JPH02244876A - Sharpness improving system for picture processor - Google Patents

Abstract

PURPOSE:To prevent a character from being made thick and to reproduce the character with high quality by extracting an edge signal from a picture signal, and correcting a color signal required for a hue identification means. CONSTITUTION:The system is so constituted that a correction means 5 is pro vided to extract an edge signal from a picture signal and to correct a color signal required for a hue identification means 3. When a character with a hue such as W, Y, M, C, B, G, R, K is included in a picture, the hue identification means 3 detects the hue and a high pass filter 2 detects the edge of the charac ter and an edge detection signal corrects a color signal required for the hue identification means 3. Thus, even when an original picture signal is unsharpened, the required color signal whose edge is sharp is obtained thereby preventing the character from being made thick.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a color image processing device such as a color copying machine, a color laser printer, or the like, and in particular, to a color image processing device such as a color copying machine, a color laser printer, or the like, which generates an edge emphasis signal using an edge detection signal and a necessary color signal, The present invention relates to an edge processing circuit of an image processing apparatus that modulates the output of a low-pass filter and combines the signal with a recording signal and outputs the signal.

[Conventional technology]

FIG. 17 is a diagram showing the configuration of a digital color image processing device, FIG. 18 is a diagram showing an example configuration of a conventional edge processing circuit,
FIG. 19 is a diagram showing the configuration of the hue detection circuit, FIG. 20 is a diagram for explaining the character thickening phenomenon, and FIG. 21 is a diagram for explaining edge emphasis processing.

Generally, a color copying machine reproduces a full-color original by performing a development process using four colors: Y (yellow a-), M (magenta), C (cyan), and K (black). At this time, a very large memory capacity is required to store full-color image data obtained by one document reading scan for four development process executions. Therefore, the developing process is executed by performing signal processing while repeatedly reading and scanning the document for each developing color.

When reading the original, a line sensor is used to optically scan B.
The read image data is captured using the color separation signals of (blue), G (edge), and R (red), and as shown in FIG.
D conversion 401, color masking (color correction 3
The toner signals YSM and C are converted into color toner signals YSM and C through the toner signal YSM and C 402, respectively. After that, the UCR403 performs black plate (K) generation and undercolor removal, and the hue separation type nonlinear filter section, T
The toner signal X of the developing color is converted into on/off binary data through an RC (tone adjustment) 41O1SG (screen generator) 411. Then, the charged photoreceptor is exposed using this binarized data to control the radar light, and images of each color are superimposed using halftone gradation to reproduce a full-color original.

Usually, in a digital color image processing device, binary images such as characters and line drawings and halftone images such as photographs and halftone printed materials are mixed together. Therefore, there is a method that performs edge enhancement processing on originals with different types of images, such as those described in 4, in order to obtain binary surface images of characters, line drawings, etc. with particularly high sharpness by introducing nonlinear filter processing. Various proposals have been made. Part 1
As an example, FIG. 17 shows a configuration example including a hue separation type nonlinear filter section.

The hue separation type nonlinear filter is generated by performing a version generation and undercolor removal processing in UCR403, f:, YS
Image data X of the development color selected according to the development process is manually inputted from the signals M and C, and is branched into two systems. One of them is subjected to smoothing processing by a smoothing filter 404, and the other is subjected to edge enhancement processing by a T-transform 406, an edge detection filter 407, and an edge enhancement LUT 40g. These outputs are finally combined by an adder 409 and output as a nonlinear filter signal. FIG. 18 shows an example of the configuration of the edge processing circuit.

In edge processing, the hue of the input image is detected by the hue detection circuit 405, and it is determined whether the developed color at that time is a required color. Here, if the input image is a black area, edge enhancement of the YSM and C chromatic signals is not performed, and the K
control to emphasize only the edges according to the amount of edges.

As shown in FIG. 19(a), the hue detection circuit 405
, M, C maximum and minimum circuit 412
, a multiplexer 413 for selecting a developed color, a subtraction circuit 414 for calculating the difference between the maximum value and the minimum value, a subtraction circuit 415 for calculating the difference between the minimum value and the developed color, and a comparator 4.
16 to 418. Comparators 416-41
8 is compared with the threshold value, and if it is larger than the threshold value, r, m, c
The outputs of 'm' and y' are respectively set to logic '1'. Then, from this output, the judgment hue is derived according to the judgment conditions shown in the same figure (b), and furthermore, the necessary color "
1” or an unnecessary color “0”. The judgment hue is W (white), YS, which is used as a normal character color.
It targets eight colors: M, C, B, G, and RlK.

As is clear from the criteria for determining necessary colors and unnecessary colors, if the hue is B, for example, the developing colors m and C are considered necessary colors,
Other colors are considered unnecessary. Therefore, in this case, in the cycle of the necessary color, the edge is emphasized by the edge emphasis LUT 408, and in the cycle of the unnecessary color, the edge is emphasized by
T408 (■) is used as a signal without edge enhancement.

[Problem to be solved by the invention]

As described above, in the edge enhancement processing, hue identification is performed by comparison with the threshold value tit, and based on the result, the edge detection signal is converted by the edge enhancement LUT to generate an edge enhancement signal. However, the MTF characteristic of IIT becomes worse as the frequency becomes higher, as shown in FIG. 20(a). Moreover, the degree of deterioration of this MTF also differs depending on the color and the main and sub-scanning directions. MTF like this
Due to this deterioration, even if the density distribution of the original is as shown in a of FIG. In hue detection, the hue is determined by comparing this signal with the threshold th. Therefore, the hue is determined when the width of W is considerably thickened like W', and this is considered as the processing range for edge enhancement. Ru. Based on this determination, an edge emphasis signal d as shown in FIG. 5(C) is added to emphasize the edge, so that it is reproduced as a bold character like C in FIG. 2(B). Furthermore, this thickening of text appears due not only to IIT but also to IOT's developing material, development method, development characteristics, etc.

Furthermore, in the above method, black character reproduction can be improved compared to other conventional methods in which all signals are emphasized in YSM and C%, but smoothed signals remain in the Y and M%C signals. That is, as shown in the edge enhancement LUT 408 in FIG. 18, necessary colors are emphasized by ■, and unnecessary colors are simply removed by ■. An edge-enhanced signal is generated that emphasizes only K without emphasizing Y, M, and C, but the smoothing filter does not emphasize Y, M, and C, as shown in FIG. , a smoothed signal is generated in which all the signals are smoothed. Therefore, when these are finally combined, smoothed signals of YlM and C remain as shown in (C) of the same figure.

Normally, even in the case of black letters, not only K but also Y, M,
Since the C signal is also included, this 7% M is placed on the edge part.
, C appear, which means that black characters cannot be reproduced with K1 color.

In such a configuration, compared to K1 color reproduction,
There is a problem in that the edges become discolored or muddy due to line thickening, registration deviation, etc., resulting in loss of sharpness and poor image quality.

The present invention is intended to solve the above-mentioned problems, and its purpose is to eliminate the thickness of characters.1. The goal is to reproduce high-quality characters. Another object of the present invention is to improve the sharpness of line drawing images containing thin lines and fine characters. Still another object of the present invention is to efficiently perform edge enhancement processing, and to reproduce a color image without edge color change or muddiness due to line thickening, registration shift, etc. Another object of the invention is to prevent unnecessary colors from being emphasized. Other objects of the present invention are photographs, characters, mesh printing,
The object of the present invention is to switch edge enhancement processing depending on a mixed image, etc., and to adjust the sharpness of a smoothed signal and an edge enhancement signal. Still another object of the present invention is to improve the reproducibility of black characters and the accuracy of edge detection.

