JP2947823B2 - Image processing device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus having a function of determining characteristics of an input image.

2. Description of the Related Art Conventionally, a document image is read, an edge area including characters, line drawings, and the like in the read image is separated from a halftone area including a photograph, and appropriate processing is performed on each area. A method has been proposed (for example, published in 1987, pages 100 to 101 of Nikkei Electronics No. 425).
etc).

[Problems to be Solved by the Invention] However, according to the related art, when a document in which a halftone is expressed by a halftone dot is read and processed, the halftone dot region is determined to be an edge region. There was a problem of adding an unnecessary process.

SUMMARY An advantage of some aspects of the invention is to provide an image processing apparatus capable of accurately performing feature determination on an input image reproduced in halftone using halftone dots.

[Means for Solving the Problems] An image processing apparatus of the present invention for solving the problems has the following configuration. That is, first detection means for detecting a density change and a direction of the density change in the vicinity of a target pixel in the input image, and the density in a direction substantially orthogonal to the direction of the density change detected by the first detection means. Second detecting means for detecting the continuity of the pixel in which the change has been detected, and third detecting means for detecting a halftone dot of the image in the input image based on the density change detected by the first detecting means And determining means for determining the characteristics of the input image based on the detection results by the second and third detecting means.

Embodiment An embodiment according to the present invention will be described below in detail with reference to the accompanying drawings. In the embodiment, an example in which the present invention is applied to a full-color digital copying machine will be described.

<Description of Overall Configuration> FIG. 3 shows an overall configuration diagram (cross-sectional view) of a full-color digital copying machine.

The copying machine according to the embodiment includes an image scanner unit 201 that reads a document and performs digital signal processing, and a printer unit 202 that prints out an image corresponding to the document image read by the image scanner unit 201 on recording paper in full color. It is composed of

In the image scanner unit 201, reference numeral 200 denotes a mirror pressure plate, and a document 204 on a platen glass (hereinafter, referred to as a platen) 203 is irradiated with a lamp 205. The reflected light is sequentially reflected by mirrors 206, 207, and 208 and guided to a lens 209 to form an image on a three-line sensor (hereinafter referred to as a CCD) 210. Red (R) and green (G), which are color components of full-color information. , Blue (B) components to the signal processing unit 211. Note that the ramp 205 and the mirror 206 have a speed v
Numerals 207 and 208 scan the front surface of the original by mechanically moving at 1 / 2v in a direction perpendicular to the electrical scanning direction of the line sensor. The signal processing unit 211 electrically processes the read signal (for example, γ conversion processing, logarithmic processing, undercolor removal processing, etc.)
Magenta (M), cyan (C), yellow (Y),
It is decomposed into black (Bk) components and sent to the printer unit 202. Also, for one original scan in the image scanner unit 201, one component of M, C, Y, and Bk is output to the printer unit.
Since it is sent to 202, one printout is completed by a total of four document scans.

M, C, Y and Bk sent from the image scanner unit 201
Are sent to the laser driver 212. The laser driver 212 modulates and drives the semiconductor laser 213 according to the input image signal. The laser light is reflected by one side of the polygon mirror 214 rotating at a constant speed, and the f-θ lens 2
15, is swept over the photosensitive drum 217 via the mirror 216.

Reference numeral 218 denotes a rotary developing unit, which includes a magenta developing unit 219, a cyan developing unit 220, a yellow developing unit 221, and a black developing unit 222. The four developing units are in contact with the photosensitive drum 217 and are formed on the photosensitive drum 217. The developed electrostatic latent image is developed with a corresponding color toner.

Reference numeral 223 denotes a transfer drum, and the recording paper fed from the paper cassette 224 or 225 is wound around the transfer drum 223, and transfers the image developed on the photosensitive drum 217 to the recording paper.

In this manner, the four colors of M, C, Y, and Bk are sequentially brought into contact with the photosensitive drum 217, so that the respective color toners are transferred to the recording paper. When the development and transfer for the four colors are completed, the recording paper is separated from the transfer drum 223, passes through the fixing unit 226, and is discharged.

<Description of Image Scanner> FIG. 4 is a block diagram of the inside of the image scanner unit 201. In FIG. 4, reference numeral 101 denotes an address counter, and a main scanning address 102 for specifying a main scanning position of the CCD 210.
Is output. That is, when the horizontal synchronization signal HSYNC is "1", it is set to a predetermined value by a CPU (not shown) and incremented by the clock signal CLK of the pixel.

The image formed on the CCD 210 has three line sensors 30
The photoelectric conversion is performed at 1,302,303, and the R component, G
Components, as component B read signals, amplifiers 304, 305, 306,
Sample hold circuits 307, 308, 309 and A / D converters 310, 31
8-bit digital image signal 313 for each color through 1,312
(R), 314 (G), and 315 (B) are output.

<Explanation of Overall Flow of Image Processing> FIG. 5 shows an electrical configuration of the entire apparatus. Note that components common to FIG. 3 are denoted by the same reference numerals.

In the figure, CLK is a clock signal for transferring pixels, and HSY
NC is a synchronization signal for starting main scanning, that is, a horizontal synchronization signal. CLK4 is a clock signal for generating a 400-line screen described later. The timing charts of CLK, CLK4, and HSYNC are as shown in FIG. 6, and each is sent from the control unit 401 to the image scanner unit 201, the signal processing unit 211, and the printer unit 202.

The image scanner unit 201 reads the original 204 and converts the R, G, and B signals as electrical signals into a color signal processing unit in the signal processing unit 211.
402 and a feature extraction unit 403. In the feature extraction unit 403, the color processing control signal generation unit 404 outputs a BL signal indicating that the current processing pixel is a black image, a COL signal indicating that the image is a tinted image, and a black image. A UNK signal indicating that there is a possibility of an image having a color or a color, a CAN signal for canceling the BL signal, or an EDGE signal indicating that the image is an edge such as a character line drawing is sent.

The color processing control signal generation unit 404 receives the above-described signal from the feature extraction unit 403 and generates a color processing control signal for the color signal processing unit 402. These are two multiplication coefficient signals GAIN1 and GAIN2 for weighting two types of image signals, a FIL signal for switching a spatial filter, and a GAM signal for switching a plurality of density conversion characteristics.

The control unit 401 sends 2 bits to each processing block.
Is sent. This signal corresponds to the developed color of the printer unit, and the PHASE signals 0, 1, 2, and 3 represent the developed colors magenta (M), cyan (C), yellow (Y), and black (Bk), respectively. means.

The color signal processing unit 402 generates a recording image signal VID for the printer unit 202 based on the PHASE signal and the color processing control signal.
Generates EO.

Based on the VIDEO signal, the printer unit 202 performs pulse width modulation (PWM) on the emission time of the laser, and outputs a copy output 406 having a gray scale expression.

Here, the printer unit 202 includes a color processing control signal generation unit 404.
From the SCR signal is input. The printer unit 202
A plurality of pulse width modulation basic clocks (screen clocks) are switched in accordance with the SCR signal to perform the optimum density expression for the original.

In this embodiment, when the SCR signal is 0, pulse width modulation is performed in units of one pixel, and when the SCR (screen) signal (to be described in detail later) is "1", pulse width modulation is performed in units of two pixels.

<Description of Feature Extraction Unit> FIG. 8 is a block diagram of the inside of the feature extraction unit 403. Reference numeral 1101 denotes a pixel color determination unit, which generates a BLP signal indicating that each pixel is black, a COLP signal indicating that the color is colored, or an UNKP signal indicating whether it is unknown, and performs area processing. Send to section 1102. The area processing unit 1102 performs BLP, COLP, and ULP in a 5 × 5 area including the pixel of interest.
NKP and G signals are determined in the area to eliminate errors, and BL,
Generate COL, UNK and generate CAN signal.

