JPS5839390B2 - charge injection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention is a novel charge injection device that enables so-called primary information processing such as smoothing of original image information, edge detection, and line segment detection in optical character reading and pattern recognition.
This invention relates to an Injection Device (hereinafter abbreviated as CD) and its driving method.

As shown in FIG. 1, CID, which has recently been widely used as a two-dimensional imaging device, has a light-receiving surface 1 made of, for example, a P-type semiconductor as a substrate, and X and Y shift registers 6.7 for addressing a predetermined row or column. and an imaging signal detection circuit 8.

One section 2 surrounded by a small dotted line in the figure corresponds to one pixel on the light receiving surface, and is separated from a pair of insulated gate electrodes 3a and 3b.

The pixels 2 are arranged in a matrix, one insulated gate electrode 3a of each pixel is connected in parallel to the bus bar 4, and the other insulated gate electrode 3b is similarly connected to the bus bar 5 in parallel.

When an image is formed on the light-receiving surface 1, charges generated by photoelectric conversion are accumulated directly under one of the pair of electrodes, for example, the electrode 3b.

Each pixel in this state is X. By specifying an address using the Y shift register 6.7, the accumulated charge of the specified pixel will be transferred to the other electrode 3.
Since the pixel is detected at point a, the image signals corresponding to the image formed on the light-receiving surface are read out in time series via the generatrix 4 and the detection device 8 by sequentially specifying the address of the pixel by performing constant XY scanning.

On the other hand, image pattern recognition methods involve relatively simple so-called primary processing such as smoothing of original image information, edge detection at black-and-white boundaries, and line segment detection, as well as pre-memorized character and Comparison of figures etc. with the information after the primary processing,
Image processing can be roughly divided into so-called secondary processing, which is more complex and advanced, such as judgment, and is performed by an electronic computer that works with the imaging system.

For this reason, conventional image signal processing first uses the above-mentioned CID.
Each pixel on the surface is scanned in an XY constant manner, and all image information on the light-receiving surface 1 is stored in a storage device (not shown).Then, an electronic computer performs the above-mentioned primary processing and secondary processing to create an image. Since the system involved transmitting signals, problems such as a decrease in the processing speed of the imaging system as a whole and an increase in the storage capacity of the electronic computer became a problem.

Originally, OID has features such as not only having a non-destructive readout function of an imaging signal but also being capable of high-speed operation, so it is conceivable to perform the above-mentioned primary processing by utilizing the features of OID.

If it were possible to perform primary processing using this OID, the electronic computer would only have to perform secondary processing, which would immediately result in a significant reduction in required storage capacity and an improvement in processing speed.

However, in order to execute this primary processing, the OID must have a function for randomly reading a desired number of pixels at a desired address. Since the above-mentioned random readout is not possible with the conventional OID using the driving method described above, there is a disadvantage that primary processing cannot be performed with the structure as it is.

By the way, the present inventor previously disclosed in Japanese Patent Application No. 54-30732 the above-mentioned two electrodes 3a in each pixel of OID.

We have proposed a pixel structure suitable for random readout of pixel signals by shielding electrode 3b from light and providing a third electrode 3c, that is, a photosensitive electrode, which has a light-transmitting window directly above it and is dedicated to photoelectric conversion.

The present invention utilizes OID with such a configuration to solve the above-mentioned conventional problems.The present invention uses the pixel group with the above-mentioned three-electrode configuration as a light-receiving surface, and uses the conventional method to enable random pixel address designation. It is possible to replace the X and Y shift registers with decoders arranged in each of the X and Y directions, and to assign a predetermined weighting coefficient to each signal group from a plurality of pixels consisting of one group or two groups. The present invention provides a novel OID equipped with a system signal detector, a differential amplifier for inverting the polarity of a group of signals from one of the detectors, and an arithmetic unit for performing calculations between the respective signal groups. This will be described in detail using the drawings from FIG. 2 onwards.

FIG. 2 schematically shows the OID according to the present invention.
The same parts as in FIG. 1 are shown with the same symbols.

