric signals
according to an incident light intensity and a plurality of electric signal recording
means for storing electric signals output from corresponding signal converting
means, wherein said signal converting means are disposed in all of or every other
square or rectangular frames on a light receptive area; and wherein a center line
of said electric signal means is inclined with respect to a line connecting two
positions where electric signals are input from two of said signal converting means
adjacent to each other in an extension direction of said electric signal recording
means to corresponding electric signal recording means. "Inclined" means that the
centerline of the above-mentioned electric-signal recording means is neither parallel
nor perpendicular to the above-mentioned straight line.
In this case, the pixels are given in a square array. A device with a square
pixel
array is easy not only in image processing after completion of image sensing but
also in creation of a cut-out line for cutting out a large format sensor for improvements
in yield.
For example, the above-mentioned electric-signal recording means may be a charge-coupled
device type electric-signal transfer path.
In this case, the most recent image information (charge) generated by the signal
converting means for each pixel can be transferred in one direction on a charge-coupled
device type electric transfer path provided in periphery at a high speed and stored,
thus providing ultrahigh-speed continuous overwrite image sensing.
The above-mentioned electric-signal recording means may also be a MOS type one.
In this case, during image sensing, charges generated at the signal converting
means for each pixel can be sequentially stored at the peripheral MOS type electric-signal
recording means, thus providing ultrahigh-speed continuous overwrite image sensing.
Preferably, the signal converting means each are divided into a plurality
of portions insulated from each other.
In this case, even if a potential gradient is provided at each of thus divided
portions of the signal converting means for the purpose of accelerating the collection
of charges, a potential between the electric-signal recording means and the signal
converting means does not become excessively large, thus securely sending electric
signals generated at the signal converting means over to the electric-signal recording
means. Also, since each divided portion of each signal converting means may be
small in area, a photodiode with each side of about a few microns to more than
10 microns, which has sure quality and is presently available, can be used to constitute
the signal converting means.
Also, in such a configuration that the electric-signal recording means is a
MOS type electric-signal recording one and the signal converting means is divided
into a plurality of portions, an amplification means is preferably interposed between
thus divided respective portions and the electric-signal recording means.
When a transistor is used as the amplification means, the larger is size perpendicular
to an intersection with a charge flowing direction (width), the higher would be
the amplification efficiency. However, if the width is too large with respect to
the current flowing directional size (length), it is difficult to maintain uniformity
of a width-directional potential. If the signal converting means is divided into
a plurality of portions, on the other hand, the amplification means each have a
smaller width but has a large total of the widths of the plurality of amplification
means and, at the same time, the width of each amplification means does not becomes
too large with respect to its length, thus acquiring uniformity of the width-directional
potential. Therefore, with such a configuration, the amplification means may have
an improved amplification efficiency and a larger allowable charge quantity.
Such a configuration is preferable that a band-shaped space is provided which
can be cut off and continues from one side to another of the light receptive area.
In this case, if the large format sensor has partially a defect, only the defect-free
portions can be cut out and used as a low-resolution, inexpensive small device.
Moreover, these portions can be combined into a device equivalent to the large
format sensor. It results in that image sensing at an ultrahigh speed continuous
images as many frames as necessary for continuous motion image reproduction at
a relatively high resolution can be executed.