Reference numeral 1103 denotes a character edge determining unit which determines whether or not the character edge is a character edge unit based on the G signal, and generates an EDGE signal. The reason for determining whether or not a character edge portion is a character edge portion only by the G signal is that the G signal is closest to the visibility characteristic among the R, G, and B signals as shown in FIG. , G signal can be represented by a character edge detection signal of a white / black image.

<Description of Pixel Color Determining Unit> FIG. 9 is a block diagram for determining the saturation of the pixel color determining unit 1101.

In the figure, 1301 is a MAX / MIN detector, 1302-1309 are selectors, and 1310-1315 are subtracters, which calculate AB for inputs A and B and output them. 1316 to 1323 are comparators, and comparators 1316 and 1319 have their input A and input B
When 2A> B, the comparators 1317, 1320, 1322, 1
323 is for A> B, and comparators 1317 and 1321 are for A
The logic level "1" is output when> 2B, and 0 otherwise. 1324-1328 is AND gate, 1329 is NOR
The gate 1330 is a NAND gate.

In the above configuration, the MAX / MIN (maximum / minimum) detector 130
For 1, the circuit shown in FIG. 10-1 was used. In FIG. 10-1, reference numerals 1350, 1351, and 1352 denote comparators, which output "1" when R> G, G> B, and B> R, respectively. And
The circuit shown in FIG. 10-1 composed of an inverter, an AND gate, and an OR gate eventually generates the following decision signals S00, S01, S02, S10, S11, S12 as shown in FIG. 10-2. generate. That is, when MAX is R or when R, G, and B are all equal, S00
= 1, S01 = S02 = 0, when MAX is G, S01 = 1, S00 = S02 = 0, when MAX is B, S02 = 1, S00 = S01 = 0, when MIN is R or R, G , B if all are equal
= 1, S11 = S12 = 0, when MIN is G, S11 = 1, S10 = S12 = 0, and when MIN is B, S12 = 1, S10 = S11 = 0.

For example, when MAX (maximum) is R, R> G and R ≧ B
Therefore, the comparator 1350 outputs “1” and the comparator 1352 outputs “0”. Therefore, AND1 outputs "1", and the output of OR1 becomes "1". Also, A
ND2 and AND3 output “0”. That is, S = 001, S01 = S
02 = 0. The result of the same determination is as shown in FIG. 10-2.

The outputs S00, S01, and S02 of the MAX / MIN detector 1301 are
, And outputs S10, S11, and S12 are input to selectors 1303-1309.

The selectors 1302-1309 are composed of an AND circuit and an OR circuit as shown in FIG. 11-1. According to this selector, the eleventh
As shown in FIG. 2, A is output when S0 = 1, S1 = S2 = 0 for inputs A, B, and C, and B is output when S1 = 1, S0 = S2 = 0, and S2 C is output when = 1, S0 = S1 = 0. In this embodiment, the inputs A, B, and C correspond to the R, G, and B signals, and S0, S1, and S2 have MAX /
S00, S01, S02 or S1 from MIN (maximum / minimum) detector 1301
0, S11, and S12 corresponded. As a result, for example, the maximum (MAX) of the input R, G, and B can be output from the selector 1302, and the minimum (MIN) can be output from the selector 1303.

The pixel color determination unit 1101 of this embodiment sets the maximum value of the R, G, and B signals to MAX, sets the minimum value to MIN,
As shown in FIG. 1, this is performed by dividing into four areas A, B, C, and D.

That is, in the achromatic region, the difference between MAX and MIN is small, and the closer to chromatic color, the larger the difference between MAX and MIN. Divides the MAX-MIN plane.

Specifically, Ka, Kb, Kc, ia, ib, ic, WMX, and WMN are defined as predetermined constants, and are divided into four areas A, B, C, and D as shown in the figure.

Here, the region A is a dark achromatic (black) region.
The condition that (MAX, MIN) is included in this area is that MIN ≦ WMN or MAX ≦ WMX, and that MAX−ka <2 · MIN MAX−kb <MIN MAX−kc <1/2 · MIN Is to meet.

The area B is an intermediate area between a dark achromatic color and a chromatic color. (MA
(X, MIN) is included in this area if MIN ≦ WMN or MAX ≦ WMX, and either MAX−ka ≧ 2 · MIN MAX−kb ≧ MIN MAX−kc ≧ 1/2 · 2MIN And MAX-ia <2 · MIN MAX-ib <MIN MAX-ic <1/2 · MIN.

Area C is a chromatic area. The condition that (MAX, MIN) is included in this area is that MIN ≦ WMN or MAX ≦ WMX and that either MAX-ia ≧ 2MIN MAX-ib ≧ MIN MAX-ic ≧ 1 / 2MIN is there.

D is a bright achromatic (white) area. (MAX, MIN)
Is included in this area is that all of MIN> WMN MAX> WMX are satisfied.

FIG. 12-2 shows output signals for each state of the areas A, B, C, and D. That is, BL1 = 1, UNK1 = COL = 0 when included in the area A, UNK1 = 1, BL1 = COL = 0 when included in the area B, COL1 = 1, when included in the area C BL1 = UNK = 0, and BL1 = 1, UNK1 = COL = 0, when included in the area D.

The above-mentioned area determination is performed by the circuits 1304 to 1330 in FIG. Selector 1302, depending on the output of MAX / MIN detector 1301,
1303 selects the MAX signal and the MIN signal from among R, G and B, respectively, and the selectors 1304-1309 also select the values of the constants ka, kb, kc, ia, ib and ic in conjunction with the selector 1303. . For example, when MAX is an R signal and MIN is a G signal, the selector 1304
KAG, 1305 is KBG, 1306 is KCG, 1307 is iAG, 1308 is iBG, 1
Reference numeral 309 selects iCG, and sets the values of constants ka, kb, kc, ia, ib, and ic, respectively. The reason why the minimum value changes the values of the constants ka, kb, kc, ia, ib, and ic according to any of R, G, and B is as follows.

In general, in the case of a full-color sensor, there is a deviation in the color balance unique to the sensor. Therefore, the determination of chromatic / achromatic colors with the same determination criterion for all colors may cause erroneous determination. Therefore, as shown in FIG. 13, the balance characteristic of the sensor is accommodated by dividing the three-dimensional space of RGB into three parts. That is, the three-dimensional space of RGB is divided into an area 5702 where MIN = R, an area 5703 where MIN = G, and an area 5704 where MIN = B, and ka, kb, kc, ia, and ib, ic
Is used.

For example, for a sensor in which the R component signal appears lower, the value of KAR, KBR, KCR, iAR, iBR, iCR in FIG. In the area shown in FIG.
It is possible to narrow the area, and it is possible to respond to various sensors finely.

 The subtractors 1310 to 1312 and the comparators 1316 to 1318 determine the magnitude relationship between MAX-ka, 2MINMAX-kb, MINMAX-kc, and 1 / 2MIN.

 The subtractors 1313 to 1315 and the comparators 1319 to 1321 determine the magnitude relationship between MAX-ia, 2MIN MAX-ib, MIN MAX-ic, and 1 / 2MIN.

 Comparators 1322 and 1323 determine the magnitude relationship between MAX, WMX MIN, and WMN, respectively.

From the above, the above area determination is performed, and the result is BL1, UNK
It is output as the judgment signal of 1, COL1.

<Description of Area Processing Unit> FIG. 14-1 is a block diagram of the area processing unit 1102 shown in FIG.