The structure and operation of each pixel 2 constituting the light-receiving surface was previously described in Japanese Patent Application No. 54-30732, but a brief explanation is as follows.

In each pixel 2 there are first and second insulated electrodes 3a, 3.
In addition to b, a translucent third insulated electrode 3c is provided, and the third electrode 3c in each pixel is electrically connected by a common bus bar (not shown), and a pulse voltage is applied from a terminal indicated by 10. It is now applied.

The upper part of the light-receiving surface is covered with a light-shielding film (also not shown) having a group of light-transmitting windows (not shown) only directly above the third electrode 3c.

While the voltage applied to the terminal 10 is of positive polarity, for example, 8V, a potential well (hereinafter simply referred to as a well) is generated directly below the electrode 3c.

Electric charges are generated in the well by the incident imaging light from the light-transmitting window, and when the voltage at the terminal 10 returns to zero, the well disappears. The electrode 3a is transferred inside and temporarily accumulated.
It is detected by the electrode 3a when it is moved into the well directly below it.

Therefore, in the following, each electrode 3 c t 3 b t 3
a will be referred to as a photosensitive electrode, a storage electrode, and a readout electrode, respectively.

The well group directly below each electrode 3b is the terminal Y of the address designation circuit 7a.

~ Y4 is generated by the voltage applied in advance to the electrodes. To create a well group directly under each electrode 3a, close the plurality of switches 9 together with the switch SW1 or SW2 and apply a positive voltage of, for example, 8V from the power source 31. A voltage may be applied to the group of electrodes 3a.

Since the well group formed directly under the electrode 3a is maintained for several tens of milliseconds even after the switch group is opened, the processes of exposure, storage, and readout can be performed within a shorter time than this. good.

Reference numerals 6a and 7a designate address designating circuits provided in place of the shift registers in FIG. 1, which are comprised of a decoder and a driver.

The circuit 6a, ? The terminal that a has, that is, Xo 2
XH, X2 t X2 tX4...and Y
.

, Y, , Y2, Y3. Select a plurality of terminals from among Y, . The pixel signal charges consisting of the two groups can be read out separately in time series onto the signal charge detection line 20 via the switches 5o-S4.

Then, the two groups of signal charges are transferred to the charge amplifier AI by selectively opening and closing the signal path introduction changeover switches sw, , sw2.
.

A2 separately as a time series, and sequentially V, to each output terminal of the double-width divider, that is, the point P1tP2.

Each output voltage, appearing as v2, is input to a differential amplifier A3, which inverts the polarity of said voltage V2, making it the opposite polarity of voltage V,.

and switch SW3, capacitor C33, and amplifier A4
It is also possible to perform a subtraction operation between the signal from the first pixel block and the signal from the second pixel block using a procedure described later.

In FIG. 2, in which 4×4 pixels are shown as the essential parts of the light-receiving surface, the operation in the case where 3×3 pixels among them are used as a basic block will be described below.

Assume that the feedback capacitance connecting the input and output terminals of each charge amplifier A, A2 mentioned above is Off, and the capacitance spanning the input terminals is Cin.
Then, the gain of the amplifiers A, , A2 is O1n10
It is determined by f.

Here, the first feedback capacitance shown as C31°C21 in FIG. 2 has the same value, and the second feedback capacitance C1□ and 0
It is assumed that 2□ is selected to be eight times the first feedback capacitance 0112012, that is, it is weighted.

For example, detection of a signal charge from one pixel Cx, 1) designated by address designating circuits 7a and 6a in FIG. 2 is performed as follows.

First, the switches SW2 and SW12 are opened and the switch S is opened by the voltage from the address designating circuit 6a.

Close.

At the same time, the switch 11 is closed, and by this action, a capacitor C11 having a value of 1/8 of CI2 is connected to the charge amplifier.

Therefore, the charge of the (1,1) pixel is detected by the charge amplifier A having a gain of 0 in 1011.

