JP3203689B2 - Image input device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image input apparatus for reading an original and converting the original into digital data.

[0002]

2. Description of the Related Art In a printer or the like which reproduces an image in full color, digital image data R, G, B of red, green, and blue (primary colors) read from a document by a CCD sensor or the like.
For the three colors cyan, magenta, and yellow C, M, and Y
Convert to (complementary color) data and reproduce the image. Therefore, data processing for converting digital data of three colors of red, green, and blue obtained by scanning a document into data of three reproduced colors for image reproduction is performed.

Here, photoelectric conversion data read by a CCD sensor or the like is first converted into a digital value by analog-to-digital conversion, and then a normalization process called shading correction is performed.

In this normalization, for example, a white level is read from a reference white plate, a black level is read in a state where no image signal is input, and read data of a document is normalized between these two levels. . This correction is
Also serves as offset removal.

[0005]

In this shading correction, when the correction reference data is determined from the sampling data for one line, if the correction reference data contains image noise, the reading of the original image will be difficult. Since image noise always appears at the same pixel position,
Highly regular image noise (streaks in the sub-scanning direction) occurs. There are various causes of image noise, such as power supply noise in the system and crosstalk noise in the clock system.
The S / N ratio is deteriorated for the image signal before the A / D conversion.

In order to remove the image noise of the image standardization correction reference data, a method of smoothing the reference data for a plurality of lines is good. However, this requires a large number of line memories.

The present inventor has proposed a method of performing smoothing processing on the main scanning side with respect to correction reference data.
This method does not provide a sufficient noise reduction effect. In addition, there is a drawback that normal shading distortion cannot be completely corrected.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an image input device capable of removing noise from standardized correction reference data using a line memory for one line.

[0009]

An image input apparatus according to the present invention comprises a reference plate, a line sensor for reading a reference plate or a document image line by line and outputting image data, and a reference plate read by the line sensor. A line memory for storing one line of image data, an initial value writing means for writing, as an initial value, one line of reference creation data for each pixel in the line memory, and one line of a reference plate read by a line sensor. Each time data is input, the data for one line of the reference plate read from the line memory and each pixel are weighted and averaged, and the result is again input to the line memory, and this is applied to a plurality of lines. Using the repeating calculating means and the weighted average data obtained by the calculating means, the image data for one line of the original read by the line sensor is corrected. Characterized in that it comprises a correction means.

[0010]

[Function] From the standardized correction reference data for a plurality of lines,
Image correction reference data is generated by weighted averaging using one line memory. With the weighted average of the data of a plurality of lines that are sequentially input, the weighted average value gradually converges each time data of one line is input. By using the converged average value for correction, the image noise of the image correction reference data is reduced, and the image data is accurately normalized.

Preferably, the initial value written by the initial value writing means is one line of reference creation data which is read first. The initial value may be an average value obtained before.

[0012]

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A digital color copying machine according to an embodiment of the present invention will be described below in the following order with reference to the accompanying drawings.

(A) Configuration of digital color copier (b) Image signal processing unit (c) Shading correction (d) HVC conversion (d-1) R, G, B reading position correction (d-2) Saturation and Hue separation (d-3) Circuit of HVC converter (e) Density converter (f) Area discriminator (f-1) Circuit of area discriminator (f-2) Under color removal / black printing automatic control ( f-3) Automatic edge emphasis control (f-4) Judgment of achromatic color edge part (g) Color correction (g-1) UCR / BP automatic processing (g-2) Masking processing (g-3) UCR / BP automatic Processing example (g-4) Mono color mode and color change (g-5) Border mode (g-6) Circuit configuration of color correction unit (h) MTF correction (h-1) Circuit configuration of MTF correction unit (h -2) Smoothing processing (h-3) Edge enhancement (h-4) Edge enhancement processing and sharpness mode (h-5) Color blur correction (h-6) Outline extraction mode (h-7) Example of automatic MTF correction (a) Configuration of Digital Color Copier FIG. 1 shows a digital color copying machine according to an embodiment of the present invention. 1 shows an overall configuration of a copying machine. The digital color copying machine is roughly divided into an image reader section 100 for reading a document image and a main body section 200 for reproducing an image read by the image reader section.

In FIG. 1, a scanner 10 includes an exposure lamp 12 for irradiating a document, a rod lens array 13 for condensing light reflected from the document, and a contact type CCD for converting the condensed light into an electric signal. The color image sensor 14 is provided. The scanner 10 uses the motor 1 when reading a document.
1, moves in the direction of the arrow (sub-scanning direction), and scans the original placed on the platen 15. The image of the document surface irradiated by the exposure lamp 12 is photoelectrically converted by the image sensor 14. The multi-valued electrical signals of the three colors R, G, and B obtained by the image sensor 14 are read by the read signal processing unit 20 into yellow (Y), magenta (M), and cyan.
(C) or black (K). Next, the print head unit 31 corrects the input gradation data according to the gradation characteristics of this photoconductor (γ
After performing correction and dither processing as necessary, the corrected image data is D / A-converted to generate a laser diode drive signal, and the drive signal generates a laser diode 221 (see FIG. (Not shown).

Laser diode 2 corresponding to gradation data
The laser beam generated from 21 exposes the photosensitive drum 41, which is driven to rotate, via a reflecting mirror 37, as shown by the dashed line in FIG. As a result, an image of the document is formed on the photoconductor of the photoconductor drum 41. Photoconductor drum 41
Before the exposure for each copy
And is charged by the charging charger 43. When exposure is performed in this uniformly charged state, an electrostatic latent image is formed on the photosensitive drum 41. Only one of the yellow, magenta, cyan, and black toner developing units 45a to 45d is selected to develop the electrostatic latent image on the photosensitive drum 41. The developed image is transferred by the transfer charger 46 onto a copy paper wound around the transfer drum 51.

The above printing process is repeated for yellow, magenta, cyan and black. At this time, the scanner 10 repeats the scanning operation in synchronization with the operations of the photosensitive drum 41 and the transfer drum 51. Thereafter, the copy paper is separated from the transfer drum 51 by operating the separation claw 47, is fixed through the fixing device 48, and is discharged to the discharge tray 49. The copy paper is fed from a paper cassette 50, and the leading end thereof is chucked by a chucking mechanism 52 on the transfer drum 51, so that no positional deviation occurs during transfer.

FIGS. 2 and 3 are block diagrams showing the entire control system of the digital color copying machine.

The image reader unit 100 is controlled by an image reader control unit 101. The image reader control unit 101 receives a drive I / O based on a position signal from a position detection switch 102 indicating the position of the document on the platen 15.
An exposure lamp 12 is controlled via a reference numeral 0103, and a scan motor driver 105 is controlled via a drum I / O 103 and a parallel I / O 104. The scan motor 11 is driven by a scan motor driver 105.

On the other hand, the image reader control unit 101 is connected to the image control unit 106 by a bus. Image control unit 1
Reference numeral 06 is connected to each of the CCD color image sensor 14 and the image signal processing unit 20 via a bus. An image signal from the CCD color image sensor 14 is input to an image signal processing unit 20 described later and processed.

The main unit 200 includes a printer control unit 201 for controlling the copying operation in general and a print head control unit 202 for controlling the print head unit 31.
Analog signals from the various sensors 44, 60, and 203 to 205 for automatic density control are input to the printer control unit 201. In addition, the printer control unit 2 is input via the parallel I / O 207 by key input to the operation panel 206.
01 is input with various data. Printer control unit 201
Is a control ROM 208 in which a control program is stored.
And a data ROM 209 storing various data. The printer control unit 201 includes various sensors 44 and 6
0, 203 to 205, operation panel 206 and data R
The data from the OM 209 controls the copy control unit 210 and the display panel 211 in accordance with the contents of the control ROM 208. Further, in order to perform automatic density compensation control, the grid of the charger is connected via the parallel I / O 212 and the drive I / O 213. High voltage unit 214 for voltage generation
And the high-voltage unit 215 for generating a developing device bias voltage.

The print head control unit 202 controls the control RO.
It operates according to the control program stored in the M216, is connected to the image signal processing unit 20 of the image reader unit 100 by an image data bus, and operates based on an image signal input through the image data bus. The gamma correction is performed with reference to the contents of the data ROM 217 storing the gamma correction conversion table. Further, when the multi-valued dither method is used as the gradation expression method, dither processing is performed, and the drive I / The laser diode driver 220 is controlled via O218 and parallel I / O219. The laser diode 221 is a laser diode driver 22
The light emission is controlled by 0.

The print head control unit 202 is connected to the printer control unit 201, the image signal processing unit 20, and the image reader control unit 101 via a bus, and is synchronized with each other.

FIG. 4 is a perspective view of the reading unit. In the reading unit, a light source (halogen lamp) 12 having a spectral distribution of three wavelengths (R, G, B) irradiates the surface of the original 91 with a light source (halogen lamp) 12, and the reflected light is transmitted by a rod lens array 13 to a contact type CC.
An image is formed on the light receiving surface of the D color image sensor 14 in a line shape at the same magnification. An optical system including the rod lens array 13, the light source 12, and the CCD color image sensor 14 includes:
The line information is scanned in the direction of the arrow in FIG.