BL1, COL1, UNK1 determined by the pixel color determination unit 1101
The signal is output by the line memories 1701, 1702, 1703, and 1704 to 1
The data is delayed line by line, synchronized by the HSYNC signal and the CLK signal, and data for five lines are simultaneously output. Here, BL1, COL1, and UNK1 are delayed by one line,
2, COL2, UNK2. BL3, COL3, UNK3 are delayed by 2 lines, respectively, and BL4, COL are delayed by 3 lines.
4, UNK4, and 4 lines delayed for BL5, CO respectively
L5 and UNK5.

The counter means (composed of a counter or the like) 1705 counts the number of BL black pixels (BL) in the 5 × 5 area shown in FIG. Similarly, the number of chromatic pixels (COL) is counted by the counting means 1705, and N
Get C. Further, the comparator 1707 compares the number NB of black pixels and the number NC of chromatic pixels in a 5 × 5 block.

Further, the output BK of the pixel color determination unit for the center pixel (pixel of interest) of the 5 × 5 area through the gate circuits 1708 to 1715
3, calculated with the result of COL3, UNK3, a BL signal indicating that the center pixel is black, a COL signal indicating that the center pixel is chromatic, and UNK indicating that the center pixel is intermediate saturation
A signal is output.

The determination criterion at this time is the first criterion (the pixel color determination unit
If the determination result of 1101) is a black pixel and a chromatic pixel, the determination is not reversed. That is, BK3 = 1
Or, when COL3 = 1, BK = 1 or COL = 1. If the result of the first determination criterion is intermediate between a chromatic pixel and an achromatic pixel, the comparator 1716
It is determined whether the number of black pixels is equal to or more than a predetermined value (NBC), and
The comparator 1717 determines whether the number of chromatic pixels is equal to or greater than a predetermined value. Further, the comparator 1707 determines whether the number of black pixels or the number of chromatic pixels is larger. When the number of black pixels is equal to or more than a predetermined value and NB> NC, the gate 1708
UNK3 becomes BL.

When the number of chromatic pixels is equal to or more than a predetermined value and NB ≦ NC,
At gate 1709, UNK3 becomes COL.

This is to remove the unevenness of the scanning optical systems 206, 207, and 208 and the color blur at the changing point of the color of the document due to the error of the imaging optical system 209.

Gates 1713, 1714, and 1715 detect that there are no more than a predetermined number of black pixels and chromatic pixels around the UNK3 signal, and output an intermediate chroma signal UNK.

 FIG. 17-3 shows a CAN signal generator.

In the BL signal generation logic shown in FIG. 14-3, if the target pixel is a black pixel, the BL signal is output regardless of the surroundings. However, if there is the aforementioned scanning speed unevenness or an imaging magnification error, a black signal due to color fringing may be generated around the color signal as shown in FIG. 14-4. Since the black signal due to the color bleeding is generated around the color signal, the light amount value becomes larger than the color signal as shown in FIG. 14-5.

Therefore, the CAN signal generation unit detects whether there is a color signal (COL) having a smaller light amount value than the pixel of interest around the pixel of interest and generates a CAN signal.

Here, a G signal closest to the above-described visibility characteristic is used as the light amount signal. This G signal is stored in one line of FIFO memory
After being delayed at 1718, 1719, and 1720, the target line G3 signal and the G2 and G4 signals one line before and after the target line G3 signal are input to the arithmetic unit 1722. At the same time, the three-line color determination signals COL2, COL3, and COL4 generated in FIG. 14-1 are input.

 FIG. 14-6 shows the details of the arithmetic unit 1722.

G2, G3, G4, COL2, COL3, and COL4 are delayed by 2 or 3 pixels by flip-flops (latch circuits) 1723 to 1734, respectively. Here, the target pixels are G32 and COL32. G32 is the peripheral pixel G2 by comparators 1737 to 1740.
2, compared with G31, G33, G42. This comparator output outputs "1" when the peripheral pixel has a lower light intensity value than the target pixel. Then, the AND gates 1741 to 1744 take AND of the color determination signals of the peripheral pixels and output a CAN signal at the OR gate 1745.

<Description of Character Edge Determining Unit> Next, the character edge determining unit 1103 in FIG. 8 will be described with reference to FIG.

A document 1901 in FIG. 16 is an example of an image having shading, and includes a character edge area 1902 and a halftone area 1903 expressed by halftone dots.

As a method for extracting edge information in an image, in the present embodiment, as shown by 1094 in FIG. 17, a steep density in a pixel block having nine neighboring pixels surrounding a target pixel _{xi, j} as one unit is used. It is determined whether or not there is a change, and the fact that a steep density change point is continuously present in a specific direction is used.

Specifically, the difference value of the pixel of interest x _{i, j and} its neighboring pixels is taken, and J ₁ = x _{i, j + 1} −x _{i, j−1} J ₂ = x _{i + 1, j} −x _{i−1, j} J ₃ = x _{i + 1, j + 1} −x _{i−1, j−1} J ₄ = x _{i + 1, j−1} −x _{i−1, j + 1} J ₅ = x _{i, j−1} −x _{i, j + 1} J ₆ = x _{i−1, j} −xi _{+ 1, j} J ₇ = x _{i−1, j−1} −xi _{+ 1, j + 1} J ₈ = x _{i−1, j + 1} −xi _{+ 1, j−1} In the magnitude determination, a determination is made as to whether or not a sharp density change exists, and further, an edge area is extracted by determining whether or not a sharp density change point is continuously present in a specific direction. A halftone dot area is determined by a specific combination of specific density changes.

Specifically clay, detection of longitudinal edge there is a high concentration to the right as shown in example area 1905 of FIG. 16, the nature of the point values of J ₁ in the above equation is greater is continuous in the longitudinal direction There is. That is, the patterns 2101, 2102 are shown in FIG. The detection of the lateral edge with a high concentration on the lower side, as shown in area 1906 utilizes the property that points larger value of J ₂ in the above equation is continuous in the transverse direction. That is, the nineteenth
These are patterns 2103 and 2104 in the figure. Further, the detection of the right oblique direction of the edge with a high concentration in the lower right direction as shown in region 1907 the property that the point values of J ₃ in the above expression is greater is continuous in the right oblique direction. Therefore, the pattern 21 in FIG.
05,2106. Further, the high density detection of the left oblique direction of the edge with the lower left direction as shown in region 1908 the property that the point value of J ₄ is large are continuous in the left oblique direction (in Figure 19 pattern 2107 , 2108). The detection of the longitudinal edge with a high concentration on the left side as shown in area 1909, a point value of J ₆ is large there is a property that is continuous in the lateral direction (in Figure 19 patterns 2111, 2112 ). Detection of right oblique direction of the edge with a high concentration in the upper left direction as shown in region 1911 the property that the point value of J ₇ is large are continuous in the right oblique direction (pattern of Figure 19 2113,211
Four). Then, the high density detection of the left oblique direction with the upper right direction as shown in region 1912 the property that the point value of J ₈ is large is continuous in the left oblique direction (Pattern 2115 and 2116 of Fig. 19) .

On the other hand, also in the halftone portion, such as shown in region 1909 to 1912, although values up to J ₁ through J ₈ increases, less continuity in a particular direction. In other words, in the halftone dot region, a characteristic pattern is indicated by a specific combination of density changes in a specific direction, so that it is distinguished from an edge of a character line drawing.

FIG. 1 is a block diagram of the character edge determination circuit 1103 in this embodiment.

In the figure, reference numeral 1801 denotes a density change point detection unit; 1802, a density change continuity and halftone dot extraction unit; 18021, an area determination unit for determining the area of the extracted halftone signal;
An edge signal generation unit that finally forms a character edge area signal EDGE based on the continuous density change detected in 802 and the halftone dot area detected in 1802.

Block diagrams of the density change point detecting section 1801 and the extracting means 1802 are shown in FIGS. 15-1 and 15-2.

