JPS61281673A - Image sensor output normalizing circuit - Google Patents

Abstract

PURPOSE:To make normalization automatic and to facilitate the replacement of the sensor by providing the highest output memory circuit, a ROM to output the correcting value based upon the highest output value and the output of respective elements and a RAM for storing the correcting value. CONSTITUTION:In the correcting data writing mode, the output So(i) of the sensor to the standard light enters the B input of a comparator 4 of the highest output memory circuit 3, and the output of a latch circuit 3 enters an A input. When the comparing result is A<B, an So(i), which comes to be the latch input by a clock LCK, comes to be the output of a new latch. In case of A>=B, the latch 3 does not act. The output of the latch 3 is compared with the next sensor output by a comparator 4. Thus, when the output of all elements are read, the output of the latch 3 comes to be the maximum value Smax of the So(i). On the other hand, to the higher order and the lower order addresses of the ROM 5, the Smax and the S(i)(So(i)) are respectively inputted, and from the ROM 5, a correcting value F(i) corresponding to respective elements is outputted. During the correcting mode, for a writing clock WCK, the F(i) is written to a RAM 6 corresponding to the address of an action element. The S(i) and F(i) are multiplied by a multiplier 7 and the normalized output can be obtained.

Description

[Detailed Description of the Invention] [1] Industrial Application Field Optical character reading that receives reflected light from an object using an image sensor in which a plurality of photoelectric conversion element groups are arranged, and outputs an image signal depending on the amount of light. It can be widely used in fields such as devices and image traffic flow measurement devices. By normalizing the output of the image sensor using the circuit of the present invention, there will be no variation in the output, and the original image can be faithfully reproduced.

[2] Conventional technology The use of image scanners has expanded rapidly in recent years, and the above-mentioned devices for optically reading characters, etc. are used for data input in so-called OA systems. Storage and facsimile transmission are becoming popular. For information on the method of inputting graphics into a computer using an image scanner and processing such as editing, demand for equipment, and trends in technology, see Nikkei Electronics, September 10, 1984, edition 1.
It is introduced on pages 61 to 191. That is, it is well known that the level of the video signal for one period of main scanning of the image sensor is distorted, causing shading. One of the causes of this is the variation in the output of each pixel of the sensor, which is due to the inherent variation in the sensitivity of each element.

This state is shown in FIG.

The electrical output from each element is approximately proportional to the amount of light received, and the proportionality constant is the sensitivity of each element, so if there is variation in sensitivity, even for a constant amount of light received, the Variations occur in the output, making it difficult to faithfully reproduce the original image.

For this reason, it is necessary to add correction to the output to compensate for variations in sensitivity to obtain a normalized output as shown in FIG.

FIG. 5 shows a conventional example of a normalization circuit. In this example, the correction value F(i) corresponding to each element of the image sensor (52) is stored in advance in the correction value memory C58) (for example, RO
The normalized output SNRM(i) is obtained by storing M=ReadOnlyMeuory) and multiplying this by the output S(i).

SNRM(i) = F(i) −5(i)
(1) F (i) is expressed using the output 5o(i) for a certain reference light amount yok. In order to improve accuracy, the reference light amount is the maximum amount of light that can actually be received, for example, in the case of OCR, the amount of light reflected on the surface of a blank document is used. Therefore, the output 5 at this time
o(i) is the maximum output of each element during operation.

Here, in order to maximize the dynamic range, the value of K is set as follows. The maximum value that can be taken as a correction value is Fr
Nax, 5o(i), if the maximum value (output of the element with the highest sensitivity relative to the reference light) is S max, then K = Fmax - Smax
(3) The formula (
It is necessary to store each correction value represented by 4), but conventionally Smax was calculated from 5o(i) as follows.

(2) Read the output 5o(i) of each element with respect to standard light using a synchroscope, etc., and find the maximum value Smax among them.

(2) Load the output 5o(i) of each element with respect to standard light into a computer and find the maximum value Smax on the computer using a program.

[3] Problems with the prior art When determining Smax as described above, the method (2) relies on manual reading of the output, so in terms of time and reliability of readings, the number of elements exceeds several hundred. cannot be applied to image sensors. In addition, method (■) uses a computer and is superior to (■) in terms of reliability and time required, but
In addition to the circuit shown in Figure 5, a circuit is required to read the sensor output into the computer, which is not only disadvantageous in terms of cost, but also has the disadvantage of not being able to replace the sensor in the field since a computer is always required. There is a flaw. (The variation in ζ is different for each sensor.) [4] Means for solving the problem FIG. 1 shows the configuration of the present invention.