It is still another object of the present invention to provide an image sensing
apparatus provided with the above-mentioned high-speed image sensor. Such an image
sensing apparatus provides a compact and inexpensive ultrahigh-speed video camera
with a frame rate of 1 million frames/sec.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a partial schematic front view showing a light receptive area of a
high-speed image sensor according to the first embodiment;
FIG. 2 is a partial schematic view showing a layout of polysilicon electrodes
in the high-speed image sensor according to the first embodiment;
FIG. 3 is a partial schematic front view showing wiring of metal wires for delivering
CCD transfer driving voltage in the high-speed image sensor according to the first embodiment;
FIG. 4 is a partial schematic front view showing a layout of metal wires for
controlling read-out operations in the high-speed image sensor according to the
first embodiment;
FIG. 5 is a partial schematic front view showing a light blocking layer in the
high-speed image sensor according to the first embodiment;
FIG. 6 is a circuit diagram showing a MOS type read-out circuit;
FIG. 7A is a perspective view showing a contact point;
FIG. 7B is a side view showing the contact point;
FIG. 8 is a partial enlarged view showing an input gate in the image sensor
according to the first embodiment;
FIG. 9 is a partial schematic front view showing a layout of pixels and CCDs
in the second embodiment;
FIG. 10 is a partial schematic front view showing a layout of pixels and CCDs
in a third embodiment;
FIG. 11 is a layout of pixels and CCDs at a center portion in the third embodiment;
FIG. 12A is a partial side view showing a photodiode having a gate region crossing
the center of the photodiode;
FIG. 12B is a cross-sectional view taken along line XVI—XVI of FIG. 12A;
FIG. 12C is a partial side view showing the photodiode having a gate region
crossing the center of the photodiode;
FIG. 12D is a cross-sectional view taken along line XVI′—XVI′
of FIG. 12C;
FIG. 13 is schematic view showing a vertical overflow/reset gate;
FIG. 14 is a partial schematic front view showing a zigzag pixel layout;
FIG. 15 is a schematic front view showing an image sensor according to the fourth embodiment;
FIG. 16 is a partially enlarged schematic front view showing the image sensor
according to the fourth embodiment;
FIG. 17 is partial schematic front view showing the image sensor according to
the fourth embodiment;
FIG. 18 is a schematic front view showing an image sensor according to the fifth embodiment;
FIG. 19 is a schematic front view showing a state where the image sensors of
the fifth embodiment are incorporated;
FIG. 20 is a partially enlarged front view showing an image sensor according
to the sixth embodiment of the present invention;
FIG. 21A is a schematic front view showing a state where the image sensors according
to the sixth embodiment are incorporated;
FIG. 21B is an enlarged view of an essential portion of a part XXIV of FIG. 21A;
FIG. 22 is a partial enlarged schematic front view showing an image sensor according
to the seventh embodiment of the present invention;
FIG. 23 is a partial schematic front view showing a conventional high-speed
image sensor;
FIG. 24 is a partial schematic front view showing another example of a conventional
high-speed image sensor;
FIG. 25A is a partial schematic view of the high-speed image sensor of FIG.
24; and
FIG. 25B is a partial enlarged view of FIG. 25A.
BEST MODE FOR CARRYING OUT THE INVENTION
First Embodiment
FIGS. 1 to 5 show a high-speed image sensor according to the first embodiment
of the present invention.
FIG. 1 shows schematically a light receptive area of a high-speed image sensor
and FIG. 2 shows an array of CCD transfer path 33 elements and polysilicon
electrodes 51
a to. 51
c. Further, FIGS. 3 through 5
each show a plane superposed on FIG. 2. Of these figures, FIG. 3 shows a first
(underlying layer) metal wire layout plane. FIG. 4 shows a second (upper layer)
metal wire layout plane. FIG. 5 shows the top plane which constitutes a light blocking
surface 50 for preventing a light from entering to anything other than photodiodes.
For easier understanding, the first metal layer shown in FIG. 3 is shown as superposed
on the layer of FIG. 2. The second metal layer of FIG. 4 is also shown as superposed
on the layer of FIG. 2. For simplification, however, the intermediate first metal
layer is not shown in FIG. 4. Similarly, FIG. 5 shows only the top plane over the
layer of FIG. 2. Thus, the high-speed-image sensor according to the present embodiment
is provided with the two metal layers except the light blocking layer 50
shown in FIG. 5, so that three-dimensional intersection of the wirings is possible,
thus arbitrarily configuring circuits regardless of the outside and the inside
of the light receptive area.