In the density change point detection unit 1801, the image signal G is delayed by the line memories 1803 and 1804, and three lines are simultaneously input to the detector 1801a, and the eight types of density change information AK1 to AK
8 is output.

here, (Where T _{1 to} T ₃ are preset thresholds).
In other words, the density is “1” when there is a steep increase in density in eight directions of right, lower, lower right, lower left, left, upper, upper left, and upper right for the target pixel, and is “0” otherwise. . Here, registers 2023, 2024, and 20 for holding the respective thresholds
25 is written from a CPU (not shown).

In order to realize this processing, the density change point detection unit of the embodiment is used.
1801a are flip-flops 2001 to 2 as shown in FIG.
006, difference calculator 2007-2014 and comparator 2015-2022
It is composed of

That is, in FIG. 18, flip-flops 2001 to 2006
Image data of a pixel shown in the pixel block 1904 by connexion FIG. 17 to clock CLK is latched, the difference calculator 2007 to 2014, the comparator 2015 to 2022 calculates the J ₁ through J ₈ described above, the determination result in AK1 to AK8 are output.

The extracting means 1802 determines that the steep density change is continuous in a direction having an angle of 90 ° with respect to the density change direction. In this embodiment, as shown in FIG. 20, the continuity of the density change in a 5 × 5 pixel block centered on the target pixel is observed. Patterns 2201 and 2202 show reference pixels when detecting a continuation of edges in the vertical direction. In other words, this is a reference pixel in the case where it is detected that three pixels in which the characteristic of the density change of the above-described peripheral pixel is AK1 or AK5 are continuous three pixels. Similarly, patterns 2203 and 2204 indicate reference pixels when detecting the continuation of AK2 or AK6.
Patterns 2205 and 2206 are for detecting a continuation of AK or AK8, and patterns 2207 and 2208 are reference pixels for detecting a continuation of AK3 or AK4.

In this embodiment, there is a reason that the pixel of interest is not brought to the center of the continuity check when extracting the continuity of the density change. That is, as shown in FIG. 21, the pixels constituting the character end are also determined as pixels included in the continuous edge.

In order to detect the continuity of the density change in the 5 × 5 pixel block area, the density change point detection unit 1801 detects
The edge values in eight directions for each pixel are stored in 4-line memory 1805 ~
Delayed by 1808. 5 formed in this way
Line density change information AK1 to AK8, BK1 to BK8, CK1 to CK8, D
Each of K1 to DK8 and EK1 to EK8 is delayed by one to five flip-flops in order to check the continuity shown in FIG. Thereafter, the NAND gates 1809 to 1824 indicate that a central pixel (CPU3, CBT3, CLF3, CRT3, CUL3, CBR3, CUR3, CBL3) forms a 3-pixel continuous edge at the end.
Generate GE signal.

For example, NAND gate 1809 is connected to AK6 (pattern 211 in FIG. 19).
It is nothing but detecting whether the feature 1) is continuous in the form of the pattern 2203 in FIG. Also, NABD gate
1810 detects that the features of AK6 are continuous in the form of a pattern 2204.

Detection at other gates is as follows. That is, the gate 1811 detects that the characteristics of AK2 are continuous in the form of 2203.

Gate 1812 detects that the features of AK2 are continuous in the form of 2204.

Gate 1813 detects that the features of AK5 are continuous in the form of 2201.

Gate 1814 detects that the features of AK5 are continuous in the form of 2202.

Gate 1815 detects that the features of AK1 are continuous in the form of 2201.

Gate 1816 detects that the features of AK1 are continuous in the form of 2202.

Gate 1817 detects that the features of AK7 are continuous in the form of 2208.

Gate 1818 detects that the features of AK7 are continuous in the form of 2207.

Gate 1819 detects that the features of AK3 are continuous in the form of 2208.

Gate 1820 detects that the features of AK3 are continuous in the form of 2207.

Gate 1821 detects that the features of AK8 are continuous in the form of 2205.

Gate 1822 detects that the features of AK8 are continuous in the form of 2206.

Gate 1823 detects that the features of AK4 are continuous in the form of 2205.

Gate 1824 detects that the features of AK4 are continuous in the form of 2206.

In this way, only the character edge portion in the character area 1902 shown in FIG. 16 is determined and output as the EDGE0 signal.

On the other hand, the portion shown in FIG. 15-2 of the extracting means 1802 is a portion for detecting a specific combination of density changes in a specific direction in an image and extracting a halftone image.

That is, this part is the NOR gates 1851 to 1858, 1859 to 186
2, 1863 to 1864, 1865, and outputs a DOT0 signal indicating that the pixel is a pixel in the halftone image.

FIG. 22-1 shows a judgment pixel group for making a dot judgment.
As shown in FIG. 22-1, a pixel group 2251, 2252, 22 composed of four pixels indicated by a thick frame around the pixel of interest 2250
At 53 and 2254, characteristic dot changes are detected to detect halftone dots.

Specifically, the output of the gate 1851 in FIG. 15-2 indicates whether there is a downward density change of at least one pixel in the pixel group 2254.

The 1852 output of the gate indicates that there is an upward density change of at least one pixel in the pixel group 2253, and the output of the gate 1853 indicates that there is an upward density change of at least one pixel in the pixel group 2254. I have. The output of the gate 1854 indicates that there is at least one pixel or more downward density change in the pixel group 2254, and the output of the gate 1855 indicates that there is at least one pixel or more right density change in the pixel group 2252. . The output of the gate 1856 indicates that there is at least one pixel or more leftward density change in the pixel group 2251, and the output of the gate 1857 indicates that there is at least one pixel or more leftward density change in the pixel group 2252. The output of the gate 1858 indicates that there is a rightward density change of at least one pixel in the pixel group 2251.

These outputs are logically operated by the gates 1859 to 1865. As a result, in the four cases shown in the patterns 2210, 2211, 2212 and 2213 in FIGS.
The output DOT0 becomes “1”.

Here, the symbol indicates that one or more rightward density changes exist in the pixel group surrounded by the thick line. The symbol indicates that one or more leftward density changes exist in the pixel group surrounded by the thick line, and the symbol indicates that one or more upward density changes exist in the pixel group surrounded by the thick line. The symbol indicates that one or more downward density changes exist in the pixel group surrounded by the thick line.

Here, the case shown in the pattern 2210 indicates that the vicinity of the position of the pixel of interest is in a halftone portion as shown in the pattern 2214 or 2215. Also, the case shown in the pattern 2211 indicates that the vicinity of the pixel of interest is located at a halftone dot portion shown in the same 2212 or 2217, and the case shown in the pattern 2212 shows the same.
Halftone dots as shown in 2218 or 2219, and pattern 22
The case as shown in FIG. 13 indicates that the target pixel is located at the halftone dot portion as shown in 2220 or 2221.

Now, this result may be output as it is as a signal as to whether or not the pixel of interest is in the halftone dot area.
In response to the DOT0 signal obtained here, the area determination unit 18021 (see FIG. 1) performs determination in a wide area.

FIG. 15-3 is a block diagram of the area judging section 18021.

1831 is one in a 4 × 3 window containing the pixel of interest.
This is a determination unit for determining whether or not there is a point where DOT0 = “1” or more. If it exists, “1” is set. If not, “0” is set.
Is output as DOT0 '. More specifically, line memories (FIFO) 18311 and 18312 each provide a delay of one line, and flip-flop 18313 simultaneously outputs D lines of three lines.
Make OT0 input. The OR gate 18314 and the flip-flops 18315, 18316, and 18317 respectively delay one clock, and the outputs thereof are OR gate 18318.
To obtain DOT0 '.

For example, for three continuous lines as shown in Fig. 15-5
Assume that DOT0 is output. However, in the figure, the hatched grid indicates “1”, and the blank grid indicates “0”.

