JPH02226966A - Original reader - Google Patents

Abstract

PURPOSE:To prevent difference from illuminance distribution at an original read position for each decomposed color, deterioration in gray balance, shading correction defect, or fog or jump of a picture by setting a correction coefficient between a 1st white board and a 2nd white board independently of each decomposed color at shading correction. CONSTITUTION:A color separation picture signal read for each line by a 3-line color CCD sensor 103 while being subjected to exposure scanning is inputted to a video processing circuit 200, in which signal processing is applied. A signal line 214 is a signal line driving the 3-line color CCD sensor 103 and a required drive pulse is generated by a pulse generator 201. A reflected light in a white board which is irradiated by a halogen lamp is subjected to read scanning by the 3-line color CCD sensor 103 to obtain a signal level with a prescribed density and the level is used to correct the white level of a video signal. Outputs from 3 line sensors 103a, 103b, 103c of the 3-line color CCD sensor 103 are amplified to a prescribed level by amplifier circuits 202, 203 and 204.

Description

DETAILED DESCRIPTION OF THE INVENTION (A) The present invention relates to a document reading device, and more particularly to a document reading device that has a plurality of image sensors arranged in parallel for each color separation and reads images line-sequentially.

[Conventional technology]

Conventionally, in image reading devices that digitally read images using an image sensor such as a COD, problems occur due to uneven sensitivity of each pixel of the image sensor, uneven illuminance of the illumination system consisting of a light source such as a halogen lamp, a reflective shade, etc. Image shading is corrected using white level correction (which uses a standard white plate as a reference for the white level when reading images).
It was possible to correct this by using shading correction).

That is, in the shading correction, as described above, during reading, prior to reading the original, a standard white plate placed outside the original position is read, and the actual image reading signal is corrected according to the reading level.

That is, R7 is the reading data of the standard white board, R1 is the uncorrected image data when the actual image is read, and the patent Ro '' (R1/R,) x 255 (1
) 255: Correction is performed by a calculation called a normalization coefficient when the image data is 8 bits, but as can be seen from equation (1), the level of the corrected image data R0 is the same as the level of the standard white plate reading data R1. greatly influenced by. In other words, if the reflection density of the standard white plate is lower than the reflection density of the white part of the image before correction (R, < R,), the value of the corrected image data R0 when reading the white part of the original is calculated from equation (1). Ro<255, and the white portion of the document does not become white, but appears as an image fog.

Conversely, if the reflection density of the standard white plate is higher than the reflection density of the white part of the image before correction (Rt>R-), the value when reading the white part of the original of the image data R0 after correction is R1≧R1. With respect to the uncorrected image data, R,=255, and the image is read as an image whose density has stopped at a low density level. Furthermore, it is very difficult to manage the reflection density of the standard white plate, and variations in the reflection density of the standard white plate cause fogging and jumps in reproduced image density, and differences in gradation reproducibility appear between devices.

Therefore, as a method that does not depend on the variation in the reflection density of the standard white plate and has a relatively stable reflection density, for example, copy paper is used as a standard white plate and a second standard white plate, and the standard white plate and copy paper are used as a second standard white plate. Some printers achieve stabilization of the white level by measuring the ratio of the reflection density of paper in advance.

That is, the copy paper read data RI), the standard white plate read data R3, the image data before correction R1, the image data after correction R0, and the ratio of the reflection density of the copy paper and the standard white plate are C.
Then, C= Ro / R,'
(2) Ro = (Ri / (Rs XC)) x2
55 (3) The correction is executed by the calculation.

Furthermore, in recent years, various color image reading devices have been developed. There are two simultaneous reading color separation methods: one is a method in which a stripe filter is configured on one line of image sensors and the color separation signals are read out point-sequentially in a time-division manner, and the other is a method in which image sensors are connected in parallel for each separated color. There is a method of having multiple lines and sequentially reading them.