(B) Configuration of Image Signal Processing Unit FIG. 5 is a block diagram of the image signal processing unit 20. Referring to FIG. 5, a print head is connected from the CCD color image sensor 14 via the image signal processing unit 20. Control unit 202
Will be described.

In the image signal processing section 20, the image signal photoelectrically converted by the CCD color image sensor 14 is converted by an A / D converter 61 into three primary colors R, G, and R of red, green and blue.
Multivalued digital image data r _{7 to 0} of B, it is transformed g _{7 to 0,} the b _{7 to 0.} The converted image data is normalized by shading correction in a shading correction unit 62. Next, the image data subjected to the shading correction is converted into data _V7-0 , _W7-0 , and _H7-0 of lightness V, saturation W, and color tone H in the HVC conversion unit 64. The converted data is output to the HVC line memory interface 66.

On the other hand, the image data RS _7-0 , GS _7-0 , B
S _{7 to 0} are the reflection data of the document 91, the density data DR _{7 to 0} of the real image by performing a log conversion in the density conversion unit 68, DG _{7 to 0,} is converted to DB _{7 to 0,} The data is output to the color correction unit 72 and the RGB line memory interface 70.

On the other hand, the data V _7-0 , W _7-0 , and H _7-0 outputted from the HVC conversion unit 64 are respectively discriminated from the area and color by the area discrimination unit 74, and the UCR / BP is determined.
The ratio is output to the color correction unit 72, and the MTF data and the control parameter NEDG (hereinafter, a signal represented by a signal name beginning with N means a negative logic signal) are output to the MTF correction unit 78. . Further, these data V, W, H
The color is determined by the color determining unit 76, and the signal NCCS is output to the color correcting unit 72.

Then, in the color correction section 72, black data generation processing and masking processing are performed simultaneously. That is, black data is generated, subtracted from the read data DR, DG, and DB, and the read data is converted into data of three reproduction colors of cyan C, magenta M, and yellow Y.

Further, in the MTF correction section 78, a digital filter is selected in accordance with the signal from the area determination section 74, and optimal smoothing (smoothing) processing or edge enhancement processing is performed. In addition, color blur correction and contour extraction are performed.

Next, the magnification / movement unit 80 changes the magnification. Further, a color balance unit (γ correction unit) 82
, The color balance is adjusted, and data is output to the print head control unit 202.

FIG. 6 shows a timing control section 84 for generating a timing signal for the image signal processing section 20. The timing controller 84 generates a CCD horizontal drive signal of the CCD sensor 14, a CCD vertical drive signal, a digitized timing signal of the A / D converter 61, and an image processing reference signal.

FIG. 7 is a circuit diagram of the HVC line memory interface 66. Image data V _{7 to 0} is output from the HVC converting unit 64, W _{7 to 0,} H _{7 to 0} and brightness average amount V _E, the selector 100, H via the bidirectional buffer 102, V, C line memory 104 Is stored in
CPU 140 that controls image signal processing unit 20 via bidirectional buffer 102 and bus gate 106 (see FIG. 9)
Is read. Note that, as shown in Table 1, H, V, and C are output through the address counter 108 and the selector 110.
The type and address of write data to the line memory 104 are controlled.

[0034]

[Table 1]

FIG. 8 is a circuit diagram of the RGB line memory interface 70. The image data DR _7-0 , DG _7-0 , and DB _7-0 output from the density converter 68 are stored in the RGB line memory 124 via the selector 120 and the bidirectional buffer 122, and are stored in the bidirectional buffer 12
2. The data is read by the CPU 140 (see FIG. 9) via the bus gate 126. Note that, as shown in Table 2, R, G, B
The type and address of write data to the line memory 124 are controlled.

[0036]

[Table 2]

FIG. 9 is a circuit diagram of a CPU peripheral circuit.
The CPU 140 has a two-line memory interface 6
The image data is monitored via 6 and 70 to detect the document size, detect a system abnormality, and automatically adjust the signal of the CCD sensor 14.

Further, a parameter signal is set according to the reading mode and the editing mode of the image processing. That is, through the decoder 142, the three parallel IO circuits 144
Outputs various parameter signals to each component of the image signal unit 20.

(C) Shading Correction The shading correction section 62 corrects the black level and the white level of the image data read by the CCD sensor 14 and normalizes the image data by the following equation.

DOUT = (DIN−BK) · 255 / WH Here, DIN is original read data r, g, b,
WH is read data of the shading reference white plate, and BK is black level data of the CCD sensor 14. This correction is performed independently for each color data r, g, b.

The BK level correction is performed to correct the difference in the background sensitivity level of each dot of the CCD sensor 14. For this reason, in the absence of incident light, the read data r, g, and b of the three primary colors of the CCD sensor 14 are stored in the FIFO memory 162, which is a line memory for one line, and are compared with the original read data DIN. I do. This processing also serves to remove offset on the digital signal.

The WH level is corrected by the CCD sensor 1
This is performed in order to remove unevenness in the main scanning due to the non-uniformity of sensitivity 4 and the spectral distribution of an optical system such as a lamp. For this reason,
The read data WH of the shading reference white plate is FIFO
The data is stored in the memory 182, and the reciprocal thereof is multiplied by the document read data for correction. This processing is performed by the CCD sensor 14.
Has the function of normalizing the three primary color data of the output, and also serves as white balance correction in which each R, G, B data has a fixed ratio with respect to a white document.

One problem in the shading correction is that, when the shading correction reference data is determined based on the sampling data of one line, if the sampling data contains image noise, the same on the corrected image data will be obtained. That is, image noise always appears at a pixel position.

There are various factors causing image noise, such as power supply noise in a system and crosstalk noise in a clock system, and the S / N ratio is deteriorated with respect to an image signal before A / D conversion. This means that the dynamic range (reference potential) of the A / D conversion is about 2.39 V,
If quantized to bits, there is a point of 1 LSB ≒ 9.3 mV.

Therefore, the correction reference data BK and WH are determined based on the sampling data for a plurality of lines, so that image noise can be reduced.

For this reason, it is conceivable to determine the correction reference data by preparing a line memory for a plurality of lines.
In the present invention, this correction is performed by using a line memory for one line to obtain a cumulative average of data of a plurality of lines.

In this embodiment, as shown in Tables 3 and 4,
There are four modes for BK level correction and WH level correction, namely, an initial value mode, a data generation mode, a holding mode, and a correction mode. These modes are BK
For the level correction and the WH level correction,
The selection is made by the mode selection signals SH ₀ to ₃ output from the PU 140.

[0048]

[Table 3]

[0049]

[Table 4]

Here, the initial value mode is a mode in which initial values are input to FIFO memories 162 and 182 which are line memories for one line.

In the data generation mode, the data in the FIFO memory and the reference data sequentially input are weighted and averaged.
This is a mode in which correction reference data is generated with high accuracy by cumulative averaging.

The correction mode is a mode in which the correction reference data is output while the correction reference data obtained in the data generation mode is held, and the BK level and the WH level are corrected for the original image signal. .

In the holding mode, no correction is performed, but the generated correction reference data is held. It is used to always access the memory to use the line memory having the DRAM structure.

Next, the circuit of the shading correction section 62 will be described. Shading correction is performed for a plurality of lines using a memory for one line. FIG.
FIG. 4 is a circuit diagram of a shading correction unit 62.

The image signals DIN _7-0 are _supplied to the first selector 16
The signal is input to a BK level FIFO memory 162 having a DRAM structure via a 0 terminal C (for an initial value mode). Further, the output of the FIFO memory 162 is the first selector 1
It is input to the A terminal 60 (for holding mode and correction mode). Further, the image signals _DIN7-0 are multiplied by 1/4 in the multiplier 164 and output to the adder 166, and the output of the FIFO memory 162 is multiplied by 3/4 in the multiplier 168 and output to the adder 166. , Adder 166 outputs the sum (weighted average) of the two to terminal B (for data generation mode) of first selector 160. Image signal DIN
_{7 to 0} are input to the subtractor 172, while the third selector 170 outputs the output of the FIFO memory 162 or “00”.
(Only in the case other than the correction mode) is selected, and the subtractor 17 is selected.
Output to 0. The subtractor 172 outputs the image signal (DIN-B) corrected at the black level BK only in the correction mode.
K) is output to the second selector 180 and the multiplier 192.