First, operations of continuous overwrite image sensing are described with
reference to FIG. 1.
Image information (charges) generated at a photodiode 30
a, which
constitutes a signal converting means, are collected at a charge collecting well
31
a and then transferred from an input gate 32
a to
a CCD transfer path 33
a which constitutes an electric-signal recording
means. In the CCD transfer Path 33
a, the charges are transferred
downwardly in the figures. This transfer of charges in the CCE transfer Path 33
a
is executed by a standard three-phase driving.
The CCD transfer path 33
a is not parallel to a perpendicular 40
connecting the input gate 32
a and an input gate 32
b included
in a pixel on the lower side of the figure with respect to a pixel including the
photodiode 30
a but inclined leftward with respect to the perpendicular
40 in the figure. With this inclination, a CCD transfer path 33
b
extending from the input gate 32
b of the lower side pixel can
be incorporated into the light receptive area.
The CCD transfer path 33
a extends straightly and obliquely downward
as long as six pixels until its bottom end, where it has a drain gate 35
a
and a drain 36
a. The image information, after being transferred
in the CCD transfer path 33
a, is discharged out of the image sensor
through the drain 36
a. As described later, the drain gate 35
a
acts also as a read-out gate. It should be noted that the drain gate 35
is supplied with an operating voltage from a metal wire 56 shown in FIG. 4.
As shown in FIG. 2, the CCD transfer path 33
a has ten CCD elements
34 for each pixel, so that six pixels correspond to 60 CCD elements. Accordingly,
the CCD transfer path 33
a always holds 60 image frames of the most
recent image information therein. In science and technological applications, the
minimum required number of continuous image frames necessary for reproduction of
motion images is 50 (<60). Thus, the motion images can be reproduced for 10
seconds at five frames/sec. in lack of smoothness.
The inclination of the CCD transfer path 33
a with respect to the
perpendicular 40 connecting the input gates 32
a and 32
b
is set in such a way that each time it advances downward by one pixel, it shifts
by one pitch of the transfer path, i.e. sum of the width of the one CCD transfer
path 33 and that of one channel stop separating two CCD transfer paths 33.
As mentioned above, one pixel length accommodates 10 CCD pixels. In addition,
both the vertical length of each CCD pixel and the pitch of the transfer path are
4.8 microns in the present embodiment. Therefore, each time the transfer path advances
by one pixel (which corresponds to 10 CCD elements in distance), it shifts leftward
by one transfer pitch as described above. The inclination is 1/10 and, in terms
of angle, arctangent 5.71 degrees. It should be noted that the breakdown of the
transfer path pitch consists of 3.2 microns of the width of the CCD transfer path
33
a and 1.6 microns of the width of the channel stop. Due the inclination
set as above, when each time the CCD transfer path 33
a advances downward
by one pixel, it shifts leftward by 4.8 microns, so that at the left of the input
gate 32
b of the lower side pixel a space is formed, which justly
accommodates one CCD transfer path 33 and one channel stop.
In such a hypothetical configuration that the CCD transfer path 33
a
connected to the upper side photodiode 30
a extends perpendicularly,
in order to provide a new CCD transfer path 33 connecting the lower side
photodiode 30
b, the lower side photodiode 30
b and the
input gate 32
b are shifted rightward by one pitch of the CCD transfer
path. In this case, centerlines of the two right and left CCD transfer paths connected
to the upper side and lower side photodiodes 30
a and 30
b
both extend perpendicularly. On the other hand, a line connecting the input
gates 32
a and 32
b is inclined rightward by 1/10, so
that the centerlines of the CCD transfer paths 33 are not parallel to the
line connecting the input gates 32
a and 32
b. Although,
such a hypothetical configuration is in principle the same as the present embodiment,
the present embodiment is superior to the hypothetical configuration because of
its square pixel array.