In the method of reading out color separation signals dot-sequentially, it is mainly put into practical use as a contact type sensor in which multiple image sensors are connected together in the main scanning direction.
Since the color separation stripe filter is configured as shown in the figure, there is no difference in the illuminance distribution on the reading document surface of the illumination system for each color.

FIG. 14 shows the illuminance distribution on the document surface of the reading device using the contact type sensor 1301 shown in FIG.
1401 is a rod array lens for forming an image on the document surface at the same magnification on a contact type sensor, and 1402 is a platen glass, which determines the illuminance in the longitudinal direction at the imaging position of each color separation stripe filter of the contact type sensor. The distributions are equal.

Therefore, in the measurement of the ratio C of the reflection density between the copy paper and the standard white plate when performing the shading correction shown in equation (3), it is not necessary to perform the measurement for each color in practice. That is,
If a stripe filter for color separation is composed of R, G, and B, for example, the output of the G filter, which has the widest spread of spectral sensitivity, will give the ratio C of the reflection density of the copy paper and the standard white plate. By measuring C in other colors,
Shading correction for other colors is performed using the reflection density ratio C of the G filter.

(The problem that the invention is trying to solve) However,
In a method in which a plurality of image sensors are arranged in parallel for each separated color and read out line-sequentially, there is a physical distance of more than ten lines between the image sensors of each color.

Therefore, the distance between the reading lines between each CCD on the actual document surface is, for example, if the pitch of the photodiode of the image sensor is 10 μm x 10 μm, the distance between each of R, G, and B CCDs is 18 lines, and the reading resolution is 400 dpi. The distance between the ins to be read is as follows, and it is necessary to use the illumination system to uniformly expose the entire width of 2.286 mm.

However, it is difficult to efficiently and uniformly expose the above-mentioned width of <2.286 mm, resulting in an illuminance distribution as shown in Fig. m15 (a). In addition, although the illuminance distribution on the platen glass is as shown in Figure 15 (a), at the mounting position of the standard white plate, the distance between the illumination system and the platen glass and the distance between the illumination system and the standard white plate are Due to some differences, the first
The illuminance distribution changes as shown in FIG. 5(b). As a result, the rate of change for each color is Rx'/RX, Gx'
/Gx, Bx /ax are not equal, and in a method in which a stripe filter is configured on a 1-line image sensor and color separation signals are read out point-sequentially in a time-division manner, the G signal is used to differentiate between copy paper (white paper) and standard white. When shading correction is performed by calculating the ratio C of the reflection density of the plate, in the case of Fig. 15(b), the rate of change in illuminance for each color becomes R<B<G, and R
, B, when reading gray having the same density as G as an image signal after shading correction, the density is read as low. In other words, gaps in the image occur, deterioration of gray balance and variations in gradation reproducibility occur, leading to deterioration of image quality.

Furthermore, the illuminance distribution in the longitudinal direction may vary slightly between the R, G, and B sensors, and conventional corrections do not provide appropriate correction, resulting in deterioration of image quality.

Here, FIG. 15 shows the change in illuminance distribution on the reading surface when the above-mentioned sensor is used, and 1402 is a platen glass, 1501 is a halogen lamp, 1502,
1,503 are reflective shades, and 1504 is a white plate. BX, GX.

Rx is R, G at the position X on the platen glass 1402
, B is the document surface illuminance of each sensor, sxl, GX'
, R,l are the document surface illuminances of the R, G, and B sensors at the position of the white plate 1504.

(Means for Solving the Problems) In order to achieve such an object, the present invention provides a document reading device that has a plurality of image sensors arranged in parallel for each separated color and reads a document line-sequentially. The white of the document is corrected using a first white board that serves as a reference for the reading level of the white part, and a second white board that corrects the reading level of the first white board, and the reading level of the first white board is adjusted. The present invention is characterized in that correction coefficients for correction using the second white plate are independently set for each separated color.

(Example) Hereinafter, the present invention will be described in detail using preferred examples with reference to the drawings.

