JPS63269679A - Image reader - Google Patents

Abstract

PURPOSE:To simplify the circuit configuration of an image reader, by making the resolution of shading data for shading correction coarser. CONSTITUTION:By reading out shading data 210 which are the reading level signals of a standard white plate for each picture element of a CCD 103 from a shading data RAM 208 in correspondence to the output of each picture element of the CCD 103, the shading contained in the original reading signals 209 is corrected by means of a shading correction table ROM 211. The shading data 210 from the RAM 208 is used as the gradation data of a half of the original reading data 209. Therefore, the shading correction table ROM 211 can be reduced in capacity.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image reading device that reads a document pixel by pixel, and more particularly to an image reading device that has a function of equalizing and correcting bright areas of a reading signal. It is.

[Conventional technology]

Conventionally, in this type of document reading device, the exposure amount of the light source,
Shading correction is performed to electrically eliminate non-uniformity in reading output due to non-uniformity in lens transmittance and photoelectric conversion characteristics. For example, a standard white plate that serves as a reference for the white level is placed outside the document reading area, and the photoelectric conversion element output from this white plate is A/D converted and stored in memory. The original reading signal is corrected based on the reading signal.

That is, the signals obtained by A/D converting the white board reading signal and original reading signal read from the memory are input to the addresses of correction tables such as R and OM, and correction outputs are obtained as data from the ROM.

[Problem to be solved by the invention] Here, the white board reading signal and the original reading signal are converted into shading correction tables R and OM as A/D converted multivalued digital signals.
In order to obtain an address signal of 1, the larger the multi-level data length of each signal, the larger the capacity of the correction table ROM is required.

For example, if the multi-value data length of the white plate reading signal and original reading signal by A/D conversion is 8 bits each, the address lines of the shading correction tables R and OM are 16 bits.
If this is necessary and the length of the corrected data is also 8 bits, the capacity of the correction table ROM will be 64 bytes. Therefore, this is a major hindrance to the reliability and miniaturization of the device, and has a negative effect on reducing the cost of the device.

[Failure to solve the problem]

In order to eliminate the drawbacks of the above-mentioned conventional example, the present invention has the following features:
By lowering the degree of resolution of the digital data of the standard white board reading signal, the configuration for correcting the original reading data is reduced.

〔Example〕

The present invention will be explained below using preferred examples.

FIG. 1 is a schematic diagram of a document reading device to which the present invention is applied. A CC is equipped with several dozen, for example, 5,000-bit light-receiving elements, in order to read the image information of the original 102 held by the original cover 100 and placed on the original platen glass 101.
An imaging device (CCD) 103 such as a D-line sensor is used. Illumination light from a light source 104 is reflected on the surface of the original 102 and passes through mirrors 105, 106 and 107 to the lens 1.
08 to form an image on the CCD 103. light source 104
, an optical unit 113 consisting of a mirror 105 and a mirror 1
The optical unit 114 consisting of 06° 107 is adapted to move at a relative speed of 2:1. This optical unit moves at a constant speed in the sub-scanning direction perpendicular to the direction in which the light-receiving elements are arranged (main-scanning direction) while being subjected to PLL control by a DC servo motor 109. The moving speed is 22.5mm/22.5mm on the outward path (as shown in the figure, arrow X direction) depending on the magnification.
It is variable from sec to 35 Qmm/sec, and is always 800 mm/sec on the return trip.

While reading the main scanning direction (hereinafter referred to as the Y direction) perpendicular to the sub-scanning direction (hereinafter referred to as the X direction) in which this optical unit moves with a resolution of 400 dots/inch, the optical unit is moved to the leftmost home position. After moving to the right to a predetermined position, it is moved back to the home position again to complete one scan.

A home position sensor 110 in which a light shielding plate 111 provided in an optical unit 113 is a photointerrupter.
The home position is detected by crossing the The standard white plate 112 is used for shading correction and light intensity control of the light source 104, and the position where the optical unit 113 is detected by the home position sensor 110 is the reading position of the standard white plate 112.

FIG. 2 is an example of an image processing block of the document reading device shown in the first figure.

The reading surface 201 reflects light from the illumination 104 and the CCD 1
03, which has document density information read by
By moving the optical unit, the standard white plate 112 in FIG.
, the original 102 becomes the reading surface.

The image signal for each line read by CCDIQ3 is
After being amplified by the amplifier 202, the A/D converter 203
As a result, a multivalued (8 bits in this embodiment) digital image signal 209 is obtained. In this embodiment, the black level of the original is “0”.
'' and the white level is output from the A/D converter 203 as "255".The black reference level and white reference level of the A/D converter 203 are given fixed values and are programmed. The offset of the amplifier 202 is adjusted so that the output of the A/D converter 203 becomes O level, which is the perfect point, when the illumination 104 is completely turned off and no light enters the CCD 103.