[Means and operations for solving the problem] For this purpose, the present invention provides a low-pass filter 1 that removes halftone dot components and smoothes a halftone image, a high-pass filter 2 that detects edges, as shown in Figure @1. Hue (WSY, M%C%B1G
Hue a that detects 8 colors of SRSK) and generates the necessary color signal
Another means 3 and an edge emphasis conversion means 4 for generating an edge emphasis signal, the edge emphasis means 6 generates an edge emphasis signal from the edge detection signal and the necessary color signal, and the synthesis means 7 converts the edge emphasis signal into a low-pass filter. In the edge processing circuit of the image processing device which modulates the output signal along with the output of the image signal, combines it with the recording signal, and outputs the signal, a correction means 5 is provided to extract the edge signal from the image signal and correct the color signal necessary for the hue identification means 3. It consists of:

This necessary color signal correction is performed by generating an edge emphasis signal from the difference between the original image signal and the output of the low-pass filter 1, and using an edge detection signal obtained by binarizing the edge emphasis signal. Further, original image signals of each color are combined to generate a luminance signal, and an edge detection signal is generated from the luminance signal by a high-pass filter 2 and a comparator.

For example, if a character with a hue of W%Y%MSC, B, G, R, etc. is included in the image, the hue detection means 3 detects the hue, and the high-pass filter 2 detects the edge of the character. However, since the necessary color signal of the hue identifying means 3 is further corrected using the edge detection signal, it is possible to obtain the necessary color signal with sharp edges even if the original image signal is dull, thereby preventing characters from becoming thick. I can do it.

Further, a high-pass filter is provided on the input side of the hue identifying means. In this way, coefficients corresponding to the MTF characteristics of IrT can be set individually for each color on the input side of the hue identifying means, and variations among colors can be eliminated.

〔Example〕

Hereinafter, the present invention will be described in detail with reference to Examples.

Table of Contents In this embodiment, a color copying machine will be described as an example of an image processing apparatus, but the present invention is not limited to this, and of course can be applied to printers, facsimile machines, and other image processing apparatuses.

First, prior to explaining the examples, a table of contents will be shown.

(1) Overview of configuration of IPS (image processing system) (II) Hardware configuration of IPS (III>IPS control (rV) Edge processing method (V) Sharpness improvement method (1) Overview of rPs configuration First, image data An overview of the configuration of an IPS (image processing system) that processes images will be explained below.

FIG. 2 is a diagram showing an overview of the module configuration of the IPS.

The color image processing device uses a CCD line sensor at the IIT (Image Human Power Terminal) to read a color original into the primary colors of light, B (blue), G (edge), and R (red), and converts these into the primary colors of toner. The image is converted into Y (yellow), M (magenta), C (cyan), and furthermore K (black or mo), and is exposed to a laser beam and developed at an image output terminal (IOT) to reproduce a color image. In this case, a copy process (
A total of four copy cycles are executed, one copy cycle for MSC and K (pitch) and one copy cycle for each of MSC and K as process colors, and by superimposing the images of these halftone dots, a full color image is created. is being reproduced. Therefore, color separated signals (B, G, R signals)
When converting into toner signals (Y, M, C, signals), how to adjust the color balance and IIT
Problems include how to reproduce the color in accordance with the reading characteristics of the IOT and the output characteristics of the IOT, how to adjust the density and contrast balance, and how to adjust edge enhancement, blur, and moiré.

The engineering PS inputs the BSG and R color separation signals from IIT, performs various data processing to improve color reproducibility, gradation reproducibility, definition reproducibility, etc., and develops color toner. It converts the signal on/off and outputs it to the IOT, and as shown in Figure 2, it performs END conversion (E q
uivalent N neutral density
y; equivalent neutral density conversion) module 301, color masking module 302, [i size detection module 303, color conversion module 304, UCR (Und
er Co1or Re1ova1; Undercolor removal) &
Black generation modicol 305, spatial filter 306, TR
C (Tone Reproduction.

n Control; color tone correction control) module 30
7. Reduction processing module 308, screen generator 309, IOT interface module 310,
It consists of an area image control module 311 having an area generation circuit and a switch matrix, an editing control module having an area command memory 312, a color palette video switch circuit 313, a font buffer 314, and the like.

Then, 8-bit data (256 gradations) for the BSG and R color separation signals from IIT are manually input to the END conversion module 301, and Y is input.

After converting into toner signals of M, C, and C, select the process color toner signal There is. Therefore, in the case of full color (four colors), after the document size is detected, the editing area is detected, and other document information is detected in the prescan, a copy cycle is started in which, for example, the process color toner signal X is set to Y. Then, every time a copy cycle is sequentially executed in which the process color toner signal X is M, signal processing corresponding to four document reading scans is performed.

11T uses a CCD sensor to read 1 pixel for each of B, G, and R at a size of 16 dots/mm, and converts the data into 24 bits (3 colors x 8 pits);
256 gradations). The CCD sensor has a BSGSR filter attached to the top and has 16 dots)
/ mm with a length of 300 mm and a density of 190
．． 16 lines/at a process speed of 5mm/sec
Since it scans m m, it scans 15 m per second for each color.
Read data is output at a speed of M pixels.

Then, at IIT, by log-converting the analog data of pixels of B and G%R, reflectance information is converted into density information, and further converted into digital data.

Next, each module will be explained.

FIG. 3 is a diagram for explaining each module constituting the IPS.

(A) END Conversion Module The END conversion module 301 is a module for adjusting (converting) an optical reading signal of a color original obtained at IIT into a gray-balanced color signal. The amount of toner in a color image is equal in the case of gray, and gray is the standard. However, the values of the B%G and H color separation signals input when a gray original is read from the 11T are not equal because the spectral characteristics of the light source and color separation filter are not ideal. Therefore, END conversion uses a conversion table (LUT: look-up table) as shown in FIG. 3(a) to balance this.

Therefore, the conversion table has the characteristic that when a gray original is read, it is always converted into B, G, and R color separation signals at the same gradation corresponding to the level (black-white) and output. , depends on the characteristics of the IIT. Also,
There are 16 conversion tables, of which 11 are for film projectors including negative film, and 3 are for normal copying, photography, and generation copying.

(B) Color masking modicoll The color masking module 302 has B, G.

By performing matrix calculations on the R signal, it is converted into signals corresponding to the amounts of Y, M, and C toners, and E N
The signal after gray balance adjustment by D conversion is processed.

The conversion matrix used for color masking is a 3x3 matrix that calculates Y and M%C purely from B, G, and R, respectively, but BSG.

Of course, various matrices may be used to take into consideration not only R but also the components of BG, GR%RB%B2, and G3R2, or other matrixes may be used. Two sets of conversion matrices are provided: one for normal color adjustment and one for generating intensity signals in monochrome mode.

In this way, when processing IIT video signals with IPS, gray balance adjustment is performed first and foremost. If this were to be performed after color masking, the conversion table would become more complex because it would be necessary to perform gray balance adjustment using the gray original in consideration of the characteristics of color masking.

(C) Original Size Detection Module Not only standard size originals but also originals of arbitrary shapes such as cutouts or other forms may be copied. In this case, it is necessary to detect the document size in order to select a paper of an appropriate size corresponding to the document size. Furthermore, when the copy paper is larger than the original size, erasing the outside of the original can improve the quality of the copy. Therefore, the document size detection module 303 performs document size detection during briscanning and platen color erasing (frame erasing) processing during document reading and scanning. For this purpose, the platen color is set to a color that can be easily distinguished from the original, for example black, and the upper and lower limits of the platen color identification are set in the threshold register 3031 as shown in Fig. 3.
Set to . During bliscanning, a signal X converted (r-converted) into information close to the reflectance of the document (using the output of the spatial filter 306, which will be described later) and the upper limit value/lower limit value set in the threshold register 3031 are used as a comparator. 3032, and an edge detection circuit 3034 detects the edge of the document and determines the coordinate X.