The image signal (DI
N-BK) is connected to the WH level FIFO of the DRAM structure via the C terminal (for the initial value mode) of the third selector 180.
Input to the O memory 182. Further, the output of the FIFO memory 182 is input to the A terminal (for holding mode and correction mode) of the third selector 180. This image signal is multiplied by に お い て in a multiplier 184 and added to an adder 18.
6, the output of the FIFO memory 182 is multiplied by に お い て in the multiplier 188 and output to the adder 186. Then, the adder 186 outputs the sum (weighted average) of the two to the terminal B (for data generation mode) of the third selector 180. The image signal corrected at the black level BK is input to the multiplier 192, while the fourth selector 19
A value of 0 selects the output of the FIFO memory 182 or “FF” (in a case other than the correction mode) and outputs the value of the WH level to the reciprocal table 194.
Outputs the reciprocal (1 / WH) to the multiplier 192.
The multiplier 192 multiplies the reciprocal by the image signal (DIN-BK) corrected at the black level BK to obtain a black level BK.
And a gradation signal DOUT corrected by the white level WH.

Each selector 160, 170, 180, 19
The selection of 0 is made by the mode selection signals SH ₀ , ₁ , SH ₃ , ₂ (see Tables 3 and 4). In black level correction, in the initial value mode (SH ₀ , ₁ = 0), the first selector 160
Selects the input at the C terminal, and the third selector 170
Input ("00") at terminal B is selected. Therefore,
The image signals _DIN7-0 are stored in the FIFO memory 162 as they are.
Is stored as an initial value.

In the data generation mode (SH ₀ , ₁ = 1),
The first selector 160 selects the input at the B terminal, and
The selector 170 selects the input (“00”) at the B terminal. Accordingly, the averaged image data of the image data of the plurality of lines that have been weighted and averaged is stored in the FIFO memory 162 as the reference data for correction. In this embodiment, the reference data sequentially input the data in the FIFO memory 162 is input as an initial value _(DIN 7~0 = BK 17~10)
Are weighted and averaged at a ratio of 3: 1. Therefore, the black level theoretically converges in 16 lines.

In the correction mode (SH ₀ , ₁ = 3), the first selector 160 selects the input at the A terminal, and the third selector 180 selects the input (black level value) at the A terminal. . That is, while the FIFO memory 162 holds the correction reference data, the subtracter 172 performs black level correction on the document image signal.

In the holding mode (SH ₀ , ₁ = 2), the first selector 160 selects the input at the A terminal, and the third selector 170 selects the input (“00”) at the B terminal.
Therefore, the correction reference data is stored in the FIFO memory 16.
2, but the subtractor 172 does not perform the correction.

The operation in each mode in the white level correction is performed in the same manner.

In this circuit, an FIF is used as a line memory.
Since the O memories 162 and 182 are used, writing and reading can be performed independently.

The weighted average of the data of a plurality of lines is nn
(N = 1, 2,...), The multipliers 164, 168,
The circuits of 184 and 188 are simplified.

FIG. 11 is a diagram showing the black correction block of the shading correction section 62 in detail. FIG.
In FIG. 1, the circuit of the flip-flop 174 and the FIFO memory 162 for adjusting the timing is also shown in more detail. Here, the clock is synchronized using the flip-flop 174, and when the input / output data is weighted averaged and normalized for the line memory 162 (for example, the signal WE is delayed by four dots by four flip-flops). Adjustment) so that the pixel position does not shift.

FIGS. 12 and 13 show (a) of FIG.
6 is a timing chart showing an operation at points (i) to (i).
Here, a ₁ , a ₂ , a ₃ ,... Are sequentially inputted reference data, and f ₁ , f ₂ , f ₃ ,.
This is an output value of 0. (C) is a weighted average value, and (i) is a correction value output from the subtractor 170. As a modified example, the average value of the correction reference data for one line is stored in a line memory in advance, and the CPU
May read this and set it as an initial value. In the flow described later, in the second and subsequent shading corrections, the previous correction value (that is, the average value) is used as an initial value. Even if this initial value is used, it takes a little time to converge, but the same effect can be obtained.

The above-described shading correction circuit may be provided for each color separation data of the input document information.
Thereby, red, green, and blue can be corrected independently.

The reference data may be corrected only for the white side or the black side by this method. For example, in the case of a monochrome image, only the white side needs to be corrected.

In the above-described shading correction, both the BK level and the WH level are independently corrected, and each correction data is created by cumulatively averaging data of a plurality of lines. This allows
High-precision correction is possible. However, in this case, the time required to generate the correction data becomes longer than before due to the addition of the BK level correction data and the addition of the cumulative averaging process. In particular, during multi-scanning (four full-color C, M, Y, and K scans, and multiple copies), the time required for correction is long, and the copy operation is affected by the rising characteristics of the exposure lamp. Would be an obstacle.

Therefore, in a modified embodiment described below, the time is shortened by performing only the cumulative averaging process of the correction data on the WH level side in the multi-scan mode. The reason why the BK level is not updated by using the held value obtained by the initialization at the time of turning on the power for the BK level is that the BK level side is not influenced by the environmental conditions and the component variation of the CCD sensor 14 is mainly affected. It is due to the cause. In contrast, W
On the H level side, turn on / off the exposure lamp during multi-scan.
Since the sensitivity of the CCD sensor 14 slowly changes due to an increase in the ambient temperature accompanying the turning off, fogging or the like occurs, and thus correction is necessary. By performing the correction in this manner, the correction data on the WH level side is updated for each scan, and a change in sensitivity is suppressed. On the other hand, correction data that is cumulatively averaged is always used, so that shading correction can be performed with high accuracy. .

FIGS. 14 and 15 show the flow of shading correction. A corresponding shading correction sequence is shown in FIG.

In this flow, when the power is turned on, the BK level correction and the WH level correction are set to the holding mode (step S2). Next, the CCD sensor 1
4 is driven (step S4), the BK level correction is set to the initial value mode, and the initial value is input to the FIFO memory (step S6). Next, regarding the BK level correction,
The mode is set to the data generation mode (step S8), and the sequentially input reference data is weighted and averaged to obtain the FIFO memory 162.
Enter again. When the average value is generated, next, the correction mode is set for the BK level correction (step S10), and the correction reference data is output for the WH level correction. next,
The exposure lamp is turned on to irradiate the reference white plate for shading correction (step S12). Next, the mode is set to the initial value mode for the WH level correction, the reference white plate is read, and input to the FIFO memory 182 as the initial value (step S14). Further, for the WH level correction, the mode is set to the data generation mode (step S16), and the sequentially input reference data is weighted and averaged to obtain the FIFO memory 182.
Enter again. When the average value is generated, the BK level correction and the WH level correction are set to the holding mode (step S
18) The correction value is held. Then, the exposure lamp is turned off, the reading is completed, the driving of the CCD sensor 14 is stopped (step S20), and the process stands by.

Next, when copying is performed, the previously determined initial value is used for the BK level. If there is a copy request (YES in step S22), the CCD sensor 14 is driven (step S24) to set the correction mode for the BK level correction (step S26), and the data from the FIFO memory 162 is used as the BK level correction reference data. Output. Then, the exposure lamp is turned on to read the shading reference white plate, and the FIFO memory 1 is read.
It inputs to 62 (step S28). Next, the data generation mode is set for the WH level correction (step S3).
0), weighted average of sequentially input reference data
Re-input to the IFO memory. The correction mode is set for the WH level correction (step S32), and data from the FIFO memory 182 is output as correction reference data. Then, the document is scanned (step S34) and read. At this time, shading correction is performed using the correction levels BK and WH. Then, the exposure lamp is turned off, and the reading is completed (step S36).

Next, it is determined whether or not a multi-scan is performed (step S38).

If it is a multi-scan, step S28
Then, the WH level is corrected again, and thereafter, the original is read.

If it is not a multi-scan, the driving of the CCD sensor 14 is stopped (step S40), the BK level correction and the WH level correction are set to the holding mode, and the correction values BK and WH are held. (Step S42). Then, the process returns to step S22 and waits for the next copy request.

As described above, during the multi-scan, the correction reference data is created in a sequence different from that at the time of power-on. However, as another sequence for shortening the correction reference data creation time, for example, the multi-scan May be used as the black level correction reference data held only in the second and subsequent scans.

(D) HVC converter The read data r, g, b of red, green, and blue are converted into hue H and saturation W for use in image data processing.

(D-1) R, G, B Reading Position Correction FIG. 17 schematically shows the CCD sensor 14. As shown in the figure, the CCD sensor 14 has red (R), green (G), and blue (B) pixels sequentially arranged in a line, and the pixels are inclined by 45 ° to prevent moiré. For this reason, at the edge portion of the image, the reading position differs depending on each of the R, G, and B color data, so that the color difference signals (WR, WB) cannot be accurately separated. To reduce this phenomenon, the reading positions of the R, G, and B pixels are corrected based on the G pixels using the following equation.

R _n = (5/8) r _n + (3/8) r _n-1 G _n = (3/4) g _n + (1/8) (g _n-1 + g _{n + 1} ) B _n = (5/8) b _n + (3/8) b _{n + 1} where n indicates a pixel number.