A COD drive signal generation circuit 204 generates CCD drive signals 205 such as a reset signal, a clock signal, a horizontal synchronization signal, etc. necessary to drive the CCDIQ3, and also generates an A/
Generation of clock signal 206 to D converter 203, CCD
C, which is an address signal for identifying each bit of 103.
A CD address 207 is generated. In this example, CCD
103 is a 5000 pixel line sensor, the CCD address 207 is 0 to 499 pixels corresponding to the 5000 pixels read out in synchronization with the horizontal synchronization signal.
Count up to 9.

208 is the standard white plate 1 from the A/D converter 204
This is a shading data RAM that stores all pixels of a digital image signal for one line of the CCD obtained by reading 12 according to the CCD 207.

The shading correction table ROM 211 is the illumination system 1
04 and lens 108 and CCD 10
Tables R and OM are used to correct non-uniformity of the read image signal 209 caused by sensitivity unevenness of each pixel of 3, amplification setting error of the amplifier 202, etc., and when reading the original,
Standard white board 1 from shading data RAM208
By addressing the correction table ROM 211 based on the shading data 210 of No. 12 and the read image signal 209, an image signal 212 with the non-uniformity of the read image signal 209 corrected is obtained.

With the above configuration, in this embodiment, the optical unit 113 is moved to a position where the light shielding plate 111 blocks the photointerrupter 110 prior to the document reading multi-scanning. Then, the standard white plate 112 is made to correspond to the reading surface 201 of the CCD 103,
Turn on the fluorescent light 104 and scan the standard white plate 112 with the CCD.
One line of image data obtained by reading in 103 is written to shading data R and AM 208.

Next, in order to read the original 102, the optical unit 113 scans the original 102 in the X direction at a predetermined speed.

At this time, in response to the output of the image signal 209 of each pixel of the CCD 103, the shading data 210, which is the reading level signal of the standard white plate 112 at that pixel, is read out from the shading data RAM 208, and the shading correction table ROM 211 reads out the original reading signal. The shading included in 209 is corrected.

The correction operation in the shading correction table ROM 211 is shown in FIG. Here, shading data RAM2
The shading data 210 from 08 onwards is half the gradation data of the original reading signal 209. That is, the original reading signal 209 is 8-bit gradation data, whereas the shading data 210 is stored in the RAM 208.
The lowest pitch of the 8-bit gradation data written in
SB) is truncated and input to the shading correction table ROM 211 as 7-bit gradation data.

This is to reduce the capacity of the shading correction table ROM 211. That is, shading data 2101. , the original reading signal 209 are both 8-bit gradation data, the shading correction table ROM2
Address number 11 is 16 bits, and the capacity of the shading correction table ROM 211 is 64 bytes. However, if the LSB of the shading data is ignored and 7-bit tone data is used, the address of the shading correction table ROM 211 will be 15 bits, and the RO
The capacity of M211 is only half that, 32 bytes.

Here, the gradation resolution of the shading data 210 is set to 1
It is expected that the accuracy of the shading-corrected data 212 will deteriorate by reducing the bits, but this will be explained below using FIG. 3.

FIG. 3 shows the shading correction operation in the shading correction table ROM 211.

Here, S is the 7-bit shading data 210 input from the shading data RAM 208 to the upper address of the shading correction table R,0M211.
corresponds to

X is an original reading signal 209, which is 8-bit data input to the lower address of the ROM 211.

The shading correction in this embodiment uses the characteristic that the output of the CCD sensor 103 generates a linear output with respect to the amount of light! When the shading data from the standard white plate 112 at a certain pixel of the CCD 103 is S, the original reading data X at that pixel is changed so that the original reading data corresponding to the shading data SK has a normalized white data value of 255. is corrected to y.

Note that the 7-bit shading data S input to the shading correction enable ROM 211 is doubled to become 8-bit shading data and treated as 2.S.

Therefore, the amount of correction required to change the original reading data having the value of 2°S to a white data value of 255 is 255-2·S. If the value of the original read data 209 at the pixel having this shading data of 2·S is defined as X, then the correction amount of The fading-corrected value y is obtained as follows.

Considering the error caused by the 7-bit shading data S, there are two possible 8-bit shading data corresponding to S: 2.S and 2.S+1.

In the case of 2.S, the above formula can be applied as is and there is no error, but the original 8-bit shading data is 2! S+1
In this case, if 2·S+1 is used, the original reading value X will be corrected as y' as shown in the following equation, which will cause an error with the corrected value y using the 7-bit shading data 210.

This error is determined as y-y'.

Here, the error becomes maximum when x=2.8, and the error at that time is as shown in the following equation.

The smaller the shading data S, the larger this error will be, but if the shading data S is half the light amount data of pure white (255) as a realistic range of shading correction (2S2128), then The correction error is y-y'=1.97.