The maximum and minimum values of y are stored in the maximum/minimum sorter 3035.

For example, as shown in Figure 3 (6), if the document is tilted or not rectangular, the maximum and minimum values (X
I, X2, V+, y*) are detected and stored.

Also, when reading and scanning a document, the comparator 303
3, the Y, M, C of the original and the threshold register 30
31 are compared, and a platen color erasing circuit 3036 erases the reading signal outside the edge, that is, the platen, to perform frame erasing processing.

(D) Color Conversion Module The color conversion module 305 enables conversion of a specified color in a specific area, and as shown in FIG. 3(C), a window comparator 3052
．． Threshold register 3051, color palette 3
053, etc., and when performing color conversion, set the upper and lower limits of each Y, M, and C of the converted color in the threshold register 3051, and set the upper and lower limits of each Y, M, and C of the converted color.
, C values are set in the color palette 3053. Then, the Nantes gate 3054 is controlled according to the area signal manually input from the area image control module, and if it is not the color conversion area, the Y, M, and C of the original are sent out as they are from the selector 3055, and when it enters the color conversion area. , the Y, M, and C signals of the original are sent to the threshold register 30.
51, the selector 3055 is switched by the output of the window comparator 3052, and the converted colors YlM and C set in the color palette 3053 are sent out.

Specified colors can be specified by pointing directly at the document with the digitizer, such as B, B, etc. around the coordinates specified during Briscan.
The specified color is m! by taking the average of 25 pixels each of G, , and R. do.

Through this averaging operation, for example, even a 150-line original can be recognized with an accuracy within a color difference of 5. To read the B, G, and Ra degree data, designated coordinates are converted into addresses from the IIT shading correction RAM and read out. Address conversion requires readjustment for registration lane adjustment, similar to original size detection. In Briscan, the IIT operates in sample scan mode. B and G read out from shading correction R and AM.

1'1 degree data is shading corrected by software, averaged, and then END correction and color masking are performed (from - to window comparator 30
It is set to 52.

Up to 8 colors out of 16.7 million colors can be registered in the color palette 3053 at the same time, and the standard colors are Y, M,
14 colors are available: C, G, B, R, their intermediate colors, and KSW.

(E) If the UCR & black generation module Y and MSC are equal in amount, the result will be gray.Theoretically, the same color can be reproduced by replacing equal amounts of Y, M, and C with black, but in reality Specifically, when it is replaced with black, the colors become muddy and the reproducibility of vivid colors deteriorates. Therefore, the UCR & black generation module 305 generates an appropriate amount of K to prevent such color muddiness, and performs processing to reduce Y, M, and C by equal amounts (undercolor removal) according to the amount. . Specifically, the maximum and minimum values of Y, M, and C are detected, K is generated below the minimum value from the conversion table according to the difference, and a certain value is determined for Y, M%C according to the amount. Undercolor removal is being performed.

In UCR & black generation, as shown in Figure 3(e), for example, when the color is close to gray, the difference between the maximum value and the minimum value becomes small, so the minimum values of Y, M, and C are directly removed and the K However, if the difference between the maximum and minimum values is large, the amount of removal is less than the minimum value of 7%M1C, and K
By reducing the amount of produced black, it is possible to prevent ink from being mixed in and the saturation of low-brightness, high-chroma colors to decrease.

In FIG. 3(f) showing a specific circuit configuration example, the maximum value/minimum value detection circuit 3051 detects the maximum and minimum values of Y and MSC, and the calculation circuit 3053 calculates the difference. K is generated by a conversion table 3054 and an arithmetic circuit 3055. The conversion table 3054 is for adjusting the value of K, and if the difference between the maximum value and the minimum value is small,
Since the output value of the conversion table 3054 becomes zero, the minimum value is directly output from the arithmetic circuit 3055 as the value of K, but if the difference between the maximum value and the minimum value is large, the output value of the conversion table 3054 becomes non-zero. Therefore, the calculation circuit 3055
The value subtracted by that amount from the minimum value is output as the value of K. Conversion table 3056 corresponds to K, Y, , M, C
This is a table for finding values to be removed from Y, M, and C in an arithmetic circuit 3059 through this conversion table 3056. Also, anime game) 305
7.3058 gates the signal and the signal after removing the Y, M, and C undercolor according to each signal of the monocolor mode and 4 full color mode, and selector 30
52.3050 is Y, M,
The choice is between C and C. In this way, colors are actually reproduced using YlM and C halftone dots, so Y
, M, and C and the generation ratio of K are set using empirically generated curves, tables, and the like.

(F) Spatial filter module In the device applied to the present invention, the IIT
Since the document is read while scanning the CCD, if the information is used as is, the information will be blurry.Also, since the document is reproduced by halftone dots, the halftone period of the printed matter and the sampling period of 16 dots)/mm Moiré occurs between the two. Furthermore, moiré occurs between the halftone dot period generated by the user and the halftone dot period of the document. Spatial filter module 306
is equipped with a function to recover such blur and a function to remove moiré. A low-pass filter is used to cut halftone dot components, and a high-pass filter is used to emphasize edges.

In the spatial filter module 306, as shown in FIG.
Y, M, C, Min and Maw-Min as shown in
A selector 3003 extracts one color of the human input signal and converts it into information close to reflectance using a conversion table 3004. This is because it is easier to pick up edges with this information, and for example, Y is selected as one of the colors. In addition, the threshold register 3001.4-bit binarization circuit 3
002, Y, M,
From C, Min and Maw-Min to Y% M, C,
,B,G.

Separates into eight hues: R and W (white). decoder 300
Reference numeral 5 recognizes the hue according to the binarized information and outputs 1-bit information as to whether or not the process color is a necessary color. .

The output of FIG. 3 (A) is manually input to the circuit of FIG. and 5×
7 digital filter 3064, modulation table 3067, and delay circuit 3065 (IiO
Edge enhancement information is generated from the output information. Mojikore
The version tables 3066 and 3067 are selected depending on the copy mode, such as photo and text only, mixed copy mode, etc.

In edge enhancement, for example, when trying to reproduce a character with an edge like ■ in Figure 3 (1) as ■, YlC is
Emphasis processing is performed as shown in , and M is not emphasized as shown in the solid line. This switching is performed by an AND gate 3068. To carry out this process, by emphasizing as indicated by the dotted line (■), the edges become muddy due to the color mixture of M as shown in (■).

The delay circuit 3065 uses a PIF03062 and a 5×7 digital filter 3064 to switch such emphasis using an AND gate 3068 for each process color.
This is to ensure synchronization with the When bright green characters are reproduced using normal processing, magenta is mixed into the green characters, causing them to become muddy. Therefore, when we recognize green as tX as described above, Y
IC is output as usual, but M is suppressed and edges are not emphasized.

(G) TRC conversion module 10T executes 4 copy cycles (in the case of 4 full-color copies) with each process color of Y, M, C, according to the on/off command from IPs, and prints a full-color original. However, in reality, in order to faithfully reproduce the colors theoretically obtained through signal processing, delicate adjustments are required that take into account the characteristics of the IOT. T
The RC conversion module 309 is intended to improve such reproducibility, and addresses 8-bit image data as shown in No. 31!1(j) using each combination of 7% M and C densities. It has a manual address conversion table in RAM, and has editing functions such as density adjustment, contrast adjustment, negative/positive inversion, color balance adjustment, character mode, and watermark composition according to area signals. This R.A.
Bits 0 to 3 of the area signal are used for the upper three bits of the M address. Furthermore, the above functions can be used in combination using the out-of-area mode. Furthermore, this R
AM, for example, consists of 2 bytes (256 bytes x 8 sides) and has an 8-side conversion table, and up to 8 sides are stored during the IIT carriage return for each cycle of YlM and C. Selected according to copy mode. Of course, if the RAM capacity is increased, there is no need to load it every cycle.