FIG. 2 is a diagram showing an example of a schematic internal configuration of an image reading device to which the present invention is applied.

ioo is an original pressure plate, 1402 is a platen glass on which the original 102 is placed, and 110 is a halogen lamp 1 for exposing the original.
501 and the first reflecting mirror 105, and 111 is a mirror unit consisting of the reflecting mirror 10.6 of 's2 and the third reflecting mirror 107. 108 is a lens unit for reducing and forming a reflected light image from the original 102 exposed and scanned by the halogen lamp 104 on the 3-line color CCD sensor 103; It moves in the direction of arrow A (sub-scanning direction) at a scanning speed of .

Color separation image signals read line by line by the 3-line color CCD sensor 103 during exposure scanning are input to a video processing circuit 200 shown in FIG. 1 and subjected to signal processing.

A signal 214 is a signal line for driving the 3-line color CCD sensor 103, and necessary driving pulses are generated by the pulse generator 201.

In FIG. 2, reference numeral 1504 is a white plate for correcting the white level of an image signal, which will be described later.
By reading and scanning with the CD sensor 103, one novel signal of a predetermined density can be obtained, which is used for white level correction of the video signal.

The video processing circuit 200 shown in FIG. 1 is controlled by a central controller CPLI 230 via a bus 231.

Next, the video processing circuit 200 will be described in detail. As mentioned above, the original 102 is first irradiated with the halogen lamp 1501, and the reflected light from the original 102 passes through the mirror units 110 and 111, the optical lens 108, and the 3
A reduced image is formed on the line color CCD sensor 103, separated into colors, and read. Three line sensors 103a and 103b of the 3-line color CCD sensor 103.

The output from 103C is the amplifier circuit 202, 203゜204
is amplified to a predetermined level.

Figure 3 shows the 3-line color CCD sensor 103 shown in Figure 2.
This is a configuration diagram of 10μm×10μm as one pixel, and 5
A line sensor 103a with 000 effective pixels,
Three lines of sensors 103b and 103c are arranged in parallel on a common chip 30 with a distance of 180 μm (18 lines), and each line sensor is dyed with R, G, and B organic dyes. In addition, each line sensor is R-CC.
D103a, G-CCD103b, B-CCD103c
shall be. Further, each COD is driven independently and synchronously.

Each CCD 103a, 103b,! From 03c onwards, video signals V,,V, are generated in synchronization with the drive pulse.

■, are outputted independently, and independent amplifiers 202, 203 .
After being amplified to a predetermined voltage value in step 204, it is converted into 8-bit digital data by A/D conversion circuits 205, 206, and 207, respectively.

Next, in this embodiment, as described above, each CCD has an interval of 18 lines (10 Atm x 18 = 180 μm) in the sub-scanning direction, and the image is scanned in advance in accordance with the reading direction A in FIG. When viewed from the CCD 103a of the Tsudo (R) filter, the CCD 103b of the Green (G) filter,
The distance between the blue CB) filter and the CCD 103c is 18 lines (180 μm) and 3B line (360 μm), respectively.
μm), and each CCD 103a, 103b.

The reading position on the document by 103c is shifted.

Therefore, in order to connect this correctly, the following correction operation is performed using memory for multiple lines. In other words,
B-CCD103c to be read last among the three CODs
In order to synchronize with
If i, in the case of 100% same size reading, R-CCD1
The buffer memory 208 of 03a contains 36 lines of G-
The buffer memory 209 of the CCD 103b has a synchronization memory for 18 lines. Furthermore, if the maximum expansion rate is 400%, each buffer memory has 208.20%
The number of lines in 9 is 36 x 4 = 144 lines, 1
QX4-72 lines are required. Here, the number of pixels of one line stored in the buffer memories 2013 and 209 is at least 4678 pixels, which is the effective number of pixels per line, when reading an A4 size 297 mm in the longitudinal direction at 400 dpi.