By the way, at this time, for the pixel of interest 1851, a logical OR is taken within a 3 × 4 window indicated by 1852, and DOT0 'is calculated.

Due to this processing, there was a sparse presence in the halftone pixel
The DOT0 signal will be converted to a relatively continuous DOT0 'signal.

Referring back to FIG. 15-3, the circuit 1832 calculates the DOT0 'signal thus generated over a wide area, and generates a DOT1 signal indicating whether or not the pixel of interest is in a dot area.

Reference numerals 18321 and 18322 denote line memories (FIFO), each of which delays one line. Reference numerals 18323 and 18324 denote calculators. As shown in FIG. 15-6, the main pixel 1861 (the i-th line in the sub-scanning and the j-th pixel in the main scanning) is set every four pixels in the main scanning and one line in the sub-scanning. DOT0 'is sampled every other time. One line before (the (i-1) th line), N is an appropriate integer, and DOT0 '= "1" at each of the pixels at the main scanning position j, j-4, j-8,..., J-4N. Sum of things SUML1
, J + 4, j + 8,..., J + 4 The sum SUMR1 of the Nth pixels for which DOT0 = "1" is obtained. Further, the sum SUML2 of DOT0 = “1” at each of the main scanning positions j, j−4, j−8,..., J−4N one line after the current pixel of interest (the (i + 1) th line) Main scanning position
DOT0 = "1" at each of the j, j + 4, j + 8,..., j + 4Nth pixels
Is obtained, and the sum SUMR2 is obtained, and each is output.

Reference numerals 18325 and 18326 denote adders, each of which calculates the sum SUM of DOT0 'on the left side of the pixel of interest as SUML ← SUML1 + SUML2, and calculates the sum of sampling SUMR of DOT0' on the right side of the pixel of interest as SUMR ← SUMR1 + SUMR2. Output.

18327 and 18328 are comparators, and 18329 is an OR gate. Reference numeral 1830 denotes a register in which a threshold value T4 is written by a CPU (not shown).

As a result, DOT1 is output as "1" only when at least one of SUML> T4 or SUMR> T4 holds, and otherwise becomes "0". DOT1 results in an area signal that becomes "1" in the halftone area. That is,
If it is determined that there are a predetermined number (T4) or more of pixels determined to be halftone pixels in a predetermined space to the right or left of the target pixel (the size is determined by N), This indicates that the position is also determined to be in the halftone dot area. That is, the halftone image is reproduced only because the halftone dots are distributed in a certain area, and from this viewpoint, the significance of this processing can be easily understood. Conversely, if a relatively small black point, such as a punctuation mark, is found in the original image, that portion may not be recognized as a halftone dot.

Next, a block diagram of the edge signal generator 18022 in FIG. 1 is shown in FIG. 15-4, and will be described below.

In the figure, reference numerals 1841 and 1842 denote line memories.
A delay corresponding to a line is given to EDGE0, and synchronization of sub-scanning of DOT1 is performed.

1843 is composed of flip-flops, EDGE0 and DOT
Synchronization of the main scanning of 1 is performed, and EDGE0 'and DOT1' are output. 1844 is an inverter, and 1845 is an AND gate.

In this configuration, EDGE0 '= "1" and DOT1' =
Output EDGE = "1" only when "0". In other words, when the target pixel has EDGE0 = "1", that is, there is a continuous density change, and DOT1 = "0", that is, when it is not a halftone dot area, it is regarded as an edge such as a character or a line drawing.
= "1".

What has been described above is the feature extracting unit 403 in FIG.

Next, the color judgment signal BL,
The operation of the color signal processing unit 402 and the color processing control signal generation unit using UNK, COL, CAN and the character edge determination signal EDGE will be described with reference to FIG.

103 is a light amount signal-density signal conversion unit, and R, 0 to 255,
The G and B signals are converted into C, M and Y signals in the range of 0 to 255 by the following equation.

Here, D _max is a constant.

The black component K included in the C, M, and Y signals is determined by the black extraction unit 104 as follows.

K = min [C, M, Y] (min […] is a function that returns the minimum value in […]) The density signals C, M, Y, and K of the four colors to which K is added are UCR / Mask units 1
In step 05, the under color is removed and the color of the developing material of the printer 202 is removed by the following equation.

Wherein a masking coefficient for _{a 11 ~a 14, a 21 ~a} 24, a 31 ~a 34 and a ₄₁ ~a ₄₄ is predetermined color contamination _{removal, U 1, U 2, U} 3 is It is a UCR coefficient obtained by removing the K component from each of the M, C, and Y color components. Here, one of M ', C', Y ', and K' is selected by a two-bit developing color signal PHASE from the control unit 401 and is output as a V1 signal. PHASE signal 0,1,
As described above, M ', C', Y ', and K' are selected in correspondence with 2,3.

Reference numerals 112 and 113 denote line delay memories. Since the feature extraction unit 403 delays the generation of the character edge determination signal by three lines and four clocks, the V1 signal and the M signal are similarly delayed by three lines and four clocks.

In addition, the color determination unit 106 delays two lines and two clocks until generating a determination output such as BL or UNK. In order to match this delay amount with the delay amount of the character edge determination unit 107, the line delay memory 120 generates signals BL1, UNK1, COL1, and CAN1 delayed by one clock and two clocks.

<Description of Weighting Addition Unit> FIGS. 23-1 to 23-7 show a color determination signal and a character edge determination signal for the “A” character read in various color states.

23-1 to 23-6 show the determination signals of the cross section of the character a shown in FIG. 23-7.

FIG. 23-1 shows a case where a black “A” character is read as black, and M2 indicating an achromatic density signal (hereinafter referred to as an ND signal).
The signal (output from the line memory 113) is read by the blurring of the reading optical system as compared with FIG. 23-7. The edge signal is formed in a state in which the edge of the character is swollen due to the continuation of the density change of AK3 and AK7, and only the BL1 signal is generated as a color determination signal.

In the present embodiment, since the M2 signal and the EDGE signal indicating the ND signal use a green (G) color separation signal, other than green characters (FIGS. 23-2 and thereafter) are also schematically shown in FIG. The same output is shown. In the case of green characters, the M2 signal and the EDGE signal are not generated.

FIG. 23-2 shows a case where an “A” character composed of color characters is read, and a COL signal indicating that the character is a chromatic color and a color pixel having a density higher than the own pixel exist around the own pixel. A CAN1 signal is generated as shown in FIG.

FIG. 23-3 shows a case in which an "A" character composed of intermediate chroma characters is read, and a UNK1 signal indicating the intermediate chroma is generated.

FIG. 23-4 shows a case in which the character “A” composed of black characters is read with a color shift, and the BL1 signal is thinner than in FIG. Signal UN
K1 occurs.

FIG. 23-5 shows a case where the character "A" composed of colored characters is read with a color shift, and the COL1 signal is thinner than in FIG. 23-2, while the UNK1 signal is generated at the character edge. . Also, CAN1
As for the signal, the portion corresponding to the outside of the character edge is narrowed by the reduced number of portions determined to be color.

FIG. 23-6 shows a case in which a color character close to the intermediate saturation is read with color misregistration, and a black judgment pixel occurs at the edge. In this case, except that the BL1 signal is generated instead of the UNK1 signal,
The same signal as in FIG. 5 is generated.

FIGS. 24-1 to 24-3 are enlarged cross sections a of black characters, middle-saturation characters, and middle-saturation characters of black characters in FIGS. 23-1, 3, and 4 respectively. . Where V2 is M,
An example of an output signal of the circuit 105 for C, Y, and BK is shown.