Also, in such a hypothetical configuration that the perpendicular CCD transfer
path 33
a connected to the upper side photodiode 30
a is
terminated at the lower end of upper side photodiode 31
a or that
of an upper side pixel including it, the CCD transfer path 33
b connected
to the lower side photodiode 31
b can be extended perpendicularly.
In this case, however, the CCD transfer path 33 accommodates only 10 CCD
elements with the number of continuous image frames being 10, so that it cannot
be a video camera because it cannot reproduce motion images.
In addition, supposing that image information generated at the lower side photodiode
30
b would be transferred to the lowering CCD transfer path 33
a,
image information from the upper side and lower side pixels is mixed, so that images
cannot be reproduced.
It is possible to permit the center line of the CCD transfer path 33
a
and the line connecting the input gates 32
a and 32
b
to extend in parallel to each other upward and downward directions respectively
in the figure. In this case, the CCD transfer path 33
b corresponding
to the lower side pixel is provided on the left side of the lowering CCD transfer
path 33
a and charges from the lower side photodiode 30
b
are sent across the CCD transfer path 33
a. This connection is
not capable of complete transfer of charge if it is given by any principle other
than the CCD principle, e.g. simple metal wires. For straddling with the CCD principle,
in the element to be straddled, the charges should be alternately transferred in
two directions like in the case of the above-mentioned image sensor shown in FIG.
23. Namely, the image information is transferred horizontally for straddling, and
vertically for sending downward. With the present level of technology it is difficult
to alternately transfer the information in two directions at a high speed.
Thus, in the high-speed image sensor according to the present invention, the
CCD transfer path 33 is inclined with respect to the line connecting the
upper and lower input gates 32
a and 32
b, it achieving
a construction in which photodiodes 30 are arranged in a square pixel array
and the CCD transfer path 33 is connected to each photodiode 30.
The pixels each have a vertical size of 48 microns (=4.8 microns×10 elements)
and the CCD transfer path 33 has a width of a length of seven CCD transfer
paths plus a length of six channel stops, i.e. 32.0 microns (=3.2×7+1.6×6).
Therefore, the photodiode 30 has a width of 16.0 microns (=48-32.0). The
pixels number 256×256 and each side of the light receptive area measures 12.288
mm. The chip size is 15 mm×15 mm. Both the chip and the camera can be manufactured
and fabricated respectively using a standard optical system.
The charge collecting well, the drain, the input gate, the drain gate, etc. are
provided at the upper region of the photodiode. The length of this region is six
microns. Therefore, the numerical aperture, a value of a photodiode's area divided
by a pixel area, is 29.2% (=16.0×(48-6)/(48×48)). It should be noted
that the effective numerical aperture could be enhanced by mounting an on-chip micro-lens.
When a desired phenomenon to be image sensed occurs, the continuous overwriting
operation is stopped so that image information held in the CCE transfer path 33
is then read out. The means for this read-out operation will be described below.
In a state where the CCD transfer path 33 has stopped its transfer operations,
one pixel (e.g., a pixel including the photodiode 30
a) is selected
of all the pixels and the drain gate 35
a at the lower end of the
CCD transfer path 33
a is opened, to perform only one step of the
transfer operation by the CCD transfer path 33. By this transfer operation,
the first piece of the image information of the 60 pieces of the image information
generated at the photodiode 30
a is discharged from the drain gate
35
a to the drain 36
a and then to one of horizontal
branched lines 58
a of a drain 58 (see FIGS. 3 and 6). All
drain lines 58 are connected to one line, thus being all connected to a
read-out circuit outside the light receptive area in read-out operations. Through
this read-out circuit, the image information is read out of the image sensor. Namely,
the read-out line for the light receptive area and the drain line are shared in
use. In addition, the drain gate also acts as the read-out gate.