The color separation signals R, G, B thus read
Are converted into H (hue), V (brightness), and C (saturation) data. This conversion is performed in the subsequent image identification processing (color change, UCR / BP ratio automatic control, HTF automatic control,
This is used to accurately perform achromatic edge determination. The relative luminous efficiency distribution has a high specific gravity for green, and is generally approximated by a ratio of R: G: B = 0.229: 0.587: 0.114 in a C light source and a 2 ° visual field. The brightness signal (V) is obtained from this distribution. Further, the color difference signal (H) is more W
B = B−V and WR = R−V. Therefore, V,
WB and WR are calculated by the following matrix operation.

[0081]

(Equation 1)

FIG. 18 shows read data of the white and black images shown on the upper side. R, G,
When the data of B is compared with the corrected data on the lower side, 3
It can be seen that the change positions of the colors match.
Thereby, the color blur described later is also improved.

(D-2) Separation of Saturation and Hue The two color difference signals WR and WB indicate the orthogonal coordinate axes of the hue plane in the color space. When this is converted to polar coordinates, the hues are well separated as shown in the Munsell color chart of FIG. 19, the length of the vector indicates the saturation (the saturation decreases toward the center), and the angle indicates the hue. Will show. Therefore,
The saturation signal [W] and the hue signal [H] are calculated by the following equations.

[0084] ^{_{W = (W R 2 + W}} B 2) 1/2 ( provided that when W ≧ 256, W = 255) H = (256/360) tan -1 (W R / W B) (d-3 20) Circuit of HVC conversion unit FIG. 20 is a block diagram of the HVC conversion unit 64. Position correction unit (table for performing the above-described reading position correction)
300 performs the above-described position correction operation on the basis of the read data r, g, and b, and performs three-color data RS, GS, and BS.
Convert to (Note that the reading position correction is not necessary depending on the type of the CCD sensor 14.) The converted data is output as it is as the lightness signal V, while the lightness / color difference separating unit (that is, the table for performing the operation of the formula 1) ) 3
02, as described above, R, G, and _B are converted into lightness V and W _R , WB. Signal W _R, W _B is further converted into the luminance signal W by the chroma extraction circuit 304, in the hue extraction circuit 306, it is converted into a hue signal H.

FIG. 21 is a circuit diagram of the saturation extracting circuit 304. Signals WR and WB are provided by absolute value circuits 320 and 3 respectively.
After being converted to an absolute value at 22, the squaring circuits 324 and 326
, And then added by an adding circuit 328.
Then, the added value is converted into a square root, that is, a saturation W by the square root table 330 and output.

Is performed. FIG. 22 shows the hue extraction circuit 3
It is a circuit diagram of 06. The signals WR and WB are converted to the hue signal H in the tan ^-1 table 340.

(E) Density Conversion Unit The density conversion unit 68 converts the output data of the CCD color image sensor 14 into an original density (OD) as viewed from the human eyes.
Is converted to have a linear characteristic with respect to. CCD
The output of the color image sensor 14 has a photoelectric conversion characteristic that is linear with respect to the incident intensity (= original reflectance OR). On the other hand, the document reflectance (OR) and the document density (OD) are-
There is a relationship of logOR = OD. Therefore, using the reflectance / density conversion table, the CCD color image sensor 1
4 is converted into linear characteristics.

FIG. 23 is a circuit diagram of the density converter 68. The red, green, and blue image data RS, GS, BS, and the brightness signal V are converted into density data RL, GL, BL, and VL, respectively, by the density conversion table 360.

After the density conversion, a negative / positive inversion circuit 362 for negative / positive inversion selection further outputs data D
Select whether to output negatively or positively for R, DG, and DB. The negative / positive inverting circuit 362 is a control signal NNE for designating whether to output a density signal as it is (B output) or to output an inverted output (A output) by an inverter.
Select by GA (A output at L). When the A output is selected, the input data is inverted. Further, the output of the negative / positive inversion circuit 362 and “00” are connected to the selector 364.
Is selected by the effective original pixel area signal NHD1. That is, outside the area where the original is read, the output is white regardless of the negative / positive output.

[00].

The brightness data V is converted to monocolor density data DV.

(F) Area discriminating section As will be described later, the color correcting section 72 controls the black reproducibility and the color saturation based on the UCR / BP ratio, and performs the border processing. Further, the MTF correction unit 78 performs edge enhancement processing. These processes must be controlled according to the nature of the image. Therefore, based on the data obtained by the HVC conversion, the region determination unit 74 automatically controls the UCR / BP ratio based on the saturation data, automatically controls the edge amount based on the brightness change amount, and performs special processing (color The control value is set and the area is determined for (bleeding correction), and the result is sent to the color correction unit 72 and the MTF correction unit 78.

FIG. 24 is a block diagram schematically showing a portion of the image signal processing section 20 related to area determination. The image data is converted into red, green, and blue density data DR,
In addition to the conversion into DG and DB, the HVC conversion unit 64 converts the brightness into brightness V and saturation W. The brightness signal V is used for edge detection by the edge detection unit, and the edge amount is used for determination by the edge determination unit. The result is sent to the color blur correction table 620. The edge amount is converted into MTFdataA in the MTF control table 412 and input to the multiplier 622.

The MIN value of the density data detected by the MIN detector is sent to the UCR / BP processor 72.
On the other hand, after the signal W is smoothed by the smoothing filter 430, the UCR / BP ratio is sent from the UCR / BP control table 432 to the UCR / BP processing unit 72 in accordance with the result. Using the UCR / BP ratio determined in this manner, the red, green, and blue density data DR, DG, and DB are converted into cyan M, magenta C, yellow Y, and black K signals in a masking processing unit.

On the other hand, the MIN value is sent to the achromatic color judging section 442, and is used for judging whether or not the smoothed saturation signal W is an achromatic color.
20.

In the color fringing correction table 620, the signals C, M, Y, and K by the Laplacian filter 610
The next differential value is corrected in accordance with the result of the achromatic color determination and the result of the edge determination, and the corrected value is multiplied by the output value of the MTF control table 412 and further smoothed signals C and
M, Y, and K are added by the adder 608 and output.

FIG. 25 is a block diagram schematically showing a portion related to area determination for automatic MTF control in the monochrome mode. The read components of red, green, and blue have edge components detected by the primary differential filters 400 and 402 and the absolute value detection circuits 404 and 406, and the values are converted to MTFdataA by the MTF control table 412.

(F-1) Circuit of Area Discriminating Unit FIGS. 26 and 27 are circuit diagrams of the area discriminating unit 74. The brightness data V from the density conversion unit 68 is divided into a primary differential filter 400 in the main scanning direction and a primary differential filter 40 in the sub-scanning direction.
After the edge component is detected by 2 and the absolute value is detected by the absolute value detection circuits 404 and 406, the respective absolute values are input to the averaging circuit 410, and the average value _VE is output. The average value V _E is converted into MTFdataA by MTF control table 412 together with the sharpness setting value SHARP _{5 to 3.}

The two absolute values are also calculated by the comparator 42
0, 422 and compared with the threshold value REF. If any of them is larger than the threshold value REF, the edge signal NED is output via the OR gate 424. When the EGEN1 signal is output, the edge signal NED is output through the AND gate 426 to NWAKU.
Output a signal. The saturation signal W is smoothed through a smoothing filter 430 and output as a signal WS.
Further, the UCR / BP control table 432
/ BPdata.

Further, the minimum value data MIN from the color correction section 72 is converted to a BK level (D) via a BK level reference table 440.
The signal WS smoothed by the smoothing filter 430 is compared with the comparator 432. When the BK level is higher, the signal NBK is output because it is a black edge. This signal NBK is obtained by adding the edge signal NED to the OR gate 424.
Is output from the AND gate 434 to another AND gate 436. And NW
When the AKU signal (border editing area), NCH signal (color change editing area), and NMONO signal (mono color editing area) are not output, the NED signal is output and the color mode signal NCMY / K signal is output. If so, AND gate 444, NAND gate 4
46, and outputs an achromatic edge determination signal NEDG via an AND gate 448.

BK level reference table 440
Determines the binarization level of WS as shown in FIG. 28 when the minimum value of DR, DG, DB detected by each correction unit is input. Due to the chromatic aberration of the Selfoc lens,
Since the focal depths of the incident R, G, and B are different, the color difference signals WR and WB are larger than usual in an achromatic original having a high spatial frequency. Therefore, the binarization level is controlled according to the minimum value MIN that is the black level output.

The above first-order differential filters 400 and 402
Has a configuration as shown in FIG. That is, data of four consecutive lines are stored in the line memories 414a and 414a.
14b, 414c, and 414d. Further, when the data of the fifth line is input, as shown in the figure, calculations are performed on the data of 5 × 5 = 25 pixels with respect to the primary differential filters 400 and 402. Here, in the filter 400 in the main scanning direction, the numerical values shown in the figure are multiplied only at both ends in the main scanning direction, and the results are added. As a result, a first derivative value (edge amount) V _{H in} the main scanning direction is obtained for the target pixel at the center of the 5 × 5 pixels. The same calculation is performed for the other filter 402 with respect to the edge amount V _{V in the} sub-scanning direction.