The operating modes of the present invention are divided into (i) correction data writing mode and (11) normal operating mode.

(1) Correction data write mode CALt7 is true during this period. Image sensor (2)
Let standard light be incident on , and let its output be 5o(i).

5o(i) is the middle comparator (4) (c
input B of the comparator). On the other hand, the output of the latch circuit (3) is input to the comparator (4), and the initial value of the output is O. When the comparison result is A<B, the latch circuit is activated. The clock LCK input to the latch circuit (3) enters the latch circuit (5o(i)), which is the latch input.
becomes the output of the new latch.

If the comparison result is A2B, the latch does not operate and the output of the latch circuit (3) remains unchanged. Latch circuit (
The output of 3) is compared with the next sensor output as the output of comparator (4).

In this way, when the outputs from all the elements of the sensor are read out once, the output of the latch circuit (3) is 5o(i)
is the maximum value Smax. After 2 days Sm
Since 5o(i) (the same B input) exceeding ax (the six inputs of the comparator) never appears, the output of the latch circuit (3) does not change.

On the other hand, the upper addresses (A13 to A7) of ROM (5) are set to Smax, and the lower addresses (A6 to AO) are set to Smax.
(i) (So(i)) is input. This ROM (5)
By writing the results for each address in advance as data, a correction value F (i) corresponding to ζ can be obtained for each element.

F (i) is an input to RAM (6), but since the write clock WCK for RAM is operating during this mode, F (i) is input to RAM (6) in accordance with the element address. written.

(11) Normal operation mode In this mode, CAL becomes a question and the contents of the latch circuit (3) and RAM (6) do not change.

The RAM is read-only, and the sensor output S depends on the amount of light.
(i) and the RAM output F (i) are multiplied by a multiplier (7) to obtain a normalized output SNRM(i).

[5] In the example shown in Fig. 1, all data is 7 bits, but this is mainly due to restrictions on the number of address bits in ROM, and here we will use 128KRO, which has recently begun to appear on the market.
M (address 14 bits, data 8 bits) is used. Currently mainstream 64 or 32 RO
M ((61) ,, (62) or (71)(7
2), (73), (74)), the data will be 6 bits, but sufficient accuracy can be obtained for uses such as OCR. (Resolution is 96=64) However, 64
Figures 6 and 7 can also be obtained using 3jlROM.
It is also possible to create a 7-bit data circuit by adding a decoder circuit (75) as shown in FIG.

Also, in the configuration of the present invention shown in FIG. 1, the comparator (4) has six inputs.

B input or ROM (5) Smax and S (i)
The allocation of addresses (A13 to A7) is not particularly fixed. In addition, in order to prevent the memory from being lost even if the circuit power is turned off once the correction data has been written, it is necessary to add a backup circuit using a battery to the latch circuit (3) and RAM (6), or It is sufficient to use an IC with a built-in backup function.

[61 Effects] The image sensor output normalization circuit of the present invention solves the problem of the prior art described in item [8], and its remarkable effects are as follows.

(b) By adding a simple circuit, normalization can be performed automatically without relying on human labor.

(b) Image sensors can be replaced and output correction values for the image sensors can be easily rewritten even in the field without the need for a computer.