(H) Reduction/enlargement processing module The reduction/enlargement processing module 308 uses the line buffer 3083
In the process of temporarily holding and transmitting data
1 for sampling pitch signal and line buffer 3083
Generate read/write addresses for. The line buffer 3083 is configured as a ping-pong buffer consisting of two lines, so that it is possible to read one line and write the next line data to the other at the same time. In the reduction/enlargement processing, the reduction/enlargement processing module 308 performs digital processing in the main scanning direction, but the scanning speed of the IIT is changed in the sub-scanning direction. The scan speed can be changed from 50% to 4 by changing from 2x speed to 1/4x speed.
Can be scaled up to 00%. In digital processing, when reading/writing data to/from the line buffer 3083, data can be reduced by thinning and complementing, and can be expanded by adding and complementing. If the complementary data is located in the middle, it is generated by weighting according to the distance from the data on both sides, as shown in FIG. 1 (1). For example, in the case of data xi', data X l , X l+1 on both sides and the distance jFld+ between these data and the sampling point
, from da, (Xt Xd2) + (Xi, + XI
) However, it is obtained by calculating d + + dz = 1.

In the case of reduction processing, data is complemented and written to the line buffer 3083, and at the same time, the reduced data of the previous line is read out from the buffer and sent. In the case of enlargement processing, the data is written as it is, and the data of the previous line is read in the opposite direction, supplemented and enlarged, and sent out.

If complementary enlargement is performed at the time of writing, the clock at the time of writing must be increased according to the enlargement ratio, but if the above method is used, writing/reading can be performed with the same clock.

In addition, by using this configuration, it is possible to perform shift image processing in the main scanning direction by reading from the middle or by reading with delayed timing, and by repeatedly reading it, it is possible to process the image in the opposite direction. Mirror processing can also be performed by reading from .

(1) Screen generator The screen generator 309 converts the gradation toner signal of the process color into an on/off binary toner signal and outputs it. Binarization processing and error diffusion processing are performed by comparison with . The IOT inputs this binary toner signal and reproduces a half-tone image by turning on and off an elliptical laser beam approximately 80 μmφ in height and 60 μmφ in width to correspond to 16 dots)/mm. ing.

First, the method of expressing gradation will be explained. Figure 3 (n)
A case will be described in which, for example, a 4×4 halftone cell S is configured as shown in FIG. First, in the screen generator, a threshold matrix m is set corresponding to such a halftone cell S, and this is compared with a data value expressed in gradation. In this comparison process, for example, if the data value is "5", the threshold matrix m
A signal to turn on the laser beam is generated in the part below "5".

A 4×4 halftone cell with 16 dots/mm is generally referred to as a halftone dot of 100 spi and 16 gradations, but this results in a coarse image and poor reproducibility of color images. Therefore,
In the present invention, as a method of increasing the gradation, these 16 dots)
A 7 mm pixel is divided vertically (in the main scanning direction) into four parts, and the on/off frequency of the laser beam for each pixel is increased by 1/40 units, that is, by 4 times, as shown in (0) in the same figure. This achieves 4 times higher gradation. Therefore, in response to this, a threshold value matrix n% as shown in FIG. Furthermore, it is also effective to employ a submatrix method to increase the number of lines.

The above example uses the same threshold matrix (IJ) with the only growth nucleus near the center of each halftone cell, but the submatrix method is constructed from a set of multiple unit matrices, ), the growth nuclei of the matrix are set at two or more locations (plurality). If this type of screen pattern design method is adopted, for example, bright areas will have 141 spi and 64 gradations, and as it gets darker, 200 spi and 12828 gradations will be used, so that the number of lines and gradations can be freely adjusted according to dark and bright areas. can be changed. Such a pattern can be designed by visually determining the smoothness, fineness, graininess, etc. of gradations.

When a halftone image is reproduced using a dot matrix as described above, the number of gradations and resolution have a contradictory relationship.

In other words, there is a relationship in which increasing the number of gradations causes a decrease in resolution, and increasing the resolution causes a decrease in the number of gradations. Also,
If the matrix of threshold data is made smaller, a quantization error will occur in the image that is actually output. In the error diffusion process, as shown in FIG.
4, correction circuit 3o95, addition circuit 3091
It uses feedback to improve the reproducibility of gradation when viewed from a macroscopic perspective. For example, it performs error diffusion processing that convolves the corresponding position of the previous line and the pixels on both sides of it through a digital filter. There is.

As mentioned above, the screen generator switches threshold data and error diffusion processing feedback coefficients for each document or area depending on the type of image, such as a halftone image or character image, to improve the reproducibility of high-gradation, high-definition images. .

(J) Area image control module The area image control module 311 has a configuration in which seven rectangular areas and their priorities can be set in the area generation circuit, and area control information is sent to the switch 7 matrix corresponding to each area. Set.

The control information includes color conversion, color mode such as monochrome or full color, modulation selection information for photos and text, TRC selection information, screen generator selection information, etc., including a color masking module 302 and a color conversion module 304. , U.C.
It is used to control the R module 305, spatial filter 306, and TRC module 307. Note that the switch matrix can be set by software.

(K) Edit control module The edit M control module reads a document that is not a rectangle but a pie chart, for example, and enables coloring processing such as filling in a specified area of any shape with a predetermined color. As shown in (e), the AGDC (
Advanced Graphic Digital
Controller) 3121, font buffer 3126, logo ROM 3128, DMAC (DMA
Controller> 3129 is connected, an encoded 4-bit area command is written from the CPU to the brain memory 3122 through the AGDC 3121, and a font is written to the font buffer 3126. Brain Memo IJ 3122
For example, in the case of rooo(lr, command O is used to output the original [draft), each point of the document can be set with 4 bits from plane O to brain 3. The decoder 3123 defoods this 4-bit information into commands 0 to 15.
The switch 7 trix 3124 is used to set commands 0 to 15 to perform fill pattern, fill logic, or logo processing. The font address controller 3125 generates the address of the font buffer y 3126 according to patterns such as halftone shade and hatching shade based on a 2-bit fill pattern signal.

The switch circuit 3127 is a switch matrix 3124
The document data x1 font buffer 3126, color palette, etc. are selected based on the fill logic signal and the contents of the document data X. Fill logic is information for filling only the background (background part of the manuscript) with a color mesh, color converting a specific part (foreground), masking, trimming, filling, etc.

In the IPS of the present invention, as described above, the IIT original reading signal is first subjected to END conversion and then color masking, and processes such as original size, border erasure, and color conversion, which are more efficient when processing with full color data, are performed. After removing the undercolor and generating ink, I focused on process colors.

However, processing such as spatial filtering, color modulation, and TRC1 reduction can reduce the amount of processing by processing process color data compared to processing full color data, and the number of conversion tables used can be reduced to 173. At the same time, we have increased the number of types to increase the flexibility of adjustment, the reproducibility of color, the reproducibility of gradation, and the reproducibility of definition.

(II) IPS hardware configuration FIG. 4 is a diagram showing an example of the hardware configuration of the IPS.

The IPS of the present invention includes two substrates (IPS-A).

The first substrate (IPS-A) is the part that achieves the basic functions of a color image forming device such as color reproducibility, gradation reproducibility, and definition reproducibility.
In addition, the parts that achieve applied and specialized functions, such as editing, are the second part.
It is mounted on the board (IPS-B). The composition of the composer is shown in Figures 4 (a) to (C), and the latter composition is the same as I! I (6
). In particular, if the basic functions can be sufficiently achieved with the first board, applications and specialized functions can be flexibly handled simply by changing the design of the second board. Therefore, if it is desired to further enhance the functionality of the color image forming apparatus, this can be achieved simply by changing the design of the other substrate.