On the other hand, 0,0)t(0,1)t(o,2).

(1,0), (1,1), (1,2), (2,0), (
2,1).

When the charges of the 9 pixels in (2, 2) are sequentially read out, integrated, and then averaged, assuming that the amount of incident light on the light receiving surface is uniform, the signal charge will be 9q in total, so the apparent appearance will be higher. An output signal 9 times as large will be output from the charge amplifier A.

In order to avoid this, by closing switches 11 and 12, 011 and a capacitor 012 having a value eight times that value are simultaneously connected to the charge amplifier A, and the gain of the amplifier A is reduced to 1/9 in advance. First, apply weights.

In this way, a balance can be achieved between the accumulation average of charges for nine pixels and the detection of only one pixel.

In addition, when reading out charges for 8 of the 9 pixels, open switch 11 and close switch 12, connect only 01□, which has a capacitance 8 times that of 011, to charge amplifier A, and set the gain to Oin/80. ．． , it is sufficient to set it as .

Note that R,, Tt2 is the capacity 011.021 j C32
A reset switch for discharging the residual charge in the amplifier A3. It is closed before performing A2, and left open at the time of detection.

Based on the above-mentioned matters, the OID according to the present invention
The primary processing operation will be explained below.

1) Smoothing Process Smoothing process is effective for erasing black spots on an object to be read, such as a document surface with a white background, compressing the band of an image signal, or roughly recognizing figures on a document surface.

Although either of the charge amplifiers A, A2 may be used to detect the charge of the pixel, here, the amplifier A1 is used.

First, a signal charge is generated by photoelectric conversion in a well made directly under the photosensitive electrode 3a, and then the terminal X of the address designation circuit 6a described above is generated.

-X out of X4. Switch S by the voltage from the terminal.

open. Next, when the reset switch R3 of the amplifier A1 is closed and then opened again, the voltage VR of the power supply 31 is transferred to the input terminal of the amplifier A, the switch SW1. Pixel (0,0) through So
The voltage is applied to the photosensitive electrode 3a in the photosensitive electrode 3a, and a well is formed directly below the electrode as described above, and the standby state for reading out the charges in the well is completed.

Next is the terminal Y of the address designation circuit 7a. -Y. If only the voltage at the Yo terminal is made zero among the voltages that appeared in each of the voltages, the well directly below the storage electrode 2b disappears, and the charge stored here moves to the well directly below the adjacent readout electrode 3a. is detected by the electrode 3a.

This detected charge is applied to switch S.

, SW, and then to the differential amplifier A3, and appears as a voltage v3 at point P3.

If the switch SW3 is turned to the point P3 side, the voltage is temporarily stored in the capacitor 013 as a charge.

A is an operational amplifier, but since it has a feedback capacitance of 014, G40 is connected to the negative input terminal of the amplifier, that is, the point P4.
This is equivalent to having a capacity of 14.

Here 04 is the open loop gain of the amplifier.

Therefore, it can be considered that when the switch SW3 is turned down to the point P4, the charge in the capacitor 013 is transferred to and stored in the capacitor G4.014.

Next, specify the address of the pixel (0゜l) adjacent to the previous pixel (010), and use the same procedure as above to first address the pixel (0,0).
) is accumulated in the capacitor G4·014 which had been accumulated.

Similarly, nine pixels (0, 2), (ItO), (
If addresses 1, IL, etc. are specified in sequence, and each time the output of the pixel is stored in the capacitor G4.014,
At the time when the signal charge of the final pixel (212) has been detected, the integration of all the charges for the nine pixels is completed.

However, since each of these charges was read out by the charge amplifier A whose gain was reduced to 179 in advance, the procedure for calculating the average value QM of the charges for 9 pixels as described above,
In other words, the following arithmetic operation has been executed.

However, i. j is a subsign that means the i row and the j row, respectively.