(c) An optimal dynamic range can be obtained by automatically detecting Smax for standard light and calculating a correction coefficient using that value.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the configuration of the present invention. FIG. 2 is a diagram showing variations in sensitivity of each element of the image sensor. Figure 3 is a diagram showing the output state of each element of the image sensor without output correction. (a) shows the maximum amount of light received ("white").
], (b) corresponds to the intermediate light intensity, and the horizontal axis corresponds to the "black" case where no light is received (cont・d). C) K corresponding unnormalized outputs of each element (aγ, (bχ and (
Indicates Cγ. FIG. 5 is a diagram showing a conventional image sensor output normalization circuit. FIG. 6 shows a 7-bit output normalization circuit using 64KROM. FIG. 7 shows a 7-bit output normalization circuit using 32KROM. l Element address counter 2 Image sensor 3 3 Latch circuit 4 Comparator 5 ROM 51 Read address counter 52 Image sensor 53 Correction value storage 54 Multiplier RAM lROM 2ROM 7 Multiplier lROM 72 〃 78 〃 74 〃 8 Maximum output storage circuit 1. Figure 2. Figure 3. Figure 4. Figure 6.Procedure Correction Good March 72, 1985 1゜Indication of Incident 1985 Patent Application No. 123794 2 Title of Invention Image Sensor Output Normalization Circuit & Relationship with the case of the person making the amendment Patent applicant location 5-15 Kitahama, Higashi-ku, Osaka Name (
213) Sumitomo Electric Industries, Ltd. President Kawakami
Tetsu Department ↓Agent Address Sumitomo Electric Industries, Ltd., 1-1-3 Shimaya, Konohana-ku, Osaka (Telephone: Osaka 461-1031) I will submit it accordingly. Full text correction゛Description 1, Title of the invention Image sensor output normalization circuit 2, Claims (1) For normalizing the output of an image sensor in which photoelectric conversion elements that output electrical signals according to the amount of received light are arranged The circuit includes a maximum output storage circuit (8) consisting of a comparison circuit (4) and a latch circuit (3), and a latch circuit (3).
The maximum output value of the output from each element and the output from each element for the standard light amount are input as addresses to the address counter (1), and the ROM is written in advance so that the correction value uniquely determined from both is set as the output. (5) G, 20M (5
) for storing the output of the image sensor (2) in a position corresponding to the address of the element of the image sensor (2) and reading it out according to the address;
output and the correction value (Fi
) A multiplier (7) for multiplying by . 3. Detailed Description of the Invention (13 Industrial Application Fields An optical character reading device that receives reflected light from an object using an image sensor in which a plurality of photoelectric conversion element groups are arranged, and outputs an image signal according to the amount of light. It can be widely used in fields such as image traffic flow measuring devices. By normalizing the output of the image sensor using the circuit of the present invention, there will be no variation in the output, and the original image can be faithfully reproduced. 2] Conventional technology The use of image scanners has been rapidly expanding in recent years, and in the same way that the devices that optically read characters and the like mentioned above are used for data input in so-called OA systems, figures can be stored as images on disks. Also, facsimile transmission is beginning to become popular. For information on the method of inputting graphics into a computer using an image scanner and editing and other processing, as well as the demand for equipment and trends in technology, see Nikkei Electronics 784, for example. September 10th edition 1
It is introduced on page 61 = page 194. Generally, the output of an image sensor is accompanied by non-uniformity called shading, and some kind of correction for this is required. One of the causes of shading is that each element of the sensor has its own variations in sensitivity. This state is shown in FIG. The electrical output from each element is approximately proportional to the amount of light received, and the proportionality constant is the sensitivity of each element, so if there is variation in sensitivity, even for a constant amount of light received, the Variations occur in the output, making it difficult to faithfully reproduce the original image. For this reason, it is necessary to add correction to the output to compensate for variations in sensitivity to obtain a normalized output as shown in FIG. FIG. 5 shows a conventional example of a normalization circuit. In this example, each element of the image sensor (52) stores the corresponding correction value F(i) in advance in the correction value storage (53) (for example, ROM=
The normalized output SNuM(j) is obtained by multiplying this by the output S (i). SNRM(') = F(i) -5(i)
(1) F (i) is calculated as F (i) =
(2) Represented by So(i). In order to improve accuracy, the maximum amount of light that can actually be received, for example, the amount of light reflected on the surface of a blank document, is used as the reference amount of light. Therefore, the output 5o(i) at this time becomes the maximum output during the operation of each element. Here, in order to maximize the dynamic range, the value of K is as follows. The maximum value that can be taken as a correction value is Fm
If the maximum value (the output of the element with the highest sensitivity relative to the reference light) among ax and 5o(i) is Smax, then K = Fmax - Smax
(3) From now on, it is necessary to store each correction value expressed by equation (4) in the correction value storage device (53) in FIG.
