JPH03219777A - Method for generating normalized curve - Google Patents

Abstract

PURPOSE:To manage the shape of a normalized curve at each density with a comparatively few parameters by deciding the shape of the normalized curve a highlight part and a middle density region with a highlight density and a dummy shadow density. CONSTITUTION:Picture information of an original picture is given to an input circuit 4 as to each color component of B, G, R and converted into digital color density signal DB, DG, RR for B, G, R. The signals DB, DG, DR are stored in frame memories 13B, 13G, 13R for each picture element at prescanning. An information processor 14 is, e.g. a microcomputer and fetches the signals DB, DG, DR from the memories 13B, 13G, 13R to generate a normalized curve for each color component. A CPU 15 and a memory 16 are used in this processing. A console 17 inputs various commands and data manually. The normalized curve generated by the unit 14 is loaded to lookup table memories 5B, 5G, 5R as digitized data for each color component.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a normalization curve generation method, and particularly to a method suitable for automatically generating a normalization curve in a plate-making scanner or the like. Also, as one of its applications,
The subject matter is a method for systematically correcting color cast only in the shadow areas of a color original image in a color image processing device such as a color scanner for plate making.

[Conventional technology]

In a color scanner for plate making, etc., after color-separating an image of a color original, each color component is converted into a halftone signal to record a duplicate image. If there is a color cast in the original image, the color separation conditions are modified based on the operator's judgment to ensure gray balance in the recorded image.

The first conventional method for correcting the color separation conditions is a process of converting the density signal for each color component obtained by reading the original image to fit the density range of the color separation device, that is, normalization. In the conversion, the highlight density value and shadow density value are set to different values for each color component.

In addition, in the second conventional method, the highlight density value and shadow density value are common values for each color component, and yellow (Y),
At the stage of obtaining halftone dot signals for each of magenta (M), cyan (C), and black (K), the halftone dot area ratio for each color is corrected.

[Problems to be Solved by the Invention] There are roughly two types of problems in such conventional methods.

One of them is that since correction of color separation conditions is basically performed based on manual operations by an operator, a skilled operator is required to make appropriate corrections.

This kind of problem is not only solved by correcting the color separation conditions.
It also exists in normalization conversion for monochrome images.

Another problem is the above-mentioned problem specific to gray balance. In other words, the above two conventional methods are suitable for cases where a color cast occurs in the low-medium density area due to bias in the light emission characteristics of the light source when photographing a subject to create an original image, or when the original image is discolored overall. , the effect of improving gray balance is recognized.

However, if the maximum expressible density of the original film itself differs for each color component, and there is a color cast only in the shadow areas, it is difficult to make appropriate corrections using these conventional methods. It is. In other words, in each of the above conventional methods, corrections are made uniformly over a wide range of densities;
If E is attempted, the effect of the modification will also be applied to the low-medium density area, resulting in loss of gray balance in the low-medium density area.

In order to prevent this situation and remove the color cast only on the high density side, it is necessary to use UCA/UCR in the color separation device.
You can use the (undercolor addition/removal) function. By using this function, it is possible to adjust the dot area ratio for each color plate for gray components having a predetermined density or higher.

However, when using the U CA/U CR function, the lower threshold of the density range to be corrected must be manually set for each color plate. Furthermore, since the amount of correction is based on the darkest part in the original image, the determination of the amount of correction largely depends on the intuition of the operator. That is, U CA
/U The color cast removal work by CR is done on an ad hoc basis and is not objective or systematic. For this reason, it is not possible to systematically and quantitatively manage the amount of color cast correction by UCA/UCR, which hinders automation of color separation conditions.

In addition, since the U CA/U CR circuit is provided in the color separation device after the black plate signal generation circuit, the U CA/U CR circuit is
Each circuit located before the A/U CR circuit must process the signal before gray balance correction, making the setting of processing characteristics complicated.

Furthermore, no technology has been developed to uniformly handle gray balance correction when the entire original image is faded, and correction of color cast only in shadow areas, and these must be considered separately. There are also problems.

[Purpose of the invention]

The present invention is intended to overcome the above-mentioned problems in the prior art, and the first object is to provide a normalization curve generation method that can be performed systematically without the need for a skilled operator and is particularly suitable for automation. The purpose of

A second object is to provide a normalization curve generation method that is applied to modifying color separation conditions for a color original image and can systematically and quantitatively manage the amount of color cast correction only in shadow areas.

A third object of the present invention is to provide signals after gray balance correction to each circuit in the color separation apparatus, thereby facilitating setting of processing characteristics in these circuits.

Furthermore, it is possible to uniformly handle gray balance