When the smoothing process for the above 3x3 pixel block is finished and the result is output from the final terminal, point P5, the next pixel group, for example (3, o ) t (3, IL (a, 2
) and the next 3x3 pixel block consisting of the top row and the other 6 pixels (not shown), the same smoothing process as above can be applied to the original image of another part (block) on the form. It can be carried out.

In this case, when the smoothing process for one block is completed and the smoothing process for the next block is to be started, the reset switch R3 of the amplifier A4 must be closed to first discharge (reset) all the charges accumulated in 04014. No.

Note that the power supply 32 connected to the positive input terminal of the amplifier A4 is
This is to remove the DC component contained in the input signal to the amplifier A4, and is added so that the DC voltage and the value are equal and opposite in polarity, so that unnecessary DC components are removed from the output of the amplifier A4. Only signal components containing no components will be obtained.

Further, the signal output appearing at the final output terminal P of the amplifier A4 can be taken out as a desired analog signal by connecting a desired smoothing circuit to the terminal P and passing it therethrough.

1i) End Detection End detection is to detect, as a line image, the end A of a pattern consisting of a diagonally lined black area on a form and a white area to the right of it, as shown in FIG. 3a, for example. .

To detect edges like Otsu-9, in other words, black and white boundaries,
First read out the signal charge of the central pixel (l, 1),
Next, the 8 pixels surrounding this (0,0), (0,1)t
(0,2)t(LO), (1,2), (2,O), (2
,1).

After reading and smoothing the total signal charge of (2t 2 ), finding the average charge QE determined by the difference between the two;
That is, it can be performed by the following calculation.

Operations such as pixel address designation and averaging of multiple pixel signals have already been described, so the detection processing procedure will be explained below.

First, address the center pixel in FIG. 2, for example (1, 1), close the switch 11 while keeping the switch 12 of the charge amplifier A1 open, and connect the capacitor 011 to the amplifier A to set its gain to 06n1011. do.

Then, the switch SW is closed and the signal charge of the pixel (l, ■) is guided to the charge amplifier A.

The pixel signal appearing as a voltage V at the output terminal, point P, is applied to the positive input terminal of the differential amplifier A3, and appears as a voltage V3 at a point P3.

If the switch SW3 is turned to the point P3 side, the voltage V3 charges the capacitor C13, but this charging charge is the result of integrating the signal time series introduced into the positive input terminal of the differential amplifier A3, so for convenience Its polarity is considered positive.

This positive polarity charge is transferred to the charge amplifier A4 when the switch SW3 is turned down to the point P4, and is stored in the equal input capacitances G, C, . . . of the amplifier A.

Next, the switch 22 of the charge amplifier A2 is closed while the switch 21 of the charge amplifier A2 remains open, and a capacitor 022 whose value is eight times that of 021 is connected to the amplifier A2, and its gain is reduced to C! 1n1022, that is, the gain of amplifier A is 178 Fct.

Then open switch SW2 and switch SW2 instead.
(t t i) The eight pixels around the pixel are sequentially addressed, and the charge of the eight pixels is sequentially transferred to the charge amplifier A2.
If the voltage is input to the output terminal of the amplifier A2, the voltage v at the point P2.
The signals for the 8 pixels appearing as 2 are applied to the negative input terminal of the differential amplifier A3.

Therefore, the signal appearing at point P3 is
The voltage that appears on the switch SW is output with the opposite polarity.
3 is again moved to the point P3 side, negative charges for eight pixels having negative polarity are accumulated in the capacitor CI3.

Here, when the switch SW3 is turned down again to point P4, the positive charge previously accumulated in the capacitor 04014 is added to the above 8
The subtraction operation is completed at this point because the negative charge for the pixel is added.

Therefore, in order to detect this subtraction result, it is sufficient to read the output voltage appearing at point P5.

By using the above operations, the above-mentioned end detection can be performed as follows.

First, if the end is located at the dotted line ■ in Figure 3a, the pixel (+, 1
), all nine pixels (0°O) to (2,2) receive the white signal of the form surface.

Below, this will be expressed as being at the white level, but for convenience, each pixel (0,0)? (Oyl), (0,2)...
...The charges generated by (2, 2) are each q.