To open the gates one by one by sequentially selecting one pixel as mentioned
above, such a MOS type circuit as shown in FIG. 6 is used. An electrical signal
(charge) consisting of image information is held at the lowest element of each
CCD transfer path. The drain gate 35 acting also as the read-out gate is
constituted by a vertical MOS switch (gate) as shown in FIG. 6. When a voltage
is fed from a vertical-scanning shift register 45 to one horizontal control
line 56
a of horizontal control lines 56, a horizontal series
of MOS gates (drain gates) 35 connected with this horizontal control line
56
a are opened simultaneously to be conductive. In this conductive
state, a voltage is sequentially sent from the horizontal-scanning shift register
46 to horizontal MOS gates 47 arranged outside the light receptive
area. Through a photodiode 30, which corresponds to a intersection of a
vertical signal-read out line connected to the horizontal MOS gate 47 to
which the voltage is supplied and the horizontal control line 56
a to
which the voltage is fed (the 60th image information storage element in the FIGS.
1 and 2, i.e. the lowest element of the CCD transfer path 33), the image
signal flows out to the drain line 58. Then, the voltage is sequentially
sent from the vertical-scanning shift register 45 to the horizontal control
lines 56, while scanning a horizontal-scanning shift register 46.
Thus, the image information of all pixels corresponding to a given point in time
is sequentially read out.
In the present embodiment, a defect on the CCD transfer path 33 has an
influence only on a pixel including one photodiode 30, the image information
of which is recorded by such a CCD transfer path 33, resulting in a dot-shaped
defect on a reproduced image. This dot-shaped defect can be corrected, thus largely
improving the manufacturing yield.
The following will describe a means for performing intermittent monitoring in
the setting of the image sensing conditions.
Between the drain 36 and the charge collecting well 31 is provided
an overflow gate 48, which is supplied with an operating voltage from the
metal wire 57 shown in FIG. 4. If excess charge is generated due to a strong
incident light during image sensing, the excessive charges are discharged via the
overflow gate 48 to the drain 36 for preventing occurrence of blooming.
During read-out operations, the overflow gate 48 is completely closed. The
overflow gate 48 also acts as a reset gate.
When setting the image sensing conditions, the overflow gate 48 is used
as a read-out MOS gate. This enables direct reading out of image information, without
passing through the CCD transfer path 33. During this direct read-out, the
drain gate 35 is closed. For example, if an overflow gate 48
a
of the photodiode 30
a is opened in FIG. 2, charges collected
in the charge collecting well 31
a are moved to the rightward drain
36 and read out of the light receptive area.
Illumination at the time of setting the image sensing conditions is
executed as a manner of intermittent illumination. To prevent occurrence of charge
due to incidence of a light during read-out, the mechanical shutter is closed during
read-out. The mechanical shutter is also closed when information is read out after
image sensing.
The read-out circuit is not limited to the above-mentioned MOS circuit. It may
be of a CMOS type. A CMOS circuit is an improved MOS circuit to reduce required
power. Since high-speed image sensing requires large power for illumination, it
is impossible with a battery power supply. Therefore, the present invention may
be of a MOS type.
An amplification circuit may be provided at a stage preceding the read-out line.
Also, a CMD etc. may be incorporated which enables converting a charge quantity
into a voltage to perform repetitive read-out without destroying image information.
A low-noise TSL-MOS type circuit may be used. With these approaches, random noise
peculiar to read-out by use of the read-out line can be reduced, thus enjoying
the advantage of low-noise transfer by use of the CCD transfer path.
As shown in FIG. 2, three polysilicon electrodes 51
a, 51
b
and 51
c vertically arranged are provided on each of CCD elements
34. The horizontal length of these polysilicon electrodes is 30.6 microns,
equal to the width of the transfer path region (vertically long surface between
right and left photodiodes). The width of these polysilicon electrodes is 1.6 microns.
The total pitch is 4.8 microns for these three electrodes.