Further, the smoothing filter 430 for smoothing the W signal from the color correction section 72 has a configuration as shown in FIG. That is, data of three consecutive lines are sequentially stored in the line memories 434a, 434b, and 434c. Then, as shown in the drawing, an operation is performed on the 3 × 3 data on the smoothing filter 430 for these data. Here, the numerical values shown in the figure are multiplied, and the results are added. As a result, a smoothed value WS is obtained for the pixel of interest at the center of 3 × 3 = 9 pixels.

(F-2) Automatic Undercolor Removal / Black Ink Printing Control As described later, the color correction section 72 outputs the black data K ′
K ′ = MIN (DR, DG, DB) is detected. From the read density data DR, DG, DB of three colors, α ·
To create black data K by subtracting K ′, β · K ′
Is output as the K amount. Here, α is the UCR ratio and determines the black amount. β is the BP ratio and lowers the color data. The UCR ratio / BP ratio has an effect on the saturation of chromatic colors and the sharpness of achromatic colors.

The UCR ratio / BP ratio in the color correction section 72 has the following trade-off relationship with respect to color reproducibility. That is, the reproducibility of black is determined by the UCR ratio / BP ratio
If (-α / β) is increased, the image is reproduced with pure black K ′, so that it is improved. On the other hand, the chromaticness of the chromatic color is reduced due to an increase in the output ratio of K ′. UCR ratio / BP
As the ratio approaches 100%, the amplitudes of the image data DR, DG, and DB after the undercolor is removed become extremely small, so that signal errors cannot be ignored and chromatic image noise cannot be ignored. Therefore, by controlling the UCR ratio / BP ratio in accordance with the original saturation so that it is closer to 0% for a chromatic original and closer to 100% for an achromatic original, the sharpness of the achromatic color can be improved. It is possible to achieve both improvement and saturation of chromatic colors, and to perform ideal color reproduction processing.

However, in the case of a color image, if the UCR ratio / BP ratio is simply set for the black amount K ′ obtained from the read data, the image density changes with respect to changes in hue and saturation. Such a process does not always work well. For example, when changing from white to red, the edges may be enhanced, but when changing from red to cyan,
It is better not to emphasize the edges because the hue changes strangely at the edges. Skin color has a particularly large effect. Therefore, it is necessary to extract and control only the change in image brightness. Therefore, in this embodiment, UCR / BP processing is performed using the saturation data.

The saturation signal W is input to a 3 × 3 smoothing filter and smoothed in order to suppress an extreme data change at the edge of the image. Next, the smoothed saturation signal WS is converted into UCR / BP data DB ₇ to ₀ by the UCR / BP control table 432 as shown in FIG. That is, UCR / B
Pdata linearly changes from 0% to 100% with respect to the magnitude of the saturation signal WS except where the signal is close to 0.

In the color correction section 72, as described later,
UCR / BP automatic processing is performed together with masking processing.

(F-3) Automatic Edge Emphasis Control The sense of density of an image is expressed by -log with respect to the incident light intensity.
Has characteristics. However, since the edge portion of the image responds to the change in brightness (not the density), performing the edge enhancement process on the data after the density conversion has the following problems. That is, since the data at the edge portion of the character / thin line changes more smoothly than before the density conversion, sufficient edge enhancement is not performed. Further, since the data change on the high density side is large and is easily detected as an edge, even image noise is emphasized. This is because the slope of the -log characteristic is small on the low density side and large on the high density side.

In a full-color image, when the data after the color correction is processed, a hue change occurs due to edge enhancement processing.

Therefore, if the amount of change in the brightness component of the image is detected and the amount of edge enhancement of each of the C, M, Y, and K data at the subsequent stage is controlled in accordance with the detected amount, the above-described problem is reduced. You.

Therefore, first, the read data of R, G, and B are converted into lightness V, and the lightness data (V _7-0 ) is input to first-order differential filters 400 and 402 in both the main and sub scanning directions. The amount of change in each direction is extracted and converted into an absolute value, and then the average value ( _VE7-O ) is obtained. The result is converted to MTF data A (D ₇ to D ₀ ) in an MTF control table 412 as shown in FIG. Here, the signals SHARP5 to _SHARP5 are BANK signals of a table, and the amount of enhancement is made variable by a sharpness signal. In FIG. 32, when SHARP = “7”, it is 1.75 times, and when SHARP = “0”, 0.2 times.
An example is shown in which the magnification is five times. That is, the average value V _E of the absolute value of the edge amount, except for the close signal is zero linearly M
Converted to TFdataA.

FIG. 33 shows the brightness distribution when the original image shown on the upper side is read by the center line and the corresponding brightness distribution.
Output of next differential filter 502 and absolute value detection circuit 40
4 shows its absolute value. After the averaging process, the absolute value is calculated as MTFdat corresponding to an appropriate SHARP set value.
aA, and sent to the MTF correction unit 78, where the Laplacian filter 6 of the distribution of the density PD of C, M, and Y
Edge emphasis is performed by multiplication with the output of 00.

(F-4) Determination of Achromatic Edge In order to improve the reproducibility of black characters or fine black lines in a full-color image, it is preferable to prevent color fringing at the achromatic edge. That is, in the achromatic edge portion, if the C, M, and Y data are erased and only the K data is edge-emphasized, the color blur at the edge portion is eliminated. First, | V _H7-0 |, | V
_V7-0 | is binarized by the threshold data REF, and the presence / absence of an edge portion is determined by the comparators 420 and 422 in either the main or sub scanning direction (NED = "L").
Yes).

Second, the signal WS is binarized by comparing the output of the BK level reference table 440 with the comparator 442 to determine whether the signal WS is an achromatic color (NBK = "").
L ”for achromatic color).

Third, NED = "L" and NBK = ""
Only when L ”(achromatic color edge portion), the MTF correction unit 7
8 to perform special processing (color blur correction)
An achromatic edge determination signal NEDG via a D gate 444
= “L” is output.

The mono color editing area (NMON
O = “L”), color change editing area (NCH = “
L ") and the border editing area (NWAKU =" L "), the NAND gate 446, AND
The achromatic edge determination signal NEDG is canceled by the gate 448. (This is because the influence of the color blur correction in the editing area is not exerted.) As described later,
When the color separation data for reproduction is K by the color correction unit 72, M
The color fringing correction in the TF correction unit 78 performs a normal area emphasizing process, so that the achromatic edge determination signal NEDG is permitted only when the color signal NCMY / K signal is "L". (NED = “L” indicates an edge portion of an image, so NWAKU is output as a border edit area signal. Here, a signal NEGEN to the AND gate 426 is output.)
Reference numeral 1 denotes a border editing permission signal. (G) Color correction Cyan, magenta, yellow,
Each of the black color data C ', M', Y ', K' is created for each scan by a frame sequential method, and full color is reproduced by a total of four scans. Here, the reason why black printing is also performed is that even if cyan, magenta, and yellow are superimposed to reproduce black, it is difficult to reproduce clear black due to the influence of the spectral characteristics of each toner. Therefore, in this full-color copying machine,
By the subtractive color mixture method using the data Y ′, M ′, and C ′ and the black printing using the black data K ′, the reproducibility of black is improved and full color is realized.

(G-1) UCR / BP Automatic Processing The color correction section 72 calculates the black amount K from the red, green, and blue components DR, DG, and DB representing the brightness on the document as follows. DR, DG, DB obtained from the density converter 68 are R, G, B
Since the respective density data of the components are present, they correspond to the C, M, and Y components of cyan, magenta, and yellow, which are the complementary colors of R, G, and B, when read by the CCD sensor 14. Therefore, as shown in FIG. 34, the minimum value of DR, DG, DB is
Since C ', M', Y 'on the manuscript are color superimposed components,
Black data K 'may be used. Therefore, in the color correction unit 72,
Black data K '= MIN (DR, DG, DB) is detected.

When the reproduction color data C, M, and Y are created, the data K 'is used and the data of C', M ', and Y' are used as α.
When subtracting K ′ to create black data K, β · K ′
Is output as the K amount. Here, α is a U
Is the CR (under color removal) ratio, β is the BP (black printing) ratio, and these are, as already described in (f-2),
The values are set not directly from the red, green, and blue density data DR, DG, and DB but by the area determination unit 74 based on the saturation data WS obtained by the HVC conversion of these values (see FIG. 31).

Further, in order to improve the saturation on the low density side, before multiplying by α and β, certain levels d ₁ , d ₂ (K
The cut data) is subtracted from the MIN (DR, DG, DB), and the UCR / BP processing is performed.

