JPS59108490A - Image pickup device - Google Patents

Abstract

PURPOSE:To improve the resolution and color reproducibility by providing a checkerswise color separation filter including two kinds of color information to an incident optical path at each horizontal line and dividing the information in response to the color of the filter for reading out it to attain accurate color separation. CONSTITUTION:The color separation filter is provided in the incident optical path so that G, R signals are stored in the n-th line of the image sensor and B, G signals are stored on the (n+1)-th line. These signals are transferred to the 1st and the 2nd shift registers 31, 32. The registers 31, 32 are outputted by horizontal transfer pulses shifted by the phase by 180 deg. with each other.

Description

DETAILED DESCRIPTION OF THE INVENTION (Technical Field) The present invention relates to an imaging device using a solid-state imaging device.

(Prior art) Conventionally, this type of device uses an X-Y address solid-state image sensor called a MOS type, or an interline type C
OD, frame transfer type COD, etc. are often used. Among them, the frame transfer type COD is structurally very simple compared to the MOS type and the interline type because it does not require a vertical transfer register in the imaging section.

Therefore, it is possible to integrate a large number of horizontal pixels corresponding to the horizontal direction of a TV screen.

FIG. 1 is a diagram showing a conventional frame transfer type COD like the one in which there is an imaging section 1 that performs photoelectric conversion, a memory section 2 that temporarily stores the charge from the imaging section, and a further section of the memory. It consists of a horizontal shift register 3 that transfers accumulated charges in accordance with TV synchronization, and an output amplifier 4 that reads out the charges as a voltage signal. CC like this
On D, color separation filters necessary to create color signals are glued or on-chip.

FIG. 2 is a diagram showing an example of the filter.

Here, a conventional frame transfer type COD using the R (red), G (green), and B (blue) stripe method, which is said to have excellent color reproducibility, will be described.

When the number of pixels in the horizontal direction of the stripe method is approximately 580 elements, the horizontal transfer frequency corresponds to 10.7 MHz, but when using a COD with this number of elements, the output signal of the CCD is normally used as a high-bandwidth signal as a luminance signal. R, G, B
Color signal separation is often performed using a sample and hold circuit for each signal with a repetition frequency of 3.58 MTfz.

In this case, the color signal is 500 Kl (
However, in this embodiment, the sampling frequency is 3.58 MHz, and the signal band can be reproduced up to the Nyquist frequency, so there is no problem. However, regarding the luminance signal, there is no problem when the subject image is close to achromatic, but for subject images with high color saturation, the sampling frequency is 3.58 MHz (Nyquist i, s
MHz), a considerable amount of aliasing distortion occurs, significantly degrading the image quality. In order to solve this drawback, the number of horizontal pixels is approximately 770, that is, 14MH.
The need for z-drive arises. With such a 14MT(z compatible CCD), the sampling frequency is approximately 4.77MT(z(
The Nyquist value is 2.4 MT(z), which makes it possible to significantly reduce aliasing distortion and cause no problem in general television receivers.

However, in COD compatible with 14 MHz, horizontal shift register, output amplifier, clock IC, color separation sample,
The following problem occurs on the hold circuit.

First, in order to separate the color signals R, G, and B from the output signal of the COD, the period of 14 MT (corresponding to the effective signal component of the z signal) must be long to some extent. At the time of a pulse, the signal component is calculated to be 2X 1818MHz'35ns, but in reality, if the total rise and fall times due to the component elements of the drive circuit are about 10n8, the remaining part is about 25n8. 1.2.Furthermore, if the clock rises and falls too steeply, dark current increases due to heat generation in the shift register, resulting in an even shorter period for the effective signal portion.Also, the frequency characteristics of the output amplifier If we also take into account the effective signal period, it becomes even shorter.In this way, 1
In the COD compatible with 4&, the period of the signal component is considerably short, and fluctuations in the sample pulse due to variations in the elements of the clock generating section or temperature fluctuations are also added, so there is a drawback that it is difficult to separate the signals. Another drawback is that it is currently quite difficult to integrate a 770-element CCD as a horizontal shift register for an image sensor of, for example, a 1/2-sized image sensor. In addition, when 770 horizontal elements are driven at 14 Wz in this way, if sampling is performed using a stripe filter as shown in Figure 2, the Nyquist frequency of each color signal will be 2.4&, which is usually sufficient. However, even this is insufficient when trying to obtain an image signal of even higher quality. Such demands are particularly strong for printers and high-definition televisions.