In this way, the R-CCD 103a that reads in advance,
When the B-CCD 103c reads the same line as the line that the G-CCD 103b has already read, the same line is read in three colors (R, G
, B), each color data is obtained.

FIG. 4 shows the configuration of the thermal correction circuits 210, 211, and 212. Note that since these have the same configuration, the thermal correction circuit 212
will be explained as a representative. The black level output of each COD has large variations between pixels when the amount of light input to the sensor is small, and if this is output as is and pressure is applied to the image, for example, if the image is printed with a printer, the data part of the image will be Streaks and unevenness occur. Therefore, it is necessary to correct the output variations in the black portion, and the circuit shown in FIG. 4 performs the correction.

That is, prior to the copy operation, without lighting the halogen lamp 1501, the CCD 103a,
The read outputs of b and c are input to this circuit. In order to store one line of this black level image signal input to the aint1 child into the black level RAM 401, the CPU 230 sets data in the latch 408, selects A of the selector 402, closes the gate 403, and stores the data in the black level RAM 401. open.

That is, the black level image signal is sent to the selector 402 and the gate 40.
4 and is manually input to the RAM 401. On the other hand, RAM401
A of the selector 40B is selected so that the count output of the address counter 405 initialized by H3YNC is manually input, and the black level image signal for one line from the gate 404 is input to the address from the address counter 405. The data is stored in the RAM 401 according to the value (the above is defined as a black reference value import mode, and the black level image signal stored in the RAM 401 is defined as black level data).

However, the black level data imported into the RAM 401 in this way is at a very small level, so the 3-line color C
It is easy to be affected by noise generated by the CD 103, amplifiers 202, 203, 204, etc., and if the data is used as it is as uncorrected data, the image of black areas may become noisy and rough, so it is preferable. do not have.

Therefore,'! The CPU 230 executes the calculation shown in the flowchart of FIG. 5 on the black level data taken into the black level RAM 401 shown in FIG. In order to import the black level data stored from addresses B to B4a7a of the RAM 401 into the work register of the CPU 230, the CPU 230 sends gate 4 to the latch 408.
03 and 404 are closed, gate 409 is opened, and data is set to select B of selector 406. As a result, the black level RAM 401 is accessed by the address from the CPU 230 address bus, and the CPU 2
Black level data (B,) to (B4sta) are read through the 30 work registers 409 and the data bus.

Next, in step 502, the black level data (from address B to B487B loaded into the work register)
B, )......(B 4a7a) is added, divided by the number of data 4678, and the result is stored in -H working RAM.
Store in. Then, in step 503, the gate 40
Latch 408 closes 3 and 409, opens gate 404, and selects B of selectors 402 and 406.
The data is set to address B of the black level RAM 401 again.

From B 4117♂, the data in the tracking RAM is written via the selector 402 and gate 404 in accordance with the address data from the CPU 230. In this manner, by taking the average value of the black level data, black level data from which noise has been removed is set in the RAM 401.

In addition, in this example, the black level data was calculated by calculating the average value of all pixels of the black level data and used as the corrected black level data. It is also possible to perform calculations using weights multiplied by .

After the black level data capture and correction operations as described above, reading of the original image is executed. When reading the original image, the CPU 230 closes the gates 404 and 409, opens the gate 403, sets data in the latch 408 to select A of the selector 406, and sets the data in the RAM 4.
Set Q1 to data read mode. As a result, the black level data in the black level RAM 401 is read out pixel by pixel according to the address value from the address counter 405.
Bl passes through the gate circuit 403 and enters the subtracter 4070B input.
It is manually inputted in synchronization with the input of the image signal to the r+ terminal.

Therefore, the output of the black correction circuit shown in FIG.
The black level data DK(i)(l
is the pixel address), and B for the blue (B) signal.
in(i) -DK (1) -B, ut(i) is obtained for each pixel (no correction mode). Similarly, the same control is performed for green (G) and red (R) in black correction circuits 210 and 211 having the same configuration as in FIG. 4.