(G-2) Masking Process Further, the color correction unit 72 determines the transmission characteristics of each of the filters R, G, B in the CCD color image sensor 14 and the reflection characteristics of each of the toners C, M, Y in the printer unit. Correction is performed so that the color reproducibility is matched to the ideal characteristic. Taking the G filter and the M toner as an example, as shown in the transmission characteristics of FIG. 35 and the reflection characteristics of FIG. 36, the characteristics of the G filter and the M toner are indicated by hatching compared to the ideal characteristics. There is a non-ideal wavelength region. Therefore, in order to perform this correction, together with the UCR / BP processing described above,
Linear correction is performed by the following masking equation.

That is, in order to reproduce full-color input data on an image, a masking operation is performed in the MTF correction unit. Masking coefficients (Ac, m, y, Bc, m, y, C
c, m, y) are set so that the average color difference is minimized over almost the entire color gamut. (Since printing is performed in a frame-sequential manner, this masking equation is executed line by line.

[0122]

(Equation 2)

K = β · {MIN (DR, DG, DB) −d ₂ } (g-3) Example of UCR / BP Automatic Processing FIG. 37 shows one line of R, B, G read for gray scale. FIG. 38 shows the data of FIG.
FIG. 11 is an enlarged view of a part of the result of performing the UCR / BP process on the data of FIG. In FIG. 37, all three colors of R, G, and B are output to the same extent. The gray scale gradually changes from black to white from left to right. However, since the saturation WS of the gray scale is 0, the UCR / BP ratio is 100%. As a result of the masking operation, the outputs of C, M, and Y become 0 as shown in FIG. 38, and the gray scale density change can be almost represented only by K. Therefore, the color reproducibility of the gray scale was improved.

FIG. 39 shows R, G, and B data for one line read for the image composed of white and black shown on the upper side. FIG. 40 shows the results of density conversion of these read data, and FIGS. 41 and 42 show the results of HVC conversion. As shown in FIG. 40, C,
Although the values of M and Y are large, the saturation of this image is small as shown in FIG. Therefore, the UCR / BP ratio is 10
The result of the UCR / BP process shown in FIG. 43 is small, and the output of C, M, and Y is small, and the image is almost represented only by K. Therefore, the black color reproducibility of the achromatic image is improved.

FIG. 44 shows R, G, and B data for one line read for the image composed of white and red shown on the upper side. FIG. 45 shows the results of density conversion of these read data, and FIGS. 46 and 4 show the results of HVC conversion.
It is shown in FIG. As shown in FIG. 45, C, M,
The value of Y is large, and not only the C value and K 'value are large in the red portion, but also the M value and Y value are not small. As shown in FIG. 47, the saturation in the red portion is large. UC in this example
In the R / BP processing, when saturation WS ≧ 85, UC
UCR / BP when R / BP ratio = 0 (%) and WS <85
The ratio was set to 100 (1-W / 85) (%). Therefore, in the result of the UCR / BP processing shown in FIG. 48, the K output is lost, and the image is represented by a chromatic color mainly composed of C.
Therefore, the vividness of the chromatic image is improved.

(G-4) Mono Color Mode and Color Change Mode The colors for reproducing the mono color data ( _DV7-0 ) are R,
G, B, C, M, Y, K can be selected. As shown in Table 5, the reproduction color is determined by the state (C, M, Y, K) of the color separation data to be transferred to the printer and the logic of the NWH signal, and the final monocolor data (MONO _{7 to 0} )
Becomes That is, 00 is selected by NWH = “L”,
When WH = "H", _DV7-0 is selected.

[0127]

[Table 5]

At this time, color change data can be selected at the same time, and full color data (PD _7-0 ) can also be selected.

These full-color data ( _PD7-0 ),
Mono-color data (MONO _7-0 ) and color change data are represented by signals NCHAN and NCC as shown in Table 6.
It is controlled by S, NMONO.

[0130]

[Table 6]

The selection of Tables 5 and 6 is performed by the color discriminating section 76 using a table (not shown).

(G-5) Bordering Mode In the bordering mode, an image is bordered as shown in FIG. With respect to the read data of the original image shown in the figure (capital letter A in Roman letters), the result of the first differentiation (edge amount) is obtained, and the absolute value is detected. An edge signal NED is output only at a portion where the absolute value exceeds a predetermined threshold value REF, and a corresponding border image is output. That is, an image having a finite width near the edge portion of the image is output, and the size of the original image in the border image is slightly reduced.

When NEDGN1 = "L", the edge signal N
ED is enabled and the border mode designation signal (NWAKU)
Is generated.

The outline colors are C, M, Y, K, R, G, B
Of the color separation data (C, M,
Y, K) as shown in Table 7, and the selector 532 determines the logic of the WCLR signal and the border edit permission signal N
EGN1, edge signal NED, edge mode designation signal N
Data is output as shown in Table 8 corresponding to WAKU.

[0135]

[Table 7]

[0136]

[Table 8]

(G-6) Circuit Configuration of Color Correction Unit FIGS. 50 and 51 are circuit diagrams of the color correction unit 72. The density signals DR, DG, and DB from the density conversion section 68 are first supplied to the minimum value detection circuit 50 in the undercolor removal / black printing section 500.
2, the minimum value MIN (DR, DG, DB) is detected. The minimum value is calculated by subtracting predetermined black cut data d in a subtractor 504,
The data is multiplied by the R / BP data (α / β) to become black data K. Further, in the subtracter 508, the black data K is subtracted from the density data DR, DG, and DB for output of the undercolor, and is output.

In the masking operation section 510, first,
The three color data DR, DG, and DB from the subtractor 508 are output to the masking data (Ac, m, y, B
c, m, y, Cc, m, y), and the result is added to the adder 514.
And cyan, magenta, and yellow data C,
M and Y are generated. Further, this data is supplied to the selector 516 by the color signal NCMY / K by the multiplier 50.
6 is selected as black data K and output as data PD.

Further, in the color change / mono color selection section 520, this data PD is supplied to the selector 522.
Is input to the A terminal. On the other hand, the brightness signal DV from the HVC conversion unit 64 and “00” are
In response to the signal NWH, the data is input to the B terminal of the selector 422 as monochrome data MONO. The color change data is input to the C terminal of the selector 522. In the selector 522, these signals are output from the signal NMON.
O, selected by NCH. The signal NCH is the signal NC
Output when both HN and NCCS are "L". In the border editing unit 530, the selector 532 outputs the input signals “00” and “
FF "is selected by the signals NWAKU and WCLR and output as signals _{PD17 to PD10} .

FIG. 52 is a circuit diagram of the register section 540. In this embodiment, the UCR / BP data (α _7-0 / β _7-0 ) and the masking data (Ac, m,
_{y 9~0, Bc, m, y} 9~0, Cc, m, y 9~0), black cutting data (d _{7 to 0)} can be set two. Therefore, the signal NGCS
When 0 is output, the data is decoded by the decoder 542 corresponding to the parameters _MA3 to _MA0, and when the signal NWR is output, the data _{MD7 to 0} output from the CPU.
Is stored in the register 544. Then, these data are output to the selector 546 and the selection signal MPX
One masking data (Ac, m, y _9-0 , Bc,
_{m, y 9~0, Cc, m} , y 9~0), UCR / BP data _(α 7~0
/ Β _7-0 ) and black cut data (d _7-0 ) (first data).

The masking data (Ac, m,
_{y 9~0, Bc, m, y} 9~0, Cc, m, y 9~0), UCR / BP data (α 7~0 / β _7~0), black cut data (d _7~0)
Can be set in the register 44 by two types on the address map of the CPU as shown in Table 9. NGCS
When 0 = "L", data is set when the MWR rises from the "L" level. At this time, MPX1
When "L", the first data is selected, and when "H", the second data is selected.

[0142]

[Table 9]

The UCR / BP data is further processed by selector 538 at α _7-0 / β _7-0 and UC
And R _{7 to 0} is selected by the selection signal MPX0, is output. Here, the latter data UCR _7-0 is stored in the area discriminating unit 7
4 is the saturation data input from FIG.

The bits of the masking data and the UCR / BP data are defined as shown in Tables 10 and 11.

[0145]

[Table 10]

[0146]

[Table 11]

(H) MTF Correction Unit The MTF correction unit 78 performs smoothing processing, edge enhancement processing, and color blur correction. The configuration of the entire image data processing can be understood by the circuit in FIG.

(H-1) Circuit Configuration of MTF Correction Unit FIG. 53 is a circuit diagram of the MTF correction unit 78. The outline of the image data processing can be understood from the block diagram of FIG.

The signal PD from the color correction section 72 is smoothed by the two-dimensional two-dimensional FIR digital filters 600 and 602 for the central pixel and the peripheral pixels by the weighted average of the target pixel and the peripheral pixels. . The output signals of both the selector 60 and the unsmoothed signal are output.
4 is input. Then, the sharp signals SHARP1,0
Are selected and output. Thereby, reduction and smoothing of image noise are realized.

The signal PD from the color corrector 72 is input to the Laplacian filter 610. The Laplacian filter 610 is also called a second-order differential filter, and has a function of extracting an edge component of an input image.