(Objective) A first object of the present invention is to provide an imaging device capable of eliminating the various drawbacks of the prior art as described above.

A second object of the present invention is to provide an image sensor and an image sensor that can omit a sample-and-hold circuit configuration for color separation.

A third object of the present invention is to provide an imaging device that can accurately perform color separation of an image sensor having a large number of elements in the horizontal direction.

A fourth object of the present invention is to provide an imaging device capable of forming an image signal with high resolution and good gray reproducibility.

(Example) The present invention will be explained below based on Examples.

In order to eliminate the above-mentioned drawbacks, this embodiment uses W13 instead of the conventional COD having one horizontal shift register.
The above-mentioned problems are basically solved by using a CCD having two horizontal shift registers as readout transfer paths as shown in Figure +a+ and setting the horizontal transfer frequency to i. That is, the pixel charge in the horizontal direction is transferred to two horizontal shift registers at a period of one pixel by the clock circuit 46 of FIG. 6 or 7, and the horizontal transfer frequency is set to i.

In this case, although a frame transfer type COD is explained in the embodiment, it is of course applicable to an interline transfer type CCD in exactly the same way.

In this embodiment, the color filter shown in FIG. 3 1b+ is attached to the CCD shown in FIG. It is driven to allocate each cell to the registers 31 and 32. As a result, the first and second shift registers 31.3
As shown in FIG. 3, 1b+, G and R signals are accumulated in the outputs s1 and S2 of 2 for the nth line, respectively, and (n+1
) For the line, B and G signals are accumulated, respectively.

Further, each register 31, 32 is driven by horizontal transfer pulses whose phases are shifted by 180° from each other.

The clock circuit 46 not only performs horizontal transfer of the CCD, but also functions as a control means for controlling the distribution of charges to the registers 31 and 32.

Figure 6 shows the first and second register outputs 81°S2 to Y
FIG. 2 is a diagram showing an example of a signal processing circuit for forming t (RY) and (B-Y) signals. Shift register 31.
After amplifying the signals 81 and 82 which are output more than 32 times by amplifiers 5 and 6, they are passed through IH delay lines 7 and 8, respectively, thereby producing signals delayed by one horizontal period. G here
Considering the signals, as shown in FIG. 3, the G signal of the n line is stored in the shift register 31°, and the G signal of the (n+1) line is stored in the shift register 32°. n
In order to match the spatial sampling of the signals of the lines and (n+1) lines, the clock periods are set so that they are read out with a shift of half a period as described above.

By switching here, a synchronized signal is obtained for each horizontal line. Similarly, a line switch 12 obtains a simultaneous signal for each line from the 12 signals. These signals are added to adder 1
Add in step 4. This added signal is 81. Since the G signal of S2 is shifted by 1/2 period, the sampling rate will be doubled after addition. The output signal of the adder 14 is passed through a low-pass filter 15 to remove clock pulses, and then amplified by an amplifier 19. This signal is
As mentioned above, since the sampling rate is high, a broadband G signal is generated, which is referred to as GW.

On the other hand, line 5WIO converts the 81% and 82 signals to 1
/2fm<αH: horizontal scanning frequency) r: By selecting, a G signal suitable for spatial sampling of the − line can be obtained. This is passed through a low-pass filter 16 to remove clock pulses, and then amplified by an amplifier 20 to obtain a G signal GL having a narrower band than the signal GW.

Regarding the R and B signals, as shown in FIG. 3 1dl, the B signal is sent to the shift register 31, and the R signal is sent to the shift register 32.
A signal is accumulated every IH.