Each color data subjected to black level correction in this way is stored in white level correction (shading correction) sections 213 and 21, respectively.
4.215 is entered, and white level correction is performed. White level correction (shading correction) using halogen lamp 1501
Each CCD 10 when irradiating a uniform white plate 112
3-a.

FIG. 6 shows an example of the configuration of a white level correction section that corrects variations in sensitivity of the illumination system, optical system, and sensor based on the white data outputted by 103-b and 103-c. The basic circuit configuration is the same as the thermal correction circuit shown in FIG. use

That is, in this embodiment, as the initial value setting mode, a blank sheet of paper (for example, copy paper) is placed on the platen glass 1402, and the ratio of reflection density between the white sheet and the white plate 1504 is determined for each color COD according to the flowchart below. The data is measured and backed up to a backup memory (not shown) of the CPU 230.

However, since the white level correcting units 213, 214, and 215 have the same configuration and operation, the white level correcting unit 213, 214, and 215
I will represent and explain.

In the flowchart of FIG. 7, in step 701, the optical system 110 is moved to below the white paper on the platen glass 101, in step 702, the halogen lamp 1501 is turned on, and in step 703, the uniform white level is output from the CCD 103c. Image data is stored in correction RAM 601 for one line via selector 602 and gate 604.

Next, in step 704, the CPU 230 uses the correction RAM 6
White image data (Wl) 1111~ stored in 01
Near the center of 467♂, for example, address W2331
White image data of NW2, 46 (W2331) ~ (Was
4a) is taken into the work register of the CPU 230 via the gate 601, and in step 705 the address W ***
White image data at r-W 2346* (W2331)
Add ~(W23411) and divide by the number of data 16,
The result (Wo) is stored in the working RAM.

Then, in step 70B, the optical system 110 is connected to the white plate 150.
CCD 103c again in step 707.
The uniform white level white image data output from
Store in 01.

In step 708, the CPU 230 stores the white image data (W2) at addresses W2331 to W2346 in the correction RAM 601.
s3+) to (W23411) to the working register, and in step 809, the white image data (W2331
) to (W2sna) are added and divided by the number of data 16, and the result (ws') is stored in the working RAM -1.

Next, in step 710, the average white image data (Wo) of the blank paper stored in each working RAM is divided by the average white image data (Ws) of the white board, and the measured value C=(Wg) for the white board is calculated. / (Wo) and is stored in the working RAM and backed up.

Here, the output measurement value of the blank paper is set to N (・type 255),
In addition, as mentioned above, a white board is a so-called gray with a lower reflection density than a white paper, but it is extremely difficult to reproduce a perfect gray, so it may have some color tinge.
If there is some discoloration, if white correction is performed using this as a white plate, the gray balance of the read image data will be disrupted. Therefore, each C of R, G, and B when reading a white board
Assuming that the output balance of the CD has collapsed, by setting the white board brightness correction factor K (= typl) for each color as a coefficient to correct the imbalance, it becomes possible to correct the gray balance described above. , the expected white paper output value N and the white board brightness correction factor are also stored in the working RAM.
and is backed up together with the white board brightness measurement value C.

Next, during actual correction, the exposure lamp 104 is turned on and the white plate 112 is
The white image data of the uniform white level outputted from each CCD is stored in the correction RAM 801 for one line via the selector 602 and the gate 604.

Next, a correction coefficient for each pixel is calculated according to the flowchart in FIG. That is, in step 801, the correction RAM 6
In order to import the white image data stored at addresses W, 01 to W4676 into the work register of the CPU 230, the CPU 230 sends the gate 6 to the latch 608.
03 and 604 are closed, gate 609 is opened again, and data is set to select B of selector 606. As a result, the correction RAM 601 is accessed by the address from the address bus of the CPU 230, and the white image data (W r ) to (W 4aya) are read into the work register of the CPU 230 via the gate 609 and the data bus.