FIG. 24 shows a case where a black character is read, and since the UCR (under color removal) processing is performed in the circuit 105, each of the M, C, and Y color components is reduced to about 20%. However, since this character is a black character, it is desirable to record using black toner as much as possible.

In addition, it is desirable to reduce the M, C, and Y color components as much as possible in the intermediate saturation generated at the edge of the black character as shown in FIG. 23-4.

Conversely, it is desirable to reduce the BK component in the intermediate chroma generated at the edge of the color character as shown in FIG. 23-5.

Also, it is desired to distinguish the black component generated at the edge of the color character from the black character edge of FIG. 23-1 as shown in FIG. 23-6.

As described above, in this embodiment, as shown in FIG. 25, the UCR / Mask circuit is used in accordance with the results of the color judgment signal and the character edge judgment signal.
Color recording signal V2 (M ', C', Y ', K') from 105 and ND
Color recording is performed by appropriately mixing the (density) signal M2.

FIG. 25 (a) corresponds to the black character EDGE in FIG. 23-1.
A "0" signal (no development) is output when the development color is M, C, or Y, and a density signal M2 is output when the development color is BK.

FIG. 25 (c) corresponds to the intermediate saturation edge of FIGS. 23-3 and 23-5. In this case, in order to emphasize the black component of the edge, half of M ', C', Y 'generated from the circuit 105 are output as color recording signals V2 for the developed colors M, C, Y, respectively. When the developed color is BK, a signal obtained by adding 50% each of the BK 'output of the color recording signal V2 and the density signal M2 is output.

FIG. 25 (f) corresponds to the non-edge portion of the black characters in FIG. 23-1. Here, in order to improve the connection of the signal with the edge portion recorded in BK single color, M ′ of the color recording signal V2,
The C 'and Y' components are reduced to 3/4, and 3/4 of the BK 'component during BK recording.
Is added to 1/4 of the density signal M2.

FIGS. 25 (b), (d), and (g) show the case where the above-described black enhancement operation is not performed by the CAN1 signal.

With reference to FIG. 24, a description will be given of a change in the image signal due to the arithmetic processing of FIG. Here, “V2 (M)” is a V2 output signal when PHASE = 0 (magenta developed color), and V2 (C), V2
(Y) and V2 (BK) are V2 output signals for cyan, yellow and black, respectively.

FIG. 24-1 is a black character portion, and a portion b is an edge portion corresponding to FIG. Here, the recording signal amounts of M, C, and Y become “0”, and the density signal M2 is output as a BK signal. The portion c is a black non-edge portion corresponding to (f) in FIG. 25, and V4 (M), which is a V4 signal of the development colors M, C, and Y.
V4 (C) and V4 (Y) are 3/4 of V2 (M), V2 (C) and V2 (Y)
And the BK signal is a value obtained by adding 3/4 of V2 (BK) and 1/4 of M2.

FIG. 24-2 shows an intermediate color character, and the d portion is an edge portion corresponding to FIG. 25 (c). Here, V4 (M), V
4 (C), V4 (Y) is 1/2 of V2 (M), V2 (C), V2 (Y)
V4 (BK) is a value obtained by adding 1/2 of V2 (BK) and 1/2 of M2.

FIG. 24-3 shows a case where intermediate saturation occurs in an edge portion of a black character. The edge portion e is subjected to the same processing as the d portion, the non-edge portion is determined by black determination (BL = 1), and the same as the c portion. Processing is performed. This reduces the color signal at the black character edge.

In the embodiment, in order to generate the V4 signal shown in FIG. 25, multipliers 114 and 115 and an adder 116 are provided in FIG.

Then, BL1, UNK1, COL1, C
Receiving each color determination signal of AN1 and the character edge determination signal EDGE, it generates multiplier coefficients GAiN1 and GAiN2 of the multiplier.

As shown in FIG. 26, the multiplication coefficient generator 108 is constituted by a ROM (look-up table), and inputs each 5-bit determination signal and a PHASE signal (2-bit) as an address. Two gain signals of three bits each
Outputs GAiN1 and GiN2.

FIG. 28 shows the relationship between the address of the ROM and the output data in the multiplication coefficient generator 108. The gain signal here is obtained by quadrupling the actual gain, and is multiplied by 1/4 in the multipliers 114 and 115.

FIG. 28 shows details of the multipliers 114 and 115. The 8-bit image signal is quadrupled and doubled by bit shift type multipliers 2901 and 2092, respectively. They are 3-bit gain signals GAiN (1), GAi
The gates 2903, 2904, and 2905 select the data according to N (2) and GAiN (0), and add the data at the adders 2906 and 2907. After that, it is multiplied by a factor of 4 by a bit shift type divider 2908 to form a 255 limiter 2909.
, All 9-bit data of 255 or more are put together into 255 8-bit data and output.

The color recording signal V2 and the density signal M2 weighted and added by the color determination signal and the character edge determination signal as described above are input to the spatial filter 117.

<Description of Spatial Filter Unit> FIG. 29 shows a configuration diagram of the spatial filter 117 in this embodiment.

The spatial filter of the embodiment is an edge-enhanced filter using a Laplacian filter of 3 × 3 pixels, and the Laplacian multiplier can be switched between two types of “1 / 2,1”.

Reference numerals 3001 and 3002 denote line delay memories, respectively, and the image signals V4, V42 and V43 for three lines generated by these are delayed by one clock at flip-flops 3003 to 3006, respectively. Here, the target pixel is V43, and V41, V42, V44, and V46 are multiplied by (−1) in multipliers 3007 to 3010 to form a Laplacian, and are added in adders 3011, 3012, and 3013, respectively. Further, the target pixel V43 is multiplied by 4 by the multiplier 3014 (this signal is V43F),
The Laplacian L is generated by adding the output of the adder 3013 in the adder 3015. This Laplacian L is a multiplier 30
It is multiplied by 1/2 by 16. At the adder 3017, the target pixel V43 and L /
2 are added to generate a weak edge emphasis signal E1. On the other hand, the adder 3018 adds the pixel of interest V43 and the Laplacian L to generate a strong edge emphasis signal E2. The two types of edge-enhanced signals and the signal SMG (to be described in detail later) obtained by subjecting the pixel of interest V43 to smoothing processing are converted into control signals DFiL
(1) Selected by DFiL (0) and output as V5 signal.

When DFiL (1) is “0” and DFiL (0) is “1”, a weak edge emphasis signal E1 is selected, and DFiL (1) is “1”.
When DFiL (0) is "1", a strong edge emphasis signal E2 is selected. When DFiL (0) is “0”, the signal SMG
Is selected and output as the V5 signal.

It is a filter control signal generator that generates a 2-bit DFiL signal consisting of the filter switching signals DFiL (1) and DFiL (0).

In the present embodiment, strong edge enhancement is applied to a black character edge portion, and strong edge enhancement is applied to a black character edge portion so as to bring out a black edge. Non-character edge portions are not edge-enhanced to prevent a change in color tone due to edge enhancement.

The edge portion of the character having the intermediate saturation and the color is configured to apply a weak edge enhancement so that a change in color tone due to the edge enhancement is not so noticeable while being recorded in the edge portion in a sharp manner. When the CAN1 signal is "1", the BL1 signal and the UNK1 signal generated due to the color shift at the edge of the color character edge are not subjected to EDGE emphasis.

FIG. 30 shows a circuit of the filter control signal generator.
The meaning of the output signal is as shown in FIG.

The area not judged as an edge by the character edge judgment unit 107 in this embodiment includes a halftone original as shown by 1903 in FIG. When this halftone original is read in units of pixels using a CCD, moiré fringes occur due to the regularity of the pixels of the CCD and the regularity of the halftone original. In order to prevent this, in the present embodiment, a smoothing process is performed on a document area that has not been determined as a character edge (an area having a high possibility of a halftone dot) to generate a signal SMG. In the present embodiment, as the smoothing filter, the target pixel as shown in FIG.
A smoothing filter is used which multiplies the surrounding four pixels by 1/8 and adds them.