Since the B signal is output to Sl only every IH, the line switch 11 selects and selects the IH-delayed signal and the original signal for each IH, and read them out simultaneously for each line.

This EV signal is passed through a low-pass filter 1 to remove the clock signal and obtain the B signal.

Similarly, the R signal is simultaneously synchronized by the line switch 13 and outputted from S2. A low-pass filter 18 removes the clock from this signal to obtain an R signal.

Process the R, B, and GL signals obtained in this way.
The YL flaw signal color difference signal (
R-YL), (B-Yt,) are generated, and at the same time, a false signal is subtracted from the above-mentioned wideband G signal Gw in the subtraction circuit 21 to generate a G high frequency signal GH. GW obtained in this way
signal and YL made from GL + B * R% signal
The flaw signal is added in the addition circuit 22 to produce Y as a luminance signal.
Get a signal. In this way, NTSC with full bandwidth
A luminance signal Y is formed. Further, these luminance signal Y and color difference signals R-YL and l1-YL are led to an encoder (not shown) to become an NT8C signal. This NT
By connecting the SC signal to a display, high-quality images with good color reproducibility can be obtained.

Incidentally, FIG. 4 is a configuration diagram of a main part of an example of the image sensor shown in FIG. 3 fa+. The diagonal line at the top right is the channel stop.
A and B indicate virtual electrode portions, and CSD indicates a portion below the transfer electrode. Reference numeral 117 denotes a separation portion for sequentially separating line information of a part of the memory into two registers 31 and 32, and the separation may be configured using a gate electrode.

2E is a transfer electrode of the memory part 2. 17E is a transfer electrode of the separation section, and 31E and 32E are registers 31.3, respectively.
20 transfer electrodes, CL is a clear gate portion, and CD is a clear drain.

Although one-phase drive is used in this embodiment, a two-phase or more drive method may also be used.
~P (at all times, when a level voltage is applied to each transfer electrode), P fcl > P ff11 > P (A
l > P (B), when low level voltage is applied, PtA)>PtB)>P (C1> P (r)l
It is configured so that

With this configuration, the charges initially at 168 and 171 of the memory portion 2 are transferred to 167 and 170, respectively, by setting the clock hand to a high level, and then when φT becomes low again, the charges are transferred to 166 and 166, respectively. 164 respectively. After that, the clock φ.

When the clock φ is set to high level, the charge of 164 is transferred to 163, and when the clock φ is set to high level, the charge is transferred to 161, and is accumulated in 160 after the fall of the clock φ.

On the other hand, if a pulse is again supplied to the clock φT while keeping the clock φS at a low level, the charge in 166 is transferred to 164. After this, the clock φ.

By supplying one pulse of , the charge of 164 is transferred to 162.

In this way, after the information of one horizontal line in a part of the memory is divided and moved to registers 31 and 32, the information is shifted by one line in the part of the memory again, and the above sequence is repeated. Part of the information can be read out by dividing one horizontal line using two shift registers.Horizontal transfer using registers 31 and 32 may be performed by any appropriate method.

Next, FIGS. 5-) to tc+ and FIG. 7 are diagrams showing a second embodiment of the present invention. Figure 5 shows the G shown in Figure 3+bl.
A W (transparent filter) filter is used for the filter 9, and FIG. 7 shows its signal processing circuit.

As with the one-type filter described above, R, B~ can be separated in exactly the same way. The signal transmitted through the W filter corresponds to the above-mentioned G signal, but since the W filter is a transparent filter, the light passing through it can be treated as a Y signal (luminance signal). As for the Y signal, Yw (wide band Y signal) + YL (low band Y signal) corresponding to GL is obtained for any bag. YL, R, and B are used in the process matrix circuit 25 to generate color difference signals R-yl, , ny, . These color difference signals and YW signals are led to an encoder (not shown) to become an NTSC signal.

Adopting such a filter configuration has the effect of increasing the light utilization rate and increasing the sensitivity.