O, (101+qO2t...q2□), and the charge at the white level is assumed to be 10q.

The time when the pixel is receiving light from the black area on the form surface is expressed as being at the black level, and the charge in that case is assumed to be zero.

Thus, the above charge q. o''Q2□ all generate a charge of 10q, so QE on the left side of equation (2) becomes zero.

Next, when the end is located at the solid line ■ in Figure 3, (0,0
) , (Ozl)t(1,0L(1ylL(2,0),
The six pixels (2jl) are all at the white level and generate a charge of 10q, but the other pixels (0,2), (l,2), (
2, 2) is at the black level and the generated charge is zero.

Therefore, QE in equation (2) is +3.75.

Furthermore, when the end is located on the dotted line ■, it is (0,0).

Three pixels (l, 0) and (2, 0) are at the white level and 1
A charge of 0q is generated, but other pixels (0°1), (Q,
2), (1,1)j(1,2)j(2,1),(2F2
) are all at black level.

The generated charge is zero, and QE in equation (2) is -3,75.

Finally, if the end is located on the dotted line ■, (op o ) ~ (
Since all the pixels in 2, 2) are at the black level and all the generated charges are zero, Qo in equation (2) returns to zero again.

Among the above cases, if the ends are in ■ and ■,
QE of +3.75 and -3.75 have opposite signs, but the polarity of the output of the signal is, for example, at point P in Figure 2.
It can be made positive by squaring it in a circuit (not shown) connected to 5 and onwards.

Since that value is maintained during the period of movement from the end part (2) to (2), this situation results in a signal output having a width of one pixel as shown in FIG.

However, X and y in FIG. 3 ayb indicate the horizontal and vertical directions of the form surface, respectively.

Once the signal detection of the form surface read out in the above-mentioned 9 pixel block is completed, the adjacent signal (3tOL(3, IL(
3,2)...A block consisting of another nine pixels consisting of (5,2) is designated anew.

When the same operation is completed, addresses of other nine pixel blocks adjacent in the vertical direction are specified.

If the same operation is carried out thereafter, the signals in Figure 3 (S) where the end (A) in Figure 3A occurs will be aligned in the vertical direction - a straight line, so the end will appear as a line image drawn in the vertical direction. Can be detected.

111) Line segment detection Line segment detection is effective when it is desired to extract only line segments in a specific direction.

The simplest example is a case where it is desired to extract only the horizontally drawn line part without detecting the vertically drawn line part in the addition symbol 10, so this case will be explained below.

First, for example, pixels arranged in the horizontal direction (1, 0L (Ll), (1
, 2), the addresses of nine pixels arranged in the pixel arrangement direction, that is, (1,0) to (1,8) are specified.

Then, the signal charges of the nine pixels are subjected to the same procedure as described above in the smoothing process, that is, the averaging shown in equation (1).

To do this, switch S is closed by the voltage from terminal X of address designating circuit 6a in FIG. 2, and terminal Y of the terminal group of address designation circuit 1a is closed.

- Set all output voltages from Y8 to zero.

On the other hand, a capacitor 012 is connected to amplifier A, and the gain is set to C1.
Set to VC42, that is, Cin/80H, and switch S.
The charges for the nine pixels are finally integrated into the equivalent input capacitance 040H of the amplifier A4 via W, , amplifier A, and A3 to complete the averaging.

Detection of the original image line segment drawn on the form can be performed as follows using the above procedure.

For convenience of explanation, it is assumed that the form surface is a black background, and a ten symbol made up of white drawn lines each having a width equal to one pixel in both the vertical and horizontal directions is written on it.

The shaded area in FIG. 4a indicates a black background, and the code entry 1. Considering the case where a pixel row composed of (lt o ) to (1,8) moves in the X direction on the horizontal line of the white ten symbol formed by each part of When the end A of the horizontal line is located at the dotted line ■, all pixels are at the black level and their output signal charge is zero, so (1)
The average charge QM given by the formula is zero.