Returning to FIG. 29, the smoothing process in this embodiment will be described.

In the illustrated adders 4201, 4202, and 4203, four pixels V41, V42, V44, and V46 around the target pixel are added. A signal V43 obtained by quadrupling the target pixel V43 by the adder 4202 with respect to the signal V43
Add F. The result is divided into a bit shift type divider 4
By performing 1/8 at 205, a smooth filter signal SMG is obtained.

In the FiLTER circuit 117, the pixel of interest is delayed by one clock from one line, so the F
The iL signal is delayed by one line and one clock in the line memory 121 (see FIG. 2) to become a DFiL signal.

Similarly, the GAM signal from the gamma switching signal generating section 110 and the SCR signal from the screen switching signal generating section 111 are also delayed by one line and one clock to become a DGAM signal and a DSCR signal.

<Description of Gamma Conversion Unit> The gamma conversion unit 118 performs image density conversion.
The gamma conversion unit 118 is composed of a ROM as shown in FIG. 32. The filtered 8-bit V5 signal is input as an address of the ROM, and the corresponding gamma conversion output is output.
It is output as an 8-bit ViDEO signal from the data terminal of the ROM.

2 which is input to the address line together with the V5 signal
As shown in FIG. 33, four types of gamma conversion characteristic curves are selected according to the bit DGAM signal.

In FIG. 33, the case of DGAM = 0 is the case of “input = output” and is applied to a non-character edge portion.
When DGAM = 1, as shown in the figure, "0" is uniformly determined for inputs 0 to j, and "255" is uniformly determined for 255-j to 255,
In the meantime, (j-255-j) corresponded to output values 0-255. In other words, the slope between them is a straight line having a conversion characteristic of 255 / (255−2j). This means that a video signal with a lower density is output for a nearby input that is a low density input, and a video signal with a higher density is output for a neighborhood input that is a high density 255.
Since the signal is output and the density change of the input near 128 which is the middle of the density is emphasized, the character edge can be recorded more sharply. That is, this DGAM = 1
Applies to color character edges.

In the case of DGAM = 2, the value of j of DGAM = 1 is replaced by a larger k, and a character edge is recorded in the sharp. However, since the linearity between the input and the output is lost, the color tone cannot be compensated. Therefore, DGAM = 2 applies to mid-saturated character edges.

In the case of DGAM = 3, the characteristic uses l which is a value larger than k, and is applied to a black character edge for which sharpness is required.

This gamma switching signal DGAM is supplied to the gamma switching signal generator 11.
0 or the GAM signal is delayed by one line and one clock by the line delay 121. The gamma switching signal generator 110 is the third
As shown in FIG. 4, the ROM is constituted by a ROM, receives a color determination signal, a character edge determination signal, and the like as an address, and outputs a GAM signal as data. The contents of the ROM table are 35th
It is as shown in the figure. As described above, the black character edge portion (EDGW = 1, BL1 = 1) has GAM3, and the intermediate saturation character edge portion (EDGE = 1, UNK = 1) has GAM = 2. Therefore, when there is a CAN1 signal indicating that BL1 = 1 or UNK = 1, GAM = 0 is set so as not to emphasize the character edge.

<Description of Pulse Width Modulation> The gamma-converted VIDEO signal is output to the PWM
Input to 9 and pulse width modulation.

Then, by controlling the lighting time of the laser 213 with the pulse width modulated signal, a copy output 406 having a gradation density expression is obtained.

 FIG. 36 shows details of the PWM modulator 119.

The VIDEO signal is converted into an analog image signal AV by the D / A converter 3701. The image signal CLK synchronized with the VIDEO signal and the screen clock CLK4 having a frequency twice that of the image signal CLK are provided with a toggle flip-flop 37.
At 02,3703, it is synchronized with HSYNC and frequency-divided by 1/2, and converted into clocks CLK4F and CLKF each having a duty of 50%. These two clocks are transformed into triangular waves by the integrators 3704 and 3705, then the wave heights are adjusted by the amplifiers 3706 and 3707 to the output dynamic range of the A / D converter, and the analog comparators 3708 and 3709 respectively. Is compared with the AV signal.

As a result, the AV signal is converted into two pulse width modulation signals PW4 and PW. Then, at selector 3710, DS
One of PW4 and PW is selected in the CR signal, and becomes the laser drive signal LDR.

FIG. 37 shows the operation timing of this circuit. As shown in the figure, a triangular wave T obtained by integrating the clock 4F obtained by dividing CLK4 by 1/2
Ri4 is a triangular wave with one pixel cycle of the image. This triangle wave is D / A
Since the triangular wave and the analog image signal AV are compared in a substantially linear manner over the entire output range of the converter, the AV signal is subjected to pulse width modulation with one image pixel period as one cycle to become PW4.

Similarly, since TRi is made of CLKF obtained by dividing the pixel clock CLK by 1/2, the AV signal is subjected to pulse width modulation with two periods of the image as one cycle by the TRi to become PW.

The PW4 signal pulse-width modulated at one pixel period is recorded by the printer at the same resolution as the clock CLK. On the other hand, when an image is recorded using the PW4 signal, the basic density unit is as small as one pixel, and the gradation expression cannot be said to be sufficient due to the characteristics of the electrophotographic process used for the printer.

On the other hand, since the PW signal reproduces the density in units of two pixels, the gradation expression is sufficient, but the recording resolution becomes half of the PW.

In this embodiment, PW and PW4 are switched for each pixel by controlling DSCR according to the type of image.

Specifically, PW4 is used for the black character edge portion and the intermediate saturation character edge portion that require resolution. PW is used for the color character edge portion and the non-edge portion in order to emphasize the color tone. However, it has been experimentally confirmed that it is better to use PW4 which emphasizes the resolution of the color character edge even if the color tone is sacrificed for a document composed of fine color characters such as a map. The signal DSCR for switching between PW and PW4 is a line delay 121 from the SCR signal from the screen switching signal generator 111.
Is delayed by one line and one clock. Details of the screen switching signal generator 111 are shown in FIG.

As a result, in the DSCR shown in FIG. 37, the portion corresponding to the black toner development in the black or intermediate chroma character edge portion becomes Low, PW4 is selected only in this section, and the LDR signal is output. In this case, even if the character edge portion is determined to be a character edge portion, in the case of a character edge portion having color misregistration (CAN = 1), PW is used to prevent deterioration of the quality of the recorded image due to the color misregistration being emphasized.
It does not use 4 signals.

That is, the need for a sharp character edge is for a black character edge, and in the case of a color character edge, the reproduction of the color tone of the document is more important.

On the other hand, as shown in FIG. 24-1, the M, C, and Y toners do not exist in the black character edge portion. BK toner hardly exists in the color characters due to the operation of the UCR circuit 105.
As shown in FIG. 24-2, the BK toner and the M, C, and Y toners are moderately present in the intermediate chroma character edge portion. Therefore, another example of the screen switching signal generator 111 is shown in FIG.

In consideration of the above characteristics, in the present embodiment, the one-pixel cycle pulse width modulation signal PW4 can be used for laser driving only when the character edge determination unit is BK toner.

As a result, a black character edge having a small number of color components can achieve the same sharpness as that of the first embodiment, and a color character edge containing a small amount of a color component has only the BK component recorded in the shape of the color character edge. Since color gradation is maintained, color reproducibility is also guaranteed.

The operation of the screen switching signal generator 111 in FIG. 39 will be described below.