In other words, even if the resolution of the kneading data S is reduced from the original 8-bit resolution to 7-bit resolution, the resulting error in the shading-corrected data 218 will be less than 2 for the 8-bit data value at most, which is not practical. It turns out that there is no problem with the above.

In the configuration shown in FIG. 2, 8-bit data is stored in the shading data RAM, and when read out, 7-bit data is formed by excluding the L8B signal. The 7-bit data excluding the LSB signal is stored in the shading data RAM 20.
The resolution may be made coarser by storing it at 8.

Further, another embodiment in which the shading correction error caused by lowering the resolution of shading data in the above embodiment is eliminated will be described below.

As mentioned above, the shading correction table ROM211
The 7-bit shading data S input 210 can be equivalent to 2S and 28+1 when converted to original 8-bit data.

If the 8-bit value is 28, the shading correction output 212 does not include any error, but in the case of 2S+1, it is corrected as y, whereas it should originally be corrected as y' as described above. The value is corrected to be thicker by y-y'.

By the way, as mentioned above, as a realistic upper limit for shading correction, 2.
When S becomes 128, which is half of the maximum value 255, and when the original reading data X is equal to 2·S, the shading correction error y-y' becomes the maximum value 1.97.

Now, when we calculate the actual y' and y, we get y’=253.03=253 y2255 becomes.

However, since the actual 8-bit digital correction value cannot express the decimal places, the value of y', 253.03, becomes 253 when the decimal places are discarded, and the error of 1.97 becomes 2.0.
becomes.

In other words, this means that even if there is an error of 0.01, it may become an error of 1.0 due to digital rounding error.

Therefore, in this embodiment, as shown in FIG. 4, the error (y-y') that may occur due to the rounding of the shading data S input 210 to 7 bits as the output of the shading correction table ROM 211. is stored in the shading correction table ROM 211, and the error is corrected by the subtracter 216.

That is, the 7-bit shading data S input 21
If 0 is 64 (2S=128), the original reading data X
If input 209 is 128, shading data zRA
The original 8-bit shading data stored in M208 can take values 128 and 129.

Here, if the 8-bit shading data is 128, there will be no correction error compared to when shading correction is performed using the 7-bit shading data S input, but if the 8-bit shading data is 129, as described above, This results in a correction error of y-y'-2. Therefore, the error amount 2 is transferred from the shading correction table ROM 211 to 2
It outputs bit error data 214 corresponding to 64 shading data S inputs 210. Then, the gate 215 determines the case where an error occurs in the LSB signal of the data from the shading data RAM2Q8.
The subtracter 216 compensates for the correction error to achieve error-free shading correction. That is, when the LSB signal is 0, the compensation operation is not performed, and when the LSB signal is 1, the gate 215 is enabled and the compensation operation is performed.

In this case, the additional capacities of the subtracter 216, gate circuit 215, and shading correction table ROM 211 are required in addition to the configuration shown in FIG. ROM is 51
However, this method requires a capacity of 288 bits.

In this way, by making the shading data applied to the shading correction table ROM 211 coarser than the resolution of the image read data, it is possible to save the address lines of the shading correction table ROM 211, that is, to reduce the capacity. The inconvenience caused by this can also be removed, and good shading correction can be performed.

Note that shading correction table 2 in this embodiment
The reference numeral 11 is not limited to a ROM, but may be, for example, a RAM, as long as it is a table from which an output can be obtained in response to an input.

Furthermore, the standard white plate 112 may not be pure white but may be gray as long as it is a standard density plate with controlled density.The shading correction in this case is shown in FIG. 3, and the target value for correcting the shading data S is The target value is not 255 but corresponds to the density of the standard density plate.

Further, the reduction in resolution of shading data is not limited to one least significant bit, but may reduce the least significant bits as much as the correction accuracy of the apparatus allows.

〔Effect of the invention〕

As explained above, in the present invention, by coarsening the resolution of shading data for shading correction, the circuit configuration for shading correction can be simplified.
This has the effect of lowering costs and maintaining the accuracy of shading correction compared to conventional discs.

[Brief explanation of the drawing]

Fig. 1 is a diagram showing an example of the configuration of a document reading device to which the present invention is applied, Fig. 2 is a block diagram of a shading correction circuit, Fig. 3 is a schematic diagram of shading correction operation, and Fig. 4 is a diagram showing other methods of shading correction. FIG. 2 is a block diagram of an implementation circuit. 103...CCD, 208 C-Ding data RA
M

Claims (1)

[Claims] An image reading device that photoelectrically converts a document pixel by pixel using a photoelectric conversion means and reads the electrical signal as multivalued digital data using an A/D converter, and has substantially uniform density information. The digital data obtained when the reading surface is read is stored in a memory, the data stored in the memory is read out during actual document reading, the document read data is corrected based on the memory read data, and the memory read data is used for correction. An image reading device characterized in that the minimum resolution of the document is made coarser than the minimum resolution of the document read data.