When the end A comes to the position indicated by the dotted line ■, the pixel (it s )
9 changes from the black level to the white level, and the amount of charge for that one pixel (9q for convenience) is detected, but the output charges of the other 8 pixels are all zero, so the charge for the one pixel 9
q is averaged over 9 pixels, and the average charge Q in equation (1)
M is only 1/9.

However, if the above end A is located on the dotted line ■, the pixel (1t7) in addition to the (1,8) pixel will output a charge of 9q, so QM will be 2/9 and when the end A reaches the dotted line ■ The pixel (L6) not shown also outputs a charge of 9q, so that the average charge QM becomes 3/9.

In this way, the average charge QM gradually increases in value. Here, assuming that the length e of the pixel row is 9 times the length of the horizontal line, that is, 81 pixels, the end portion i is When reaching above the dotted line ■, that is, when all nine pixels are located at the white level, the value of QM becomes 9/9=1.

In other words, while the drawn line moves 81 times pixel by pixel in the direction of the pixel column and all the pixels are white, the average charge output of QM=1 is calculated as follows: The position is maintained until the end position Q reaches the position indicated by the dotted line ■.

When the end portion Q is located at the dotted line ■, only the pixel (l, 8) out of the total nine pixels changes from the white level to the black level, so the output charge of the pixel (1, 8) becomes zero.

However, the other eight pixels are still at the white level and each outputs a charge of 9q.

Therefore, the average charge QM decreases to 8/9 again, and when the end portion Q is located at the dotted line ■, 0M decreases further to 7.
/9.

Thereafter, as the end portion Q moves to the right, that is, in the X direction, the average charge QM gradually decreases, and finally, when the end portion Q reaches the dotted line ■, QM becomes zero.

If the change in the average charge QM due to the above process is plotted, it will be trapezoidal as shown by the solid line in Figure 4.

As mentioned above, it is assumed that the length of the horizontal line is selected to be 9 times the length e of the pixel row, so the length eb of the flat part of the curve in Figure 4 drawn by the above average charge QM is the pixel length. The length of the column is equal to e.

This is because each pixel in the pixel column outputs the same charge of 9q 81 times while the drawn line moves in the X direction.

For reference, if the drawn line length and pixel column length are equal, there is only one chance for nine pixels to output the same charge of 9q, so the change in 6M in this case is shown by the dotted line in Figure 4. It becomes an isosceles triangular shape with one oblique side, and there is no flat part.

Incidentally, the lengths ec and ea in the X direction of the hypotenuse portions of the QM, that is, the gradually increasing portions and the gradually decreasing portions in FIG. 4, are both equal to the length e of 9 pixels.

Furthermore, in the above case, the pixels (o t 1 ) located above and below the pixel (1, 1) in FIG.
) and (2,l) are not addressed.

Therefore, even if the vertical drawn line part 2 and the part E of the above-mentioned cross symbol are located on both pixels, and both of these pixels output a charge 9q, the charge will not be detected, and therefore nothing will be added to the above-mentioned average charge QM. It will not contribute.

On the other hand, the pixel rows (0,0) to (0,8), not shown, including the above pixel rows (o, i) arranged horizontally directly above the pixel rows (LO) to (1°8), are set as addresses. If you specify it again, the above two parts of the vertically drawn lines of the ten symbol will be detected.

Then, pixel rows (2,0) to (2,8), not shown, including the pixel rows (2,l), which are also arranged in the horizontal direction directly below the pixel rows (1t o ) to (i,s), are similarly addressed. Even when the image is corrected, the portion indicated by E among the vertically drawn lines of the + sign is detected in the same way.

However, in each of these cases, even if the vertical drawn line portions 2 and 5 move in the X direction, the detected charge is only the charge 9q for one pixel, and this is the same as the other 8 pixels at the black level and the white level. Since the average charge is calculated for a total of 9 pixels including 1 pixel in , the resulting average charge QM never takes a value of 179 or more.