The output of the selector 612 is a Laplacian table (color blur correction table) together with a signal NEDG for performing color blur correction (see (h-5)) and a selection signal MODE indicating a reading mode (photo / standard mode). ) 620, and converts the filter output D in accordance with the input value as shown in FIG. When NEDG = “L” and MODE = “H”, color fringing correction is performed.

The output D of the Laplacian table 620
Further, in the multiplier 622, the MTFdat for edge control created from the brightness data V by the area determination unit 74
The product of a and a is obtained and input to the adder 608 together with the smoothed data of the original image from the selector 604. This MTFdataA is an amount changed by the edge amount detected by the primary differential filter, as shown in FIG.

As a result, smoothed and edge-enhanced image data can be output based on the edge amount detected in the edge image portion by the Laplacian filter. That is, in the flat part of the image, the image is eliminated by smoothing, and in the edge image part, the density of the edge is prevented from dropping by the edge detection value by the Laplacian filter. The correction amount at this edge is 1 in the HVC converter.
The brightness is adjusted according to the magnitude of the edge of the brightness detected by the second derivative filter. A specific example of this automatic MTF correction is
This is described in (h-7).

The edge detection signal RAP output from the Laplacian filter 610 is compared with a threshold value REF in an edge level determination circuit 630 to generate an image outline extraction signal NRAP.

At the selector 632, the image data and the EDG
And NRAP = “L” and NEGEN
If 2 (outline extraction signal) = "L", EDG data (outline data) is selected.

FIG. 54 shows a register 640 for storing various control data and its peripheral circuits. Register 64
0, the data MD and the write signal NWR are input to the register from the CPU.
MTFdataB, REF, and EDG are stored corresponding to the signals decoded by GCS1 and MA1,0.
Note that MTFdataB and MTFda of the register 640 are used.
taA is selected by the selector 644 based on the MPX2 signal, and is output as MTFdata.

(H-2) Smoothing Processing As described above, the signal PD from the color correction section 72 is
Two types of two-dimensional FIR digital filters 600 and 60
2, smoothing is performed by the weighted average of the pixel of interest and its surrounding pixels. Both output signals are input to the selector 604 together with the signal that is not smoothed. And
Either is selected according to the sharp signals SHARP ₁ , ₀ and output to the adder 608. Thereby, reduction and smoothing of image noise are realized.

FIG. 55 shows this smoothing processing more specifically. That is, data of four consecutive lines are sequentially stored in the line memories 606a, 606b, 606c, and 606d. Further, when the data of the fifth line is input, the first smoothing filter 600 performs an operation on the data of 5 × 5 = 25 pixels as shown in FIG. The calculation is performed in the 2 smoothing filter 602. In the calculation, the data is multiplied by the numerical value shown in the figure, and the result is added. As a result, a smoothed value is obtained for the pixel of interest. Incidentally, Table 12 shows the selection of the output of the selector 604 by the selection signals SHARP _{1, 0.}

[0159]

[Table 12]

(H-3) Edge Emphasis The signal PD from the color correction section 72 is also input to the Laplacian filter 610. Laplacian filter 61
0 is also called a second derivative filter, and has a function of extracting an edge component of the input image. As will be described later, when the processing result of this filter and the input image are added, sharpening (edge enhancement) of the image can be performed.

FIG. 56 shows the Laplacian filter 610 specifically. This filter is a 5 × 5 filter. In the calculation, the numerical value shown in the figure is multiplied by the corresponding data, and the result is added.

The output result of the Laplacian filter 610 indicates that when edge enhancement is not required,
Is forcibly cleared by selecting "00" with the NSHARP3 signal at

(H-4) Edge Enhancement Processing and Sharpness Mode The output of the selector 612 is a Laplacian table 620 together with a signal NEDG for performing color blur correction and a selection signal MODE indicating a reading mode (photo / standard mode).
And converts the filter output D according to the input value, as shown in FIG.

As shown in FIG. 57, the photograph mode (MOD
At E = “L”), the output D is proportional to the input A. On the other hand, in the standard mode (MODE = "H"), NEDG
When "L" (black edge enhancement), the output D is always -64.
It is. When NEDG = “H”, ordinary edge enhancement is performed, and when the input A is large, the output D is constant when the input A is small.

The output D of the Laplacian table 620
Further, in the multiplier 622, the MTFdat for edge control created from the brightness data V by the area determination unit 74
The product of a and a is obtained and input to the adder 608 together with the smoothed data of the original image from the selector 604. As a result, the smoothed data and the edge amount are added, and edge-enhanced image data can be output.

FIG. 58 shows the density conversion data and the result of the Laplacian filter 610 for the image shown on the upper side. When the two are added in the adder 608,
As shown on the lower side, the value of the edge portion of the image becomes larger than the actual value, and edge enhancement can be performed.

As shown in Table 13, SHARP
_{Reference numerals 5 to 0 denote} a bank of the MTF control table of the area determination unit 74, selection of the smoothing filters 602 and 604 of the MTF correction unit, and ON / O of the output of the Laplacian filter 610.
The FF is controlled to correspond to a sharpness mode SHARP specified from the outside.

[0168]

[Table 13]

(H-5) Color Bleeding Correction The chromatic color reproduction data C, M, and Y at the black edge portion of the original image are erased by using the achromatic edge determination signal NEDG generated by the region determining unit 74, and the color blur is corrected. Perform bleed correction. As described above, the signal NEDG and the read mode signal MODE (“L” for the photograph mode and “H” for the standard mode) are input to the Laplacian table (color blur correction table) 620.

Here, when NEDG = “L” (achromatic color edge portion) and MODE = “H” (standard mode), color fringing correction is performed. At this time, no matter what the input value A is, the result of the multiplication with the MTF data in the multiplier 622 at the next stage is forced to be a negative value so that the output D = −64. Therefore, when PDs ₁₇ to ₁₀ are C, M, and Y data (NCMY / K = “L”), PD
_Since -64 * (MTF data) is subtracted from ₁₇ to 10, the C, M, and Y data at the black edge portion is erased, and color blur is prevented.

Further, the intensity of the Laplacian filter is selected by the Laplacian table 620 according to the logic of the MODE signal, and the edge enhancement is made stronger in the standard mode than in the photograph mode.

Next, an example of color fringing correction will be described. FIG.
When the above-described automatic UCR / BP processing is performed on the read density data of R, G, and B shown in FIG. 9 in the color correction unit 72, an output result as shown in FIG. 60 is obtained. (The UCR / BP ratio was processed at around 80%.) The horizontal axis represents the pixel number. Here, a color bleeding phenomenon occurs at the edge portion (see the portion 〇). That is, since C, M, and Y also have an edge portion at the edge portion of the K data, chromatic colors also overlap in addition to black, and the color blurs, and the edge is not reproduced clearly. Next, when edge enhancement is performed by MTF correction without performing color fringing correction in the MTF correction unit 78, the result is as shown in FIG. 61, and the color fringing becomes more severe (see the portion 〇). Therefore, when color blur correction is performed in the standard mode at the achromatic color edge portion by the NEDG signal and the MODE signal, as shown in FIG. (See section 〇).

(H-6) Contour Extraction Mode The edge detection signal RAP output from the Laplacian filter 610 is compared with a threshold value REF in an edge level determination circuit 630 to generate an image contour extraction signal NRAP. That is, first, the edge detection signal RAP is
If it is negative, it is converted to an absolute value; if it is positive, it is set to "00". Next, the value is compared with the threshold value REF to obtain 2
If the value is higher than a certain level, NRA = "L""
Is output to the selector 632 (see FIG. 63). In the selector 632, the image data and the EDG data are input,
NRAP = “L” and NEGEN2 (contour extraction signal)
If "L", EDG data (contour data) is selected.

FIG. 63 shows an example of contour extraction. Regarding the density conversion data of the original image in the figure (Roman capital letter A), the result of the second differentiation is obtained. If the result is negative, the result is converted into an absolute value. An NRAP signal is output only in a portion where the result exceeds a predetermined threshold value REF,
This shows the contour part.

REF _17-10 (binary level for contour extraction), EDG data (contour data) and MTF
dataB (edge emphasis control data) is set as shown in Table 14 on the address map of the CPU. This data is set when NWR rises from "L" when NGCS1 = "L".

[0176]

[Table 14]

[0177] For MTFdata, see Table 15
As shown in (1), the case where the edge enhancement is automatically controlled by the MTFdataA of the area determination unit 74 (photo / standard mode) and the case where the edge enhancement is manually controlled by the set value of the register (MTFdataB) (map mode) are divided.

[0178]

[Table 15]

Note that the bit definition of MTFdataB is
It is shown in Table 16.

[0180]

[Table 16]

In the map mode, MTFdataB = “80” in order to process with priority given to fine line reproducibility.
(1), MODE = "H" (standard mode), NBKEN
= "H" (no color blur correction).

(H-7) Example of Automatic MTF Correction FIGS. 64 to 68 show examples of automatic MTF correction for an original of 1 line / mm. FIG. 64 shows brightness data and its first derivative. The horizontal axis represents a pixel number. The brightness change corresponding to each line is obtained. FIG. 65 shows the density data and the result of the second derivative.