Next, FIG. 8 is a diagram showing a third embodiment of the present invention. In this embodiment, the outputs of registers 31 and 32 are recorded independently without being subjected to correlation processing. .82 are connected to line switches 100.82 after passing through amplifiers 5.6 and 100.82, respectively. 101.

This line switch 100.101 is a clock circuit 46
Yo! It is switched every Ill/2fH.

Also, 102.105 is a low pass filter, 103.10
5 is a process circuit, 104.107 is a modulation circuit, 108
1 is a mixer, 109 is a recording circuit, 110 is a recording head, 1
11 is a recording medium, 112 is a 2-y switch, and 1
/2 can be switched with fu. 113 is an oscillator that outputs carriers f, , f, , f;;

Further, the head 110 switches the recording track for each IH.

With this configuration, the G signal is recorded IH by IH, and the R signal and B signal are recorded line sequentially. Here, for example, the modulation circuits 104 and 107 are FM circuits,
f is several Mllz, and ft, f; are each several hundred KI (set to z).

With this configuration, the resolution in the vertical direction is slightly reduced, but the band of each color signal is increased by 3 at the Nyquist frequency.
．． Since the color spreads to 5811 tlz, color reproducibility is extremely improved. Furthermore, since vertical and horizontal correlations are not used, false signals do not occur.

(Effects) As explained above, according to the present invention, (1) The sample and hold circuit for color separation can be omitted or simplified.

(2) It is possible to prevent the deterioration of the scene due to high-speed readout of the horizontal shift register, and it is possible to reduce the inaccuracy when separating the color signal from the dot sequential signal.0 (8) The drive frequency of the horizontal shift register can be substantially reduced. can be lowered.

(4) Since only two horizontal shift registers are used, the configuration is simpler than when a horizontal register is provided for each color.

(5) By matching the cycle of the color filter repetition 1 to pattern and the horizontal line signal input to both shift registers, the processing circuit configuration of the image sensor output is simplified, and the color reproducibility is improved. Excellent image quality with even higher resolution and sensitivity can be obtained.

(6) Since a checkered color filter is used, it is possible to increase the frequency of each color, and since signal processing is performed without using vertical correlation, a wideband color signal can be obtained. , excellent color reproduction,
Moreover, false signals are less likely to occur.

(γ) Of course, the use of vertical correlation has many effects, such as the ability to easily form high-band luminance signals and standard color television signals.

[Brief explanation of drawings]

Figure 1 is a configuration diagram of frame transfer type 00D, Figure 2 is R2
A diagram showing an example of a repeating color filter of O and B, FIG. 3(a)
3(b) is a diagram illustrating the configuration of an image sensor suitable for the present invention, FIG. d) is a diagram illustrating the present invention in detail, FIG. 4 is a main part configuration diagram of an example of an image sensor, and FIGS. 5(a) and (b). (C) is a diagram illustrating the second color filter suitable for the present invention and the horizontal line signal distribution method, FIG. 6 is a block diagram of a signal processing circuit according to an embodiment of the present invention, and FIG. A signal processing circuit block diagram of the second embodiment of the invention, FIG. 8 is a signal processing block diagram of the third embodiment of the invention, 5.6 is an amplifier,
7 and 8 are IH delay lines, 9 to 13 are line switches, 14 is an adder, 15 to 18 are low pass filters,
19. 20 is an amplifier, 21 is a subtraction circuit, 22 is an addition circuit, 23 is a process matrix circuit, 24 is a process circuit, 25 is a process matrix circuit, 104, 10
7 is a modulation circuit. Patent Applicant: Canon Co., Ltd. 3rd Wa (b,) 3z 83μ5 (C) n line box T]! "Ding η [■ Tsukasa → 5t (fr + I) line G (x & G52 83 wa (d) n line RRRR52

Claims (1)

[Scope of Claims] An image sensor that converts an optical image into electrical information; A checkered color separation filter that is provided in the optical path of incidence to the sensor and that contains two types of color information for each horizontal line; An imaging device having two horizontal scanning means for dividing and reading out one horizontal line of information of the sensor according to the color of the filter, and a recording means for independently recording the outputs of both horizontal scanning means. .