In other words, the average charge QM obtained from each part 2 or E of the vertically drawn line is approximately 10% of the average charge obtained from the horizontally drawn line.
It is.

Here, we set a technique often used in signal processing, that is, a slice level. For example, at the level indicated by the dashed-dotted line in Figure 4, which corresponds to the charge of 7 or 8 pixels, the second
If the output voltage at point P in the figure is sliced, the vertically drawn line portions 2 and 5 will be ignored and only the horizontally drawn line will be extracted.

Moreover, the gradually increasing part ea and the gradually decreasing part ec of the curve in Figure 4 cause long blurred parts of the image at both ends of the horizontal line in the longitudinal direction. Portions near the ends A and C of the drawn line can be reproduced as clear images.

Above, we have described an example using a 3x3 image as the basic number of pixels, but the processing performed using this CID is not limited to 3x3, but can be expanded to more or less. This can also be done using n×m pixels as the basic number of pixels.

In that case, the two systems of detection charge amplifiers AHI A
It is necessary to set a value for the feedback capacitance of 2 according to the given expansion case, in other words, to increase the weighting during detection. 11
, 12 or 21°22 and the number of switches 0
11 j 012 or 021. This can be easily done by appropriately setting the number of capacitances C2□ or their values.

Furthermore, in the explanation of Fig. 4, we took as an example the case where multiple pixels are specified in one row horizontally, but in contrast, it is also possible to specify multiple pixels in one column vertically, or even in a diagonal direction. This is also possible by operating the address designation circuits 6a and 7a.

For example, if you specify the address of one vertical column of pixels,
The distribution of multiple ten symbols and only one symbol is effective in knowing where the ten symbols are scattered on the form surface where the ten symbols are written.

To do this, it is sufficient to ignore the horizontal drawn lines of the ten symbol and at the same time extract only the vertical drawn lines to determine whether the symbol is a ten symbol or one symbol.

Similar determination is performed by recognizing the alphabet V, X, or W by specifying the diagonal address of a plurality of pixels once or several times and reading out the signal charges.

According to the OID and its driving method according to the present invention described above, the imaging system It is possible to improve the overall processing speed and reduce the required storage capacity of the interlocking computer, and is expected to have significant practical effects.

[Brief explanation of drawings]

Fig. 1 is a schematic diagram showing the structure of a conventional OID, Fig. 2 is a schematic diagram showing the structure of a new OID according to the present invention, and Fig. 3 a
) b is the pattern edge detection means on the form surface using CID in FIG. 2, and FIG. 4 a) and b are the line segment detection means on the form surface using 0IDi/i: in FIG. It is a figure for explaining. 1: Light receiving surface, 2: Unit pixel, 3at3bt3c: Accumulation,
Each electrode for detection and photosensitivity, 4, 5: bus bar, 6at7a
: address designation circuit, 10: voltage application terminal to electrode 3C group,
11.12, 21.22. SW. SW2 t SWs pso=s4 t R1t R2
t R3: Switch, 20: Detection line, 31, 32: DC power supply, 011 j C12t 013 y 014
: Capacity, (o t o ) ~ (3 3) 2 pixel number, e
: Pixel row length, A, C: End of the original image, B, N, G: Part of the horizontal drawn line of the original image, 2, E: Part of the vertical drawn line of the original image,
x, y: horizontal and vertical directions.

Claims (1)

[Claims] 1. In a charge injection device mainly consisting of a light-receiving surface in which unit pixels each having a photosensitive electrode, a storage electrode, and a detection electrode are arranged in a matrix, and a horizontal and vertical address specifying means for specifying an image at a desired address, The detection electrode is connected to a signal charge detection line via an address designation switch, and the detection line is connected to two series of charge amplifiers via a signal path introduction changeover switch. A charger comprising a differential amplifier in which a plurality of feedback capacitors that can be shortened by a reset switch are connected to each of the capacitors via a switch in series, and to which the outputs of the two series of charge amplifiers are input. Injection device.