The IPS board has a CPU bus (as shown in Figure 4).
Address bus ADR5BUS, f-9 bus DATABU
S, control bus CTRLBUS) is connected, and I
Video clock IIT-VCLK, line synchronization (main scanning direction,
A horizontal synchronization) signal IIT-LS and a page synchronization (sub-scanning direction, vertical synchronization) signal IIT-PS are connected.

Since the video data is subjected to vibe line processing after the END conversion section, processing is required at each processing stage, resulting in a data delay in units of locks. Therefore,
The line synchronization generation and fail check circuit 328 generates and distributes horizontal synchronization signals in response to delays in each process, and performs a fail check on the video clock and line synchronization signals. Therefore, the line synchronization generation & fail check circuit 328 includes video clock I
IT-VCLK and line synchronization signal 11T/LS are connected, and the CPU buses ΔDR3BUS, DATABUS, CTRLBL are connected so that internal settings can be rewritten.
IS) and chip select signal C8 are connected.

r IT's bidet TF-9B, G, and R are END variant R.
It is manually operated by OM321. The END conversion table may be configured to be D-coded from the CPU using RAM, for example, but there is almost no need to rewrite it while the device is in use and image data is being processed.
Two 2-byte ROMs are used for each of G, R, and R, and a ROM-based LUT (look-up table) method is adopted. It has 16 conversion tables and is switched by a 4-bit selection signal ENDSet.

The output of ROM321 after END conversion is 3 for each color.
Three calculation LSI322s with two Xl matrices
connected to a color masking section consisting of Operation LST
Each bus of the CPU is connected to 322, and 7
) Qx coefficient can be set. A setup signal SU and a chip select signal C3 are connected to switch from image signal processing to a CPU bus for rewriting by CPtJ, and a 1-bit switching signal MONO is connected to select a matrix.

In addition, the power down signal PD is manually input to stop the internal video clock when the FIT is not scanning, that is, when not performing image processing.

B, G, R to Y, M by the calculation LSI 322.

The signal converted to C is inputted to the DOD LS 1323 after color conversion processing through the color conversion LSI 353 of the second board (IPs-8) shown in <ci) in the figure. The color conversion LSI 353 has four color conversion circuits consisting of a threshold register for setting non-conversion colors, a color palette for setting conversion colors, a comparator, etc., and the DOD LSI 323 has a document edge detection circuit, a frame eraser, etc. It owns circuits, etc.

The output of the DOD LSI 323 that has undergone frame erasure processing is U
It is sent to LEi+324 for CR. This LSI is
Including the R circuit, the black generation circuit, and the necessary color generation circuit,
Outputs each number 4z of process color x1, necessary color Hue and edge corresponding to the toner color in the copy cycle. Therefore, this LSI has a 2-bit process color designation signal C0LR and a color mode signal (4C
OLR, MONO) are also manually operated.

The line memory 325 stores the process color X, necessary color Hue, and edge color output from the UCR LSI 324.
It consists of a FIFO for storing four lines of data in order to input each signal of e to a 5×7 digital filter 326, and a FIFO for matching the delay. Here, for Process Color X and Edge, 4
The lines are accumulated and a total of 5 lines are sent to the digital filter 326, and the necessary color Hue is delayed by a FIFO, synchronized with the output of the digital filter 326, and sent to the LSI 327 for MIX.

The digital filter 326 is a 2×7 filter LS
There are two sets of 5X7 filters (low pass LP and bypass HP) each consisting of three I filters, -4, which processes process color X, and the other which processes edge.

The MIX LSI 327 performs halftone removal and edge emphasis processing on these outputs using a conversion table, and mixes them into process color X. Here, EDGE and 5harp are manually input as signals for switching the conversion table.

The TRC 342 consists of a 2 byte RAM holding eight conversion tables. The conversion table is configured to be rewritten using the carriage return period before each scan, and is switched by a 3-bit switching signal TRC3el. Then, the processing output from here is sent to the LS for reduction processing from the transceiver.
Sent to I345. The reduction/enlargement processing section has an RA of 8 bytes.
A ping pong buffer (line buffer) is constructed using two M344s, and an LSI 343 generates a resampling pitch and a line buffer address.

The output of the reduction/enlargement processing section returns to the EDF LSr 346 through the area memory section of the second board shown in FIG. 3(d). E
LS for DF 346 is a DF that holds the information of the previous line.
It has an IFO and performs error diffusion processing using information from the previous line. Then, the signal X after the error diffusion processing is converted to
S I 34? is output to the IOT interface.

In the 10T interface, the signal from the SG LSI 347 input as a 1-bit on/off signal is transferred to the LSI.
349, and sends it to the IOT in parallel in 8 bits.

In the second substrate shown in FIG. 4, the data actually flowing is 16 dots/mm, so the reduced LSr35
4, the image is reduced to 1/4, binarized, and stored in the area memory. Enlargement decoding LS r 359 It has an I-ban RAM 360, and when reading area information from the area memory and generating commands, it enlarges to 16 dots/mm, and it can also be used to generate logo addresses, color palettes, and fill patterns. We are taking steps to deal with the occurrence of. DRAM
356 stores 4-bit coded area information composed of four planes. ACDC 355 is a dedicated controller that controls area commands.

(III) IPS Control (A) VCPU In the present invention, the V CP tJ manages and controls the image data processing system consisting of the IIT and the IPS.

At each stage of image data processing in the IPS, a conversion table (LUT) is used, as described above, to provide flexibility in processing such as image data conversion and correction. In other words, by using a conversion table, you can freely set data for nonlinear conversion and correction, and
By setting the value of the calculation result in advance, a desired calculation value can be obtained quickly by simply reading the conversion table without performing calculation processing. Furthermore, by preparing multiple tables and configuring them so that they can be selected according to the type of image, it is possible to convert and correct image data according to photos, text, printing, and a mixture of these. It is possible to guarantee the reproducibility of specific images depending on the original. Moreover, by using a conversion table, the number of gates and memory capacity in processing circuits for conversion and correction can be reduced, and desired data can be obtained by reading out table data using human data as an address. It is possible to increase the processing speed. The VCPU sets and controls various tables and registers of the LSI that makes up the IPS, and also controls the II
It also controls the image data processing system of T.

VCPU board with VCPU installed (VCPUPWBA)
is an analog board (AN) from the perspective of image data flow.
It is connected after ALOG PWBA), and in addition to VCPU, ITG (IIT timing generator) and SHC (
Various circuits such as a shading correction circuit) are also incorporated. vcpu is the L that configures IPS as mentioned earlier.
In addition to setting and controlling various tables of the SI, it also controls the M(M of the TG and SHC, as well as each circuit built into the analog board).

Therefore, the bus shown in FIG. 4 is the bus of this VCPU, and not only data settings for each register, memory, etc. of the LSI, but also settings for other LSIs are performed from the VCPU through this bus.

The VCPU has basic parameters, and depending on execution conditions such as copy mode,
Write processing is executed at the time of T's carry 1 return (backscan). For example, before prescanning, predetermined data is written to each register and table depending on the copy mode and type of prescanning, and before each copying scan, data corresponding to each developed color M1C, etc. is written. Predetermined data is written to each register and table. Therefore, in a screen generator that changes the screen angle according to the developed color, data is rewritten every time a copy scan is performed. In addition to these, the writing process by the VCPU also includes color masking and UCR.
, TRC, etc. tables, registers, etc., so in order to efficiently write these in the short carriage return time, the VCPU calculates the next data to be written during the scan. ing.

(B) IPS control system configuration Figure 5 shows the layer structure of the IPS control system, Figure 6 shows the configuration of the IPS control system, and Figure 7 shows the S
FIG. 8 is a diagram for explaining the open communication between the YS and the TPS, and is a diagram for explaining the relationship between the scan operation and the IPS settings.