Next, in step 802, white image data (Wl) from addresses Wl to W<aya taken into the work register
For ~(W 46,8), use the expected blank output value N backed up in the working RAM, the white board brightness correction factor K, and the measured white board brightness value C to obtain E, m (to NxCx) / W 1 (i-1 to 4678) A shading correction coefficient is calculated for each pixel.

Then, in step 703, the gates 603 and 609 are closed, the gate 604 is opened, data is set in the latch 808 to select 8 of the selectors 602 and 606, and the addresses Wl to W of the correction RAM 601 are set again. ,
. Data in the working RAM is written via selector 602 and game 1-604 according to address 2 data from CPU 230.

When reading the original image, the correction coefficient E is stored in the correction RAM 60 in synchronization with the human image data DI input to the Bin terminal.
1 to the multiplier 607 via the gate 603,
The calculations D, wD, and XE are performed for each pixel and output as corrected data.

As described above, the black level and white level corrections based on various factors such as the black level sensitivity of the human image system, dark current variations, optical system light intensity variations, and white level sensitivity are applied to the read data of the original image of each COD. The data of each color proportional to the amount of input light, which is executed uniformly across the main scanning direction, is
The signals are input to logarithmic conversion circuits 216, 217, and 218, respectively, in accordance with the relative luminous efficiency characteristics of the human eye.

Here, each color data output from the logarithmic conversion circuits 21EJ, 217, and 218 for each B, G, and R is, for example,
It corresponds to the density value of the output image by subtractive color mixing of the three color toners of yellow, magenta, and cyan.
Since the toner amount of M) corresponds to the toner amount of cyan (Cy) for R, from now on, the color image data will be Ye
, M, and Cy.

Each color component image data from the document obtained by logarithmic conversion by logarithmic conversion circuits 216, 217, and 218, that is, Y
Color correction 219 is performed on the e component data, M component data, and Cy component data. That is, as shown in FIG. 9, the spectral characteristics of the color separation filter arranged in the 3-line color CCD sensor 103 of this embodiment have unnecessary transmission areas such as diagonal lines, and the spectral characteristics are not transferred to the transfer paper Since the colored toners (Y, M, C) generally have unnecessary absorption components as shown in FIG. 10, each color component data Yi, Mi
, Ci, the masking circuit 219 calculates a linear equation for each color and performs color correction.

Furthermore, by each color component data Yl, M+, Ci, Y
Min (Y
i , Mi , Ci ), this is used as a smear (black), and then black toner is added (smear removal), and undercolor removal (smear removal) is performed to reduce the amount of each coloring material added according to the added black component.
tJcR) operation is performed in the black extraction and UCR circuit 220.

Each color signal obtained by the masking correction and black drawing ratio processing by the masking circuit 219 and the UCR and smearing processing by the UCR circuit 220 is
After correction processing and, if necessary, scaling control in the main scanning direction by the scaling circuit 212, edge enhancement and smoothing processing are performed by the filter/sharpness circuit 213.

[Other Embodiment 1] FIG. 11 shows another embodiment of the white correction section (shading correction section).

In the white correction unit shown in FIG. 6 described above, prior to the copying operation or the reading operation, the white image data of one even-white novel read from the white plate 112 is stored in the correction RAM 601 for one line.
Next, the CPU 230 calculates a shading correction coefficient El for each pixel using the white image data Wl, the expected blank output value N, the white board brightness correction factor K, and the measured white board brightness value C, and stores it again in the correction RAM 601. .

Then, when actually reading the original T14, the correction coefficient E is synchronized with the input image data DI input from the B in4 child.
, is output from the correction RAM 601 and is subjected to correction calculation by the multiplier 607.