In order to transfer charges in the CCD transfer path 33, it is necessary
to give a voltage for driving the polysilicon electrodes 51
a through
51
c. Reference numerals 52
a through 52
c in
FIG. 3 each indicate an aluminum wire for feeding the voltage. These aluminum wires
52
a through 52
c constitute a set. Each of these aluminum
wires 52
a through 52
c feeds first-phase, second phase,
and third-phase driving voltage. Reference numerals 53
a, 53
b,
and 53
c in FIG. 3 indicate contact points. FIGS. 7A and 7B show a
transfer path from one of the aluminum wires 52
a through 52
c
via the contact points 53
a through 53
c to the polysilicon
electrodes 51
a through 51
c. The contact points 53
a
through 53
c are to be provided in the channel stopper, so that
the width of the channel stopper is set little wider (1.6 microns as mentioned
above, double the design rule value of 0.8 microns).
FIG. 8 is an enlarged view of an input gate 32. At the input gate 32,
the first-phase and second-phase polysilicon electrodes 51
a and 51
b
extend to the charge collecting well 31, so that when they are supplied
with a driving voltage, a potential barrier between the charge collecting well
31 and the CCD transfer path 33 is lowered, thus transferring charges
into the CCD transfer path 33.
Second Embodiment
FIG. 9 shows a second embodiment of the present invention. It differs from the
first embodiment in a respect that in place of the MOS switch type read-out circuit,
a horizontal CCD transfer path 80 is provided between the photodiodes. The
drain 36 is used for continuous overwriting, blooming control, etc.
As mentioned above, it is difficult to prevent incomplete transfer due to a rapid
change in the transfer direction during the ultrahigh-speed image sensing. Therefore,
if charges are transferred from the above downward obliquely on the CCD transfer
path 33 and bent at an acute angle and then to the horizontal CCD transfer
path 80, the image information stored at an element of the horizontal CCD
transfer path 80 is deteriorated. This image information, however, corresponds
to beginning few frames of an image, so that the image information corresponding
to 60 frames except the beginning few frames is stored on the CCD transfer path 33.
The layout of the charge collecting well 31 and the overflow gate 36
acting also as the reset gate is the same as that in the first embodiment. The
layout of the electrodes, the metal wires, etc. is also the same as that in the
first embodiment. The driving voltage for the horizontal CCD transfer path 80
may be fed by either a thin metal wire passing just below the photodiode 30
or a transparent polysilicon wire. After image sensing, on the other hand, read-out
can be performed at a lower speed, thus enabling almost the complete charge transfer
even perpendicularly. Also, the transfer direction does not change alternately
but is fixed at a constant direction, so that the design of the electrodes does
not become so complicated.
In the high-speed image sensor according to the present embodiment, CCDs is used
to completely transfer image information to the outside of the light receptive
area, thus obtaining a high-quality image. Also, the construction can be simpler
because no read-out lines are necessary, thus improving the image quality.
Third Embodiment
FIG. 10 shows a pixel layout in a high-speed image sensor according to a third
embodiment of the present invention. As can be seen from FIG. 10, the present embodiment
can be obtained by changing the layout of the pixels of the image sensor provided
with the CCD transfer path having a moderate curve shown in FIGS. 24 and 25 into
a square layout. Also, it can be obtained by changing the acute-angle corner in
the second embodiment shown in FIG. 9 into a moderate-curve corner.
The four image sensors obtained by inverting and rotating the image sensors shown
in FIG. 10 can be butted along vertical and horizontal center lines 83 and
84 as shown in FIG. 11. Such a construction makes it possible, in addition
to the advantage of the above-mentioned square layout of the pixels, to cut and
butt the image sensor into ¼ with the center lines 83 and 84
as mentioned later. It should be noted that the CCD element pitch is 4.8×4.8
microns and the number of pixels is 256×256.
So a CCD type image sensor with a CCD transfer path having an oblique or curved
portion in the light receptive area has not been mentioned. In particular, a CCD
type transfer path having its repetitive moderate curve in the light receptive
area is not known to those skilled in the art. The present embodiment, however,
even as