Necessary information is set according to the copy mode and other modes by the IPS control system consisting of each LSIft and VCPU74a of the above IPS, but the layer of the It has a three-tiered structure: a layer, and below that is a Senjiba layer and an application layer. The monitor is
It creates communication clocks and other necessary locks, and also periodically runs the application and counts the timer to monitor whether the work registered by the application is being executed accurately.Furthermore, when an interrupt occurs, it sets the address of the interrupt. It functions as the original O5, which calls the ``O5'' and executes predetermined tasks. The application has a fixed job to perform on the rPS, and
The IPS system handles the execution and termination of the LSI through handshaking.

The configuration of the IPS control system (VCPU) is as shown in Figure 6.
, lower module module that has the function of setting necessary information corresponding to each LSI such as CC (color masking), PS system, sending buffer, receiving buffer that instructs the content, timing, etc. to be set for each of these lower modules. , consisting of a monitor. The IPS system performs NV depending on the situation such as power-on, start, cycle change, etc.
Instructs setting items from M information and ROM parameters and performs handshake. That is, the IPS system sequentially calls each module to instruct execution of LSI settings. In contrast, when each module finishes execution, it reports its completion to the IPS system.

Communication is performed between the monitor and the IPS system via a transmission buffer and a reception buffer T. For example, the monitor is S
When a certain command is sent from YS, a reception interrupt is generated, and the command is set in the reception buffer and a flag is set.

Then, the PS system saves the receive buffer)
It looks at the given command and calls the necessary lower module. When the lower module reports the completion of execution, ■P
The S system sets completion information in the transmission buffer and raises a flag, and the monitor reports the completion as IPS READY to sys'' by a transmission interrupt.

Between SYS and LPS ('v'cP[J), the seventh
As shown in the figure, first, when the power is turned on, the NVM command is sent and the NVM command necessary for IFS, such as the register adjustment value and magnification adjustment value, is sent.
VM information is set in IPS. The completion of this setting is reported to the SYS by the end thread and r P S It, IPS READY. Then, by copy start, S
If there is a CAM INF command, a BASiCC0PY command, and a 4i collection, an EDIT MODE command is sent.

The 5CAN INF command gives information on whether it starts with a normal copy scan, a scan that detects the original, and other types of copy scans. , contrast, density, sharpness mode), paper size, magnification, etc., and EDIT
The MODE command provides information such as area specification coordinates, editing function, color conversion, etc. if an area is specified. Then, once the required run length is taken, the SYS
IPS sends MC5TOP to S, IPS sends IPS REA
Respond to this with DY.

The BASICCOPY command has a number indicating the development order. This development order has several combinations such as 4-color full color, 3-color full color, monocolor, etc. In IPS, as shown in Figure 7 Q')), information on the number of development cycles and the development order is Hold it on the ROM table,
Information in the ROM table is selected by setting the number to the pointer. Then, the development cycles are counted by the IIT Page Sync PS using a counter.

For example, the number of development cycles is "4" and the development order is "1 (M) 2 (C) 0 (Y) 3 (
K) When the number J is low, the count "1" gives M, the count "2" gives C, and when the count reaches "4", the next one is 1.
1" and turn it until the IPS 5TOP command is issued.

Next, the lower modules of the IPS system will be explained. These operations are performed by M and CS as shown in Fig. 8(a).
Assuming that scanning is performed in the development order of Y%, processing for setting the LSI is performed one cycle prior to this development cycle.

For example, setting up spatial filters, TRC, etc. takes time to calculate the setting values, so the IPS system
The next color is determined, the lower module calculates and sets the data corresponding to the next color, and the OCR is used to select the developing color. At the start, when a BASICCOPY command is received, the signal to be developed first is determined from the color mode number, and Y, for example, is entered in the development cycle variable. As shown in the same figure, etc., for END, table selection, color selector, development color Y set, spatial filter, filter 1 coefficient multiplication table set, TRC, Y table calculation and Start by setting the magnification for R/E, setting the buffer output effective range and image shift amount for the line buffer, and setting the TRC table, spatial filter mode, and color mode for area memory 1. Since the contents of each process to be performed at the time are stored in the ROM, each lower module is sequentially called according to the contents, instructions for setting each LSI are given, and setting information is passed. When each setting is completed and a job end is received from each lower module, the IPS system sets the developing color in the SE bit of the UCR control register, changes to the next color for data calculation processing, etc., and sends the IPS to SYS. Return R, EA, DY.

In this way, the IPS system receives an interrupt at the rising edge of the IIT page sync PS and precedes the developed color by one cycle. For example, for a spatial filter, the magnification, sharpness adjustment value, sharpness mode,
Information on development colors is passed, and information on color balance adjustment, contrast adjustment, density adjustment, character mode, negative/positive inversion, watermark composition, monocolor, OHP, etc. is passed to the TRC.

(rV) Edge processing type digital color copiers are capable of handling photographs, fine prints, characters,
Various manuscripts such as line drawings are input. As previously explained in the section regarding the spatial filter module, in the image processing apparatus according to the present invention, the edge enhancement signal and the smoothing signal are appropriately mixed to improve reproducibility corresponding to the type of image.

FIG. 9 is a diagram for explaining one embodiment of the edge processing method of the present invention, and FIG. 10 is a diagram showing an example of the configuration of an edge processing LUT.

As shown in FIG. 9(a), the edge processing circuit of the present invention basically includes a bypass edge detection filter 361, a hue detection circuit 362 that detects the hue of the original image signal, and an edge enhancement LUT (lookup table). )■ and an edge attenuation LUT■, it is equipped with an edge processing LUT 363 that converts the edge detection signal, detects edges, and generates an edge processing signal that emphasizes necessary colors and attenuates unnecessary colors. ing. The switching signal of the hue detection circuit 362 indicates whether the developed color is a necessary color or an unnecessary color, and the switching signal of the necessary color causes the edge processing LUT 363
is switched to the edge enhancement LUT■, and the edge processing LUT363 is switched to the edge attenuation LUT by the unnecessary color switching signal.
Switch to UT■. In the edge processing LUT 363, when switched to the edge enhancement LUT ■, the edge detection signal is converted into a signal that emphasizes edges, and when switched to the edge attenuation LUT ■, the edge detection signal is converted into a signal that attenuates edges. is converted to Therefore, when an edge is detected, the edge is emphasized if the color is a necessary color, and the edge is attenuated if the color is an unnecessary color.

The edge processing LtJT363 uses the beak axis as the output value of the edge detection filter, that is, the value representing the degree of edge, and the vertical axis as L.
UT output value, for example, if the filter output value is ±100
The following is determined as a halftone image part (- sets the emphasis amount to 0, ±1
1] The value outside of 0 is determined to be positive/negative and is given an emphasis value or an attenuation value.

Therefore, the edge attenuation LOT2 gives an attenuation value when the output value of the filter is outside ±100, as shown in Figure (a). In other words, in this case, L[
, JT's output value is set.

According to the above configuration, for example, for a filter input signal with a black hue as shown in FIG. M and C are made unnecessary colors, and not only are they not emphasized, but they are also attenuated according to the amount of edges, as shown by the solid line. The edge processing LUT 36 converts the edge detection signal into an edge processing signal that emphasizes necessary colors and attenuates unnecessary colors according to the amount of edges.
It is 3. On the other hand, the smoothing circuit generates smoothed signals for each color as shown in FIG. 9(C). Therefore, in the smoothed signal, the edge portions of each color are blurred, but by combining this portion with the edge enhancement processed signal, unnecessary colors are almost attenuated and the filter output is not output, as shown in (d) of the same figure. A signal can be obtained, and black characters can be reproduced almost exclusively in K1 color. This eliminates the color mixture at the edge portion as described in the section of (H(F)), and YSM%C, B%G.

With RSW, it is possible to reproduce clear characters even at the edges.