On the other hand, in FIG. 11, the corrected data D0 is determined from the correction coefficient E+ and the level of the input image data D.
is calculated by CP (7230) and corrected RAMll0I
The input image data D, which is input from stn@child when reading the original, is manually input to the address terminal of RAMll0I, and the corrected data D0 corresponding to the manually inputted image data D1 previously stored in the correction RAMll01 is sequentially input. The image is output and shading correction is completed.

(Other Embodiment 2) FIG. 12 shows another embodiment of the white correction section (shading correction section).

In the white correction unit shown in FIG. 6 described above, before copying or reading, the white image data of the uniform white level read from the white plates 1 and 12 is stored in the correction RAM 601 for one line, and then the white image data is stored in the correction RAM 601 by the CPU 230. A shading correction coefficient El is calculated for each pixel using the image data Wl, the expected blank output value N, the white board brightness correction factor K, and the measured white board brightness value C, and is stored again in the correction RAM 601.

During actual document reading, the correction coefficient E is added from the correction RAM 601 in synchronization with the input image data DI input manually from the Bin terminal, and the multiplier 607 performs correction calculations.

On the other hand, in FIG. 12, the above-mentioned expected white paper output value N, white board brightness correction factor K, and white board brightness measurement value C are
Latch circuits 1202, 12 that latch from the U data bus
04. Multiplier 1201, 1 that performs a correction operation on each latched coefficient N, C and human image data D1.
2031205, a correction RAM 1211 that stores white image data when reading a white board, and outputs shading correction data using the white image data output from the correction RAM 1211 and the calculation result of each correction coefficient N, KC and human image data as an address. The difference is that correction is performed on all hardware.

The correction ROM 1207 stores the image data value before correction R1, the image data value after correction R0, and the image data value R3 when reading the white board.
In this case, Ros=(R,/R,)*255 255; Data that can be expressed by a normalization coefficient is written as the contents of the correction ROM. The details of the operation are the same as those of the previous embodiment, so a description thereof will be omitted.

(Effects of the Invention) As explained above, by setting the correction coefficient between the first white plate and the second white plate independently for each separated color during shading correction, the illuminance at the document reading position for each separated color can be adjusted. Deterioration of gray balance, poor shading correction, and image fog caused by differences in distribution, differences in illuminance distribution for each separated color between the first white plate and second white plate, and differences in color tone between white plates, etc. It is possible to prevent deterioration of image quality such as black spots.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a video processing circuit representing this embodiment. FIG. 2 is a diagram showing an example of a schematic internal configuration of an image reading device applicable to the present invention. FIG. 3 is a configuration of a 3-line color CCD sensor. Figure 4 is a block diagram of the black correction circuit, Figure 5 is a flowchart showing the procedure for point correction, Figure 8 is a block diagram of the white correction circuit, and Figures 17 (a) and (b) are diagrams of the white correction circuit. Flowchart 1 shows the procedure for measuring the reflection density between the 1°'s2 white plates, Fig. 8 is a flowchart showing the procedure for white correction, Fig. 9 shows the spectral characteristics of the color filter, and Fig. 10 Figure 11 is a diagram showing the spectral characteristics of color toners, Figure 11 is a configuration diagram of a white correction circuit in another embodiment, Figure 12 is a configuration diagram of a white correction circuit in another embodiment, and Figure 13 is a conventional example. 14 is a diagram showing the illuminance distribution on the document surface in the conventional example, FIG. 15 is a diagram showing the illuminance distribution on the document surface in this embodiment, 103 is 3 Line color CCD sensor, 210-212
is a black correction circuit, 213 to 215 are white level correction circuits, and 6
01 is a correction RAM.

Claims (1)

[Claims] It has a plurality of image sensors arranged in parallel for each separated color, an illumination means for illuminating the original, and a condensing lens for condensing the reflected light from the illuminated original onto the image sensor. The white of the document is corrected using a first white board that serves as a standard for the reading level of the document, and a second white board that corrects the reading level of the first white board. A document reading device characterized in that correction coefficients to be corrected using a second white plate are set independently for each separated color.