The gate 4401 decodes that the 2-bit PHASE signal is "3", that is, that the developed color is BK. The output permission signal of the NAND gate 4402 is used. The other gates in this figure are the same as those in FIG. 38,
The SCR signal becomes "0" only in the BK development color in the character edge part.

In the above-described embodiment, the determination is made for a full-color image, but it is needless to say that the determination is also effective for a white / black image.

Further, in the above-described embodiment, since the G signal is used, there is a problem that a character edge is hardly detected in, for example, a yellow character. Therefore, an arithmetic unit 10000 is provided as shown in FIG. 41, and an operation is performed from the R, G, B3 signals, for example, as Y = aR + bG + cB (a, b, c are constants).
There is also a method of detecting an edge for Y.

As described above, according to this embodiment, it is possible to reliably identify a halftone region having a flat density, a halftone dot for representing a halftone in an input image, and an edge of a character / line drawing, and to determine the identification result. Since a visible image is formed based on the original image, it is possible to reproduce an image faithful to the original.

In the embodiment, a color digital copying machine has been described as an example, but a monochrome copying machine may be used. Also, it goes without saying that the principle of the present invention may be applied to an apparatus other than a copying machine.

[Effects of the Invention] As described above, according to the present invention, it is possible to accurately determine the characteristics of an input image reproduced in halftone using halftone dots.

[Brief description of the drawings]

FIG. 1 is a block diagram of a character edge judging section in this embodiment, FIG. 2 is a block diagram of a color copying machine in the vicinity of a color processing section in this embodiment, and FIG. 3 is a block diagram of a color copying machine in this embodiment. FIG. 4 is a block diagram showing the internal arrangement of an image scanner in this embodiment. FIG. 5 is a diagram showing the overall flow of an image signal in the embodiment. FIG. 6 is a diagram showing CLK, CLK4 and CLK in FIG. FIG. 7 is a diagram showing the timing of the HSYNC signal, FIG. 7 is a diagram showing the wavelength and sensitivity characteristics for each of the R, G, and B color components, FIG. 8 is an internal block diagram of the feature extraction unit in FIG. 5, and FIG. 8 is a block diagram showing the internal arrangement of the pixel color determination unit in FIG. 8, FIG. 10-1 is a circuit diagram of the MAX / MIN detector in FIG. 9, and FIG. 10-2 is an input / output condition in FIG. FIG. 11-1 is a diagram showing a circuit configuration of the selector in FIG. 9, and FIG. FIG. 11 is a diagram showing the relationship between input data and output data in FIG. 11-1, FIG. 12-1 is a diagram showing the contents of an image area in the embodiment, and FIG. 12-2 is a diagram in FIG. FIG. 13 is a diagram showing a correspondence between each image area and an output signal, FIG. 13 is a diagram showing a relationship between R, G, B in a three-dimensional space, and FIG. 14-1 is a part of an area processing unit in FIG. FIG. 14-2 is a block diagram showing an area to be processed by the area processing unit, FIG. 14-3 is a block diagram showing another part of the area processing unit in FIG. 8, and FIG. 14-4. FIG. 14 is a diagram for explaining the occurrence of a black signal due to color fringing around the color signal, FIG. 14-5 is a diagram showing a change in the light amount value due to color fringing, and FIG. 14-6 is a diagram 14-3. 15-1 and 15-2 are block diagrams of the density change point detecting unit and the extracting means in FIG. 1, and FIG. FIG. 5-3 is a block diagram of the area judging unit in FIG. 1, FIG. 15-4 is a block diagram of the edge signal generating unit in FIG. 1, and FIG. 15-5 is included in the halftone image. FIG. 15-6 is a diagram showing the dots spatially transformed. FIG. 15-6 is a diagram showing the principle for judging whether or not the pixel of interest is within the halftone dot image area based on FIG. 15-5. FIG. 17 is a diagram showing the difference in density change direction in a minute portion between a character and an image of a halftone dot, FIG. 17 is a diagram showing a pixel block which is a minute portion in FIG. 16, and FIG. 18 is a diagram in FIG. FIG. 19 is a diagram showing types of patterns in an edge of a character and line drawing, FIG. 20 is a diagram showing a pattern group of continuity of density change in the embodiment, and FIG. 21 is a character end portion. Fig. 22-1 shows that it is advantageous to determine that the image is a continuous edge. FIG. 22-2 (a) to (d) are diagrams showing halftone dot patterns, FIG. 23-1 to FIG. 23-7 are diagrams showing judgment signals according to various character types, 24-1 is a diagram showing details of FIG. 23-1, FIG. 24-2 is a diagram showing details of FIG. 23-3, FIG. 24-3 is a diagram showing details of FIG. 23-4, FIG. 25 is a diagram showing the relationship between the decision signal and the V4 signal output from the adder in FIG. 2, FIG. 26 is a diagram showing the circuit configuration of the multiplication coefficient generator in FIG. 2, and FIG. FIG. 26 is a diagram showing the relationship between an input signal and an output signal in FIG. 26. FIG. 28 is a block diagram of a multiplier in FIG. 2, FIG. 29 is a block diagram of a spatial filter in FIG. 2, and FIG. FIG. 31 is a circuit configuration diagram of the filter control signal generator in FIG. 2, FIG. 31 is a diagram showing the relationship between input and output data of the filter control signal generator in FIG. 30, and FIG. 32 is FIG. FIG. 33 is a diagram showing the γ conversion characteristics in FIG. 32, FIG. 34 is a diagram showing the configuration of the γ switching signal generator in FIG. 2, and FIG. 35 is the input / output of FIG. FIG. 36 is a block diagram of a PWM modulator in FIG. 2, FIG. 37 is a timing chart in PWM conversion, and FIG. 38 is a circuit of an SCR switching signal generating circuit used in PWM conversion. FIG. 39 is a circuit diagram of another SCR switching signal generation circuit used for PWM conversion. FIG. 40 is a diagram showing a weight ratio of each pixel related to smoothing processing. FIG. 41 is another embodiment. 1 is an overall configuration diagram of the device. In the figure, 105: UCR / Mask circuit, 106: Color judgment unit, 107 ...
... Character edge determination unit, 108 ... Multiplication coefficient generation unit, 109 ...
Filter control signal generator 110, gamma switching signal generator 111, screen switching signal generator 201 image scanner 202, printer 401, controller 40
2 Color signal processing unit, 403 Feature extraction unit, 404 Color processing control signal generation unit, 1301 MAX / MIN detector, 1302 to 130
9 ... Selector, 1310-1315 ... Subtractor, 1316-1323 ...
… Comparator, 132-1328 …… AND gate, 1801 ……
... Extraction means, 18021... Area determination section, 18022... Edge signal generation section.

Continuation of the front page (56) References JP-A-63-263974 (JP, A) JP-A-63-82058 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) G06T 7 / 40

Claims (5)

(57) [Claims] A first detecting means for detecting a density change and a direction of the density change in the vicinity of a pixel of interest in an input image; and a direction in a direction substantially orthogonal to the direction of the density change detected by the first detecting means. Second detecting means for detecting continuity of the pixel in which the density change is detected, and third detecting means for detecting a halftone dot of the image in the input image based on the density change detected by the first detecting means. Based on the detection result of the second and third detection means,
An image processing apparatus comprising: a determination unit configured to determine a characteristic of the input image. 2. The image processing apparatus according to claim 1, wherein said third detecting means detects halftone dots by detecting a change in density in a two-dimensional direction. 3. An image processing apparatus according to claim 1, wherein said first detecting means detects a density change and a direction of the density change based on the input multi-valued image data. 4. An image processing apparatus according to claim 3, wherein said image data is a green component. 5. The image processing apparatus according to claim 3, wherein said image data is a luminance component.