Note that the above two edge processing LITs may be constituted by one LUT by operating them.

An example of this is shown in FIG. 10(b). In this example, when applying to data with 256 gradations, 12
-126 ~ by setting the minimum resolution to 2 gradations at 8
0 to +126 is an unnecessary color area for attenuation, and -
256 to -1, 28 and 128 to 254 are the necessary color areas for emphasis.

FIG. 11 is a diagram showing an example of a hardware configuration using [, Sl] of the nonlinear filter section, and is a diagram for explaining the operation of the circuit shown in FIG. 12 or FIG. 11.

This example uses UCR-LSI365 and print color No. 47d.
Edge signal edge and required color (hue) from ata 7g
The signal Hue is separated from the image data data and the edge signal e.
dge are manually input to digital filters 366 and 367, respectively.

The digital filter (ME-LOT) 366 is a low-pass filter that has a characteristic of dulling (blurring) the edge portions of the image as shown in '1126(a). Sharpness adjustment is achieved by changing the settings of this parameter and changing the characteristics. Further, the digital filter (USM-LIJT) 367 is a high-pass filter, and has a characteristic of extracting edge portions of an image as shown in FIG. 12(b). Then, image data data, output signals of digital filters 366 and 367, necessary color signals) 1ue are MIX-LSI 368
are input and mixed. This MIX-LSI 368 has a smoothing LIT (ME-MODU-LUT) and an edge processing LtJT (USM-MODU-LUT).

The smoothing processing LUT has two strength and weakness tables as shown in the 12th rgJ (C), and performs modulation corresponding to the strength and weakness as shown in FIG.
For example, T can be used as an edge emphasis LUT as shown in FIG. 12(d).
Hold a table of three strong, medium and weak cards as shown in the figure (f
), the modulation corresponds to strong, medium and weak frequencies. In addition, as explained earlier, each edge processing LtJ
In T, the LUT for edge emphasis is set using the hue signal) 1ue.
Switching between the edge attenuation LUT and the edge attenuation LUT is performed.

In Fig. 11(a), addition is performed through an LUT for edge emphasis, or output is directly set to zero, or LtJT is used for edge attenuation.
This figure shows a configuration applied to the circuit of FIG. 18 or 9 which performs addition through two LUTs, or switches and controls two LUTs. On the other hand, in Fig. 11 and others), the output data width from the digital filter 367' is reduced by 1 bit, and the hue signal Hue is added to it by 1 bit to make 8-bit data, which is used as manual data for the MIX-LSI 368'. It is the first in MIX-LSI368'.
Hue signal) 1ue with one LUT as shown in Figure O et al.
Edge addition control is performed by selectively using the necessary color area (LUT for emphasis) and the unnecessary color area (LUT for attenuation) by turning on/off the .

That is, the data width from the edge detection unit 367' is the sign s (+or-) of 1 bit and the data of 7 bits (dJ
sdJsdad+do), the lowest pitch is)d. , shift the whole bit to the right one by one, and add a Hue signal to the most significant bit. Therefore, the input data after changing the pit assignment is r Hs dJ4dsdJ
+deJ has an 8-bit configuration.

When applied to the conventional example shown in Fig. 18, the edge data can be added through the emphasis LUT by preparing only one emphasis LUT, or it can be added by directly outputting zero, so unnecessary colors can be avoided. An edge attenuation LUT cannot be used. However, as mentioned above,
By configuring the UT and the edge attenuation LUT into one LUT (Fig. 10 et al.), and manipulating the data width from the edge detection section and adding the necessary color signal Hue,
The present invention can be realized by simple hardware changes as shown in FIG.

Note that edge enhancement based on hue determination has a great effect on improving the reproducibility not only of black characters but also of colored characters. For example, if the magenta M character is created manually, the 121st! As shown in l(g), the area where the hue is determined to be M is emphasized, and the other areas are attenuated, so that the gentle gradient part is shaved off. Therefore, compared to the conventional method that emphasizes only the necessary color portions, the outer W portion is removed as an unnecessary color, making it possible to reproduce sharp characters. Moreover, as described above, the same effect can be obtained for 7%M, C% and %B%G, R, W for hue detection.

(V) Sharpness Improving Circuit The sharpness of an image is expressed as the rise of the outline of the image, the presence or absence of fine contrast, large-area contrast, or quality of black reproducibility. By performing the edge processing as described above, it is possible to eliminate the muddiness of the characters at the edges and make the edges sharp, but it cannot solve the problem of increasing the size of the characters.

Therefore, in the present invention, a sharpness improvement circuit is added to the edge processing circuit described above to sharpen edge detection in hue determination, thereby preventing the thickening of characters and improving the reproducibility of fine characters.

131st! I is a diagram showing the configuration of one embodiment of the sharpness improvement method of the image processing apparatus according to the present invention, and FIG. 14 is a diagram for explaining the sharpness improvement effect.

13th I! In I, the selector 372 includes the UCR circuit described above, and is a circuit that performs black plate generation, undercolor removal, and selects a development color. Low bass filter 373
is a smoothing filter that performs signal processing to remove halftone dots and reproduce a smooth halftone image, and the γ conversion 374, edge detection filter 375, and LUT 376 are processing circuits that detect edges and generate edge emphasis signals. be.

In the above circuit, the additional circuit of the present invention is the subtraction circuit 3
77, multiplication circuit 378, addition circuit 379, comparator 380. It consists of 383 circuits,
A hue signal correction circuit is configured for the nonlinear filter processing circuit.

A subtraction circuit 377 is connected to the input side and output side of the low-pass filter 373, and calculates the difference between the original image signal and the output signal of the low-pass filter 373. A multiplication circuit 378 is connected to the output side of the subtraction circuit 377. The adder circuit 379 adjusts the gain by multiplying the difference between the original image signal and the output signal of the low-pass filter 373 by
It is connected to the output side of the subtraction circuit 77 and the input side of the low-pass filter 373, and adds the output signal of the subtraction circuit 377 and the original image signal. Then, the comparator 380 compares the output (:) of the adder circuit 379 with the threshold value th and converts it into a binary value, and the AND gate circuit 383 compares the binary output of the comparator 380 with the hue signal (required color signal )
The hue signal is corrected by performing an AND operation with

Next, to explain the overall operation, in the subtraction circuit 377,
Edge detection is performed by determining the difference between the output signal of the low-pass filter 373 and the original image signal. That is, the combination of the low-pass filter 373 and the subtraction circuit 377 shown in LP in FIG. 14 creates a high-pass filter with the characteristic HP. Since the edge detection signal obtained by the subtracting circuit 377 is too weak to be used as it is, the multiplier circuit 378 multiplies it by 2 to adjust the gain.

Furthermore, since the DC component is cut in the output of this multiplier circuit 378, by adding the original image signal to this gain-adjusted edge detection signal in an adder circuit 379, the result as shown in e of FIG. 14(b) is obtained. An edge-emphasized signal E dg containing a DC component and emphasizing and sharpening the edge portion.
Generate e Edge =k (a-b) +a. Then, this signal Ec1ge is compared with a threshold value th by a comparator 380 to generate a binary signal shown by diagonal lines in FIG. Performs a logical AND operation. By turning on/off the corrected hue signal Hue, the output of the edge detection circuit 375 is converted by the LUT 376, and is synthesized with the output of the low-pass filter 373 by the synthesis circuit 381, thereby producing a hue signal with sharp edges and no thickening. can be generated. Furthermore, by applying Moe's table to this LUT 376, edge addition control can be performed by selectively using the necessary color area (LUT for emphasis) and the unnecessary color area (LLIT for attenuation). Note that the gain is changed depending on each color of Y, M, and C.

FIGS. 15 and 16 are diagrams showing other embodiments of the sharpness improvement method of the image processing apparatus according to the present invention.

In the case of the above configuration, the problem of font enlargement can be efficiently solved by simply adding a simple circuit to both the input and output sides of the low-pass filter 373, but there is no flexibility in adjustment. Therefore, the 15th embodiment shows another embodiment of the present invention in which adjustment can be made individually according to each color.
FIG.

In other words, the thickness of the letters is determined by IIT's MTF Special Admiration and IO
It varies considerably depending on the development characteristics of T. For example, in IIT, as shown in FIG. 33(a), there are differences in MTF characteristics depending on the main scanning direction and the sub-scanning direction, and depending on the color, green is relatively good, but blue and red are bad. Therefore, if processing parameters cannot be changed due to these variations, variations will occur depending on the scanning direction and color. Also, in IOT, development characteristics vary depending on the scanning direction, frequency, contrast, etc.

Therefore, the sharpness of characters reproduced between colored characters also changes.

In the above embodiment, the smoothing low-pass filter 373
Since the high-pass filter for edge detection is created by combining the subtraction circuit 377 with the subtraction circuit 377, it is not possible to ignore the smoothing low-pass filter 373 and freely change the characteristics of the high-pass filter for edge detection. In other words, the characteristics are determined by the low-pass filter 373.

Therefore, as shown in FIG. 16, a high-pass filter 385 is separately connected to the input side of the hue discrimination circuit 382 so that the characteristics of the high-pass filter can be freely changed for each original image signal. When the high-pass filter 385 is provided for hue detection in this way, it is possible to freely set the gain for each color! : By changing the parameters individually for each developing color, it is possible to reproduce well-balanced and sharp characters and line drawings in YSM, C, and other colors. That is, as explained in the explanation 1 above, in the past, the hue λ was classified using a distorted signal, but since a signal with sharp edges is manually inputted in the hue discrimination circuit 382, it is possible to obtain a hue signal without thickening of characters. I can do it. Although this increases the scale of the circuit configuration, it is possible to set and change filter parameters for each color, and edge enhancement with well-balanced color can be performed.

Further, in the example shown in FIG. 16, in the additional circuit shown in FIG.
79 is a synthesis circuit 386 for each color signal, and a high-pass filter 3
It was changed to 87. The synthesis circuit 386 generates an equivalent luminance signal that is easy to detect edges by performing the calculation of A=k l Y+, 2 M+, and 3 C, for example, and passes this signal A through the noise filter 38.
7 human power signal. In this case, each coefficient is, for example,
=0.11, k2=0.59, and ks=0.3, and by changing these coefficients individually, the characteristics of each color can be adjusted. Furthermore, the high-pass filter 387 is set with a coefficient including direct current, and is switched for each developing color. With this configuration, the coefficients of the synthesis circuit 386 and the noise bus filter 3 for each color are
87 coefficients can be changed.

〔Effect of the invention〕

As is clear from the above description, according to the present invention, edges are detected using a high-pass filter and the hue (No. 3) is corrected, so it is possible to eliminate thick characters and
Originals such as i1 strokes can be reproduced with high sharpness.

Further, by changing the filter coefficients according to MTF deterioration of the FIT and variations in the development characteristics of the IOT, it is possible to stably reproduce characters and line drawing images.

[Brief explanation of drawings]

FIG. 1 is a diagram for explaining an embodiment of the sharpness improvement method of an image processing apparatus according to the present invention, and FIG. 2 is a diagram for explaining an embodiment of the sharpness improvement method of an image processing apparatus according to the present invention, and FIG. 3 is a diagram for explaining each module that makes up the IPS, FIG. 4 is a diagram showing an example of the hardware configuration of the IPS, and FIG. 5 is a diagram showing the layer structure of the IPS control system. Figure 6 is a diagram showing the IPSM control system, Figure 7 is a diagram explaining communication between SYS and IPS, and Figure 8 is a diagram showing the relationship between scan operation and IPS settings. ! 9 is a diagram for explaining one embodiment of the edge processing method of the present invention, FIG. 10 is a diagram showing an example of the configuration of an edge processing LUT, and FIG. 11 is a diagram showing a configuration example of the edge processing LUT. FIG. 12 is a diagram showing the hardware configuration; FIG. 12 is a diagram for explaining the principle of improving color character reproduction; FIG. FIG. 14 is a diagram for explaining the sharpness improvement effect, FIGS. 15 and 16 are diagrams showing the configuration of other embodiments, FIG. 17 is a diagram showing the configuration of a digital color image processing device, and FIG. 18 19 is a diagram showing a configuration example of a conventional edge enhancement processing circuit, FIG. 19 is a diagram showing a configuration of a hue detection circuit, and FIG.
FIG. 0 is a diagram for explaining thickening of characters, and FIG. 21 is a diagram for explaining edge emphasis processing. DESCRIPTION OF SYMBOLS 1...Low pass filter, 2...High pass filter, 3...Hue identification means, 4...Edge detection means, 5
. . . correction means, 6. edge enhancement means, 7. composition means. Applicant Fuji Xerox Co., Ltd. Agent Patent Attorney Ryukichi Abe (5 others) Figure 1 Figure 3 (d) (e) Figure 3 (f) C-Pi Figure 3 (k) Figure 3 ( n) Fig. 3 (p) (q) Fig. 4 (c) Fig. 5 M6 Fig. 7 (b) Fig. 9 (b) (ko) g@m (d) Area 1 - 11 Figure (a) (b) L-+++-++-----J Figure 12 (a) Figure 12 (b) Figure 12 (f) Figure 12 (9) Figure 12 (c) Figure 12 Figure (d) Figure 12 <e> Noooo (- Figure 15

Claims (8)

[Claims] (1) A low-pass filter that removes halftone components and smoothes the halftone image, a high-pass filter that detects edges, a hue identification unit that detects hue and generates necessary color signals, and edge enhancement that generates edge enhancement signals. In an edge processing circuit of an image processing device, which has a conversion means, generates an edge emphasis signal from an edge detection signal and a necessary color signal, modulates it together with a low-pass filter output, combines it with a recording signal, and outputs it, the edge signal is extracted from the image signal. A sharpness improvement method for an image processing device, characterized in that a necessary color signal of a hue identification means is extracted and corrected. (2) A claim characterized in that an edge emphasis signal is generated from the difference between the original image signal and the output of the low-pass filter, and the necessary color signal is corrected using an edge detection signal obtained by binarizing the edge emphasis signal. Item 1. Sharpness improvement method for an image processing device according to item 1. (3) A luminance signal is generated by synthesizing the original image signals, and an edge detection signal is generated from the luminance signal using a high-pass filter and a comparator to correct the necessary color signal. Sharpness improvement method for image processing equipment. (4) The sharpness improvement method for an image processing apparatus according to any one of claims 1 to 3, wherein the generation coefficient of the edge signal is changed depending on the color. (5) A low-pass filter that removes halftone components and smoothes the halftone image, a high-pass filter that detects edges, a hue identification unit that detects hue and generates necessary color signals, and edge enhancement that generates edge enhancement signals. In an edge processing circuit of an image processing device that has a conversion means, generates an edge emphasis signal from an edge detection signal and a necessary color signal, modulates it together with a low-pass filter output, combines it with a recording signal, and outputs it, the input side of the hue discrimination means A sharpness improvement method for an image processing device, characterized in that a high-pass filter is provided in the image processing device. (6) A sharpness improvement method for an image processing apparatus according to claim 5, characterized in that coefficients of a high-pass filter provided on the input side of the hue identifying means can be changed for each color. (7) The sharpness improvement method for an image processing apparatus according to claim 1 or 5, further comprising a plurality of selectable edge processing lookup tables as the edge emphasis conversion means. (8) The sharpness improvement method for an image processing apparatus according to claim 7, wherein the edge processing lookup table is configured to emphasize necessary colors and attenuate unnecessary colors.