FIG. 66 shows an automatic M based on this density data.
The result of the TF correction is shown. Processing equivalent to edge enhancement is performed. A sharper boundary is better for an image of a line pair. Therefore, the reproducibility of the image is good.

For comparison, FIG. 67 shows the result of the smoothing process, but the reproducibility of the line pair is not good. Also,
FIG. 68 shows the result of the conventional edge enhancement. Data equivalent to the example is obtained.

FIGS. 69 to 72 show examples of automatic MTF correction for a halftone dot document (screen 133 lines). FIG. 69 shows the density data. The horizontal axis represents a pixel number. Moiré is seen.

FIG. 70 shows an automatic M based on this density data.
The result of performing TF correction is shown. Moiré is not emphasized.

For comparison, FIG. 71 shows the result of edge emphasis, in which moiré is emphasized. FIG.
Shows the result of the smoothing process. The density change is small.

FIGS. 73 to 77 show examples of automatic MTF correction for a black line original having a width of 1 mm. Figure 73
Indicates lightness data and its first derivative. A brightness change corresponding to the black line is obtained. FIG. 74 shows the density data and the result of its second derivative.

FIG. 75 shows an automatic M based on this density data.
The result of the TF correction is shown. In the flat part of the image, the smoothing effect is large, and in the edge part of the image, the inclination of the edge is sharp. Therefore, the reproducibility of the image of the black line is good.

For comparison, FIG. 76 shows the result of the smoothing process, in which the slope of the edge is gentle. FIG. 77 shows the result of edge enhancement. The change is large in the flat part of the image.

As described above, the automatic MTF correction of the present invention can achieve the optimum MTF correction for any image.

Although the MTF correction described above is performed in the full color mode, the same MTF correction can be performed in the mono color mode. In this case, FIG.
The circuit configuration shown in FIG. In the monochrome mode, the read data can be handled as brightness data, so that the configuration can be simplified.

[0193]

In the shading correction for correcting the normalization of the image reading value, the correction reference data can be created with higher precision than the reference creation data for a plurality of lines by using the line memory for one line.

[Brief description of the drawings]

FIG. 1 is a sectional view of a full-color copying machine.

FIG. 2 is a block diagram of a part of a control system.

FIG. 3 is a block diagram of another part of the control system.

FIG. 4 is a perspective view of a reading unit.

FIG. 5 is a block diagram of an image signal processing unit.

FIG. 6 is a diagram of a timing control unit.

FIG. 7 is a circuit diagram of an HVC line memory interface.

FIG. 8 is a circuit diagram of an RGB line memory interface.

FIG. 9 is a diagram of a CPU peripheral circuit.

FIG. 10 is a circuit diagram of a shading correction unit.

FIG. 11 is a diagram of a black correction block of a shading correction unit.

FIG. 12 is a timing chart at points (a) to (i) in FIG. 11;

FIG. 13 is a timing chart at points (a) to (i) in FIG. 11;

FIG. 14 is a part of a flowchart of shading correction.

FIG. 15 is a part of a flowchart of shading correction.

FIG. 16 is a diagram showing a shading correction sequence.

FIG. 17 is a diagram of a CCD sensor.

FIG. 18 is a diagram showing read data of an upper image and data after position correction.

FIG. 19 is a Munsell color chart.

FIG. 20 is a block diagram of an HVC conversion unit.

FIG. 21 is a block diagram of a saturation extraction circuit.

FIG. 22 is a circuit diagram of a hue extraction circuit.

FIG. 23 is a circuit diagram of a density conversion unit.

FIG. 24 is a simplified block diagram of a part related to area determination of an image signal processing unit.

FIG. 25 is a simplified block diagram of area determination for MTF automatic control in a monochrome mode.

FIG. 26 is a circuit diagram of a part of an area determining unit.

FIG. 27 is a circuit diagram of a part of an area determining unit.

FIG. 28 is a graph showing a relationship between a BK level (D) and a minimum value (MIN).

FIG. 29 is a diagram of a primary differential filter.

FIG. 30 is a diagram of a smoothing filter.

FIG. 31 is a diagram of conversion by a UCR / BP control table.

FIG. 32 is a diagram illustrating a change in an emphasis amount according to an MTF control table.

FIG. 33 is a diagram showing reading and processing of the image shown on the upper side.

FIG. 34 is a diagram of black data.

FIG. 35 is a graph showing characteristics of a G filter.

FIG. 36 is a graph showing characteristics of M toner.

FIG. 37 is a diagram of grayscale read data.

FIG. 38 is a table for automatically converting the data shown in FIG. 37 to an automatic UCR / BP.
It is a figure of one copy of the result of having processed.

FIG. 39 is a diagram of read data of an image composed of white and black by each element of the CCD sensor (the address is indicated on the horizontal axis).

FIG. 40 is a diagram obtained by converting the data of FIG. 39 into a density.

FIG. 41 is a diagram obtained by subjecting the data in FIG. 39 to HVC conversion.

42 is a diagram of the lightness edge and the saturation of the data in FIG. 41.

FIG. 43 is a diagram showing a result of the automatic UCR / BP processing of the data in FIG. 39;

FIG. 44 is a diagram of read data of an image composed of white and red by each element of the CCD sensor (address is indicated on the horizontal axis).

FIG. 45 is a diagram in which the data of FIG. 44 is converted into a density.

FIG. 46 is a diagram obtained by subjecting the data of FIG. 44 to HVC conversion.

47 is a diagram of lightness edges and chroma of the data in FIG. 46.

FIG. 48 is a diagram showing a result of the automatic UCR / BP processing of the data shown in FIG. 44;

FIG. 49 is a diagram illustrating data processing for bordering an image in the bordering mode.

FIG. 50 is a circuit diagram of a part of a color correction unit.

FIG. 51 is a circuit diagram of a part of a color correction unit.

FIG. 52 is a circuit diagram of a register unit.

FIG. 53 is a circuit diagram of an MTF correction unit.

FIG. 54 is a diagram of a register and its peripheral circuits.

FIG. 55 is a diagram of a smoothing filter.

FIG. 56 is a diagram of a Laplacian filter.

FIG. 57 is a diagram of processing by a Laplacian table.

FIG. 58 is a diagram showing edge enhancement processing by a Laplacian filter.

FIG. 59 is a diagram of read data of R, G, and B.

60 is a diagram showing a result of the automatic UCR / BP processing of the data in FIG. 59.

FIG. 61 is a diagram showing a result of the automatic MTF processing of the data in FIG. 59;

62 is a diagram showing a result of color fringing correction of the data in FIG. 59.

FIG. 63 is a diagram showing a contour extraction process in a contour extraction mode.

FIG. 64 is a diagram showing brightness data of a document of 1 line pair / mm and its first derivative.

65 is a diagram of the density data of the data of FIG. 64 and its first derivative.

FIG. 66 is a diagram showing a result of automatic MTF correction of the data in FIG. 64;

FIG. 67 is a diagram of a data smoothing process (comparative example) of FIG. 64;

68 is a diagram showing a result of conventional edge enhancement (comparative example) of the data of FIG. 64.

FIG. 69 is a diagram of density data of a halftone dot document.

70 is a diagram showing a result of automatic MTF correction of the data in FIG. 69.

71 is a diagram showing a result of edge enhancement processing (comparative example) of the data in FIG. 69.

FIG. 72 is a diagram showing a result of the data smoothing process (comparative example) of FIG. 69;

73 is a diagram showing the brightness of a 1 mm-wide black line original and its 1
It is a figure of the data of the next differentiation.

FIG. 74 is a diagram of the data of its second derivative of the density of the data of FIG. 73.

75 is a diagram showing the result of automatic MTF correction of the data in FIG. 74.

76 is a diagram of a result (comparative example) of the data smoothing process in FIG. 74.

77 is a diagram of the conventional edge enhancement (comparative example) of FIG. 74.

[Explanation of symbols]

65: region determination unit, 66: color correction processing unit, 67: M
TF correction section, 81: smoothing processing section, 82: under color removal / ink printing control section, 83: color correction masking control section, 84: edge detection section, 85: MTF correction control section,

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int. Cl. ⁷ , DB name) H04N 1/40-1/409 H04N 1/46 H04N 1/60

Claims (1)

(57) [Claims] 1. A reference plate, a line sensor that reads a reference plate or a document image line by line and outputs image data, and a line memory that stores image data of one line of the reference plate read by the line sensor. An initial value writing means for writing, as an initial value, one line of reference creation data to a line memory for each pixel; and, each time data of one line of a reference plate read by a line sensor is input, An arithmetic unit that weights and averages the read data of one line of the reference plate and each pixel, and inputs the result again to the line memory, and repeats this for a plurality of lines; a weighted average obtained by the arithmetic unit Correcting means for correcting one line of image data of the original read by the line sensor using the data. Power equipment.