Smax was found from o(i) as follows. (2) Read the output 5o(i) of each element with respect to standard light using a synchroscope, etc., and find the maximum value Smax among them. (2) Load the output 5o(i) of each element with respect to standard light into a computer and find the maximum value Smax on the computer using a program. [3] Problems with the prior art When determining Smax as described above, the method (2) relies on manual reading of the output, so in terms of time and reliability of readings, the number of elements exceeds several hundred.・Cannot be applied to image sensors. In addition, method (2) uses a computer and is superior to (2) in terms of reliability and time required, but in addition to the circuit shown in Figure 5, it requires a circuit to read the sensor output into the computer, which reduces cost. Not only is it disadvantageous, but it also requires a computer.
A fatal flaw is that the sensor cannot be replaced in the field. (The variation differs from sensor to sensor.) [4] Means for solving the problem FIG. 1 shows the configuration of the present invention. The operating modes of the present invention are divided into (i) correction data writing mode and (11) normal operating mode. (1) Correction data write mode CAL is true during this period. Standard light is made incident on the image sensor (2), and its output is set to 5o(i). 5o(i) enters the B input of the comparator (4) constituting the highest output storage circuit (3). On the other hand, the output of the latch circuit (3) is input to the comparator (4), and the initial value of the output is O. Clock LCK that operates the latch circuit when the comparison result is A<H
enters the latch circuit (3), and 5 becomes the latch input.
o(i) becomes the output of the new latch. If the comparison result is A2B, the latch does not operate and the output of the latch circuit (3) remains unchanged. Latch circuit (
The output of 3) is compared with the next sensor output as the output of comparator (4). In this way, when the outputs from all the elements of the sensor are read out once, the output of the latch circuit (3) is 5o(i)
is the maximum value Smax. S after 2 late days
Since 5o(i) (input B) exceeding max (eight power of the comparator) never appears, the output of the latch circuit (3) does not change. On the other hand, the upper addresses (A13 to A7) of ROM (5)
is Smax, and 5(i
)(So(i)) is input. This ROM (5
), the correction value F (i) corresponding to each element can be obtained by writing the results for each address in advance as data. F (i) is an input to RAM (6), but since the write clock WCK for RAM is operating during this mode, F (i) is input to RAM (6) in accordance with the element address. written. (11) Normal operation mode In this mode, CAL becomes false and the contents of the latch circuit (3) and RAM (6) do not change. The RAM is read-only, and the sensor output S (i) and RAM according to the amount of light are
get. [5] Embodiment In FIG. 1, all data is 7 bits, but for this purpose it is necessary to use a 14-bit ROM address for the capacity 128 bits. Currently 64 bits or 8
2-bit ROM ((61), (62) or C7L), (72), (73), (74
)), the data will be 6 bits, but this will still provide sufficient accuracy for applications such as OCR (resolution is 26=64). However, 64 or 82KROM
Even if you use the decoder circuit (
By adding 75 pins, it is also possible to create a 7-pit data circuit. In addition, in the configuration of the present invention shown in FIG.
Input or ROM (5) 7) Smax and S (i
) The allocation of addresses (A13 to AO) to the addresses (A13 to AO) is not particularly fixed. In addition, in order to prevent the memory from being lost even if the circuit power is turned off once the correction data has been written, the latch circuit (3) and RAM (6!11:)! You can either add simultaneous backup using the 2-channel system, or use a memory IC with a built-in backup function. [6) Effects The image sensor output normalization circuit of the present invention solves the problem of the prior art in item [31], and its remarkable effects are as follows. (b) Automatically perform normalization without relying on human labor by adding a simple circuit. (b) It is possible to easily replace the image sensor and rewrite the output correction value for the image sensor in the field without the need for a computer. (c) Automatically detect Smax for standard light,
By calculating a correction coefficient using this value, an optimal dynamic range can be obtained. BRIEF DESCRIPTION OF TABLE DRAWINGS FIG. 1 is a diagram showing the configuration of the present invention. FIG. 2 is a diagram showing variations in sensitivity of each element of the image sensor. Figure 3 shows the output status of each element of image sensor d without output correction.
Figure 4 shows the normalization corresponding to Figure 3, G, b, and C. Figure 5 shows a conventional image sensor output normalization circuit. Figure 6 shows a 7-bit output normalization circuit using 64K ROM. Figure 7 shows a 7-bit output normalization circuit using 32K ROM. 1 element address counter 2 image sensor 8 8 latch circuit 4 comparator ROM 51 read address counter 52 image sensor 53 correction value storage 54 Multiplier 6 RA, M ROM 2ROM 7 Multiplier ROM 72 〃 73 〃 74 〃 8 Maximum output storage circuit

Claims (1)

[Claims] (1) A circuit for normalizing the output of an image sensor in which photoelectric conversion elements are arranged to output electrical signals according to the amount of received light, and is a maximum output memory consisting of a comparison circuit (4) and a latch circuit (3). Input the maximum output value of the output from the circuit (8) and the latch circuit (3) and the output from each element for the standard light amount as an address to the address counter (1),
The output of the ROM (5) is stored in a ROM (5) written in advance so as to output a correction value uniquely determined from both, and the output of the ROM (5) is stored in a position corresponding to the address of the element of the image sensor (2). It is equipped with a RAM (6) for storing correction values that is read out in accordance with An image sensor output normalization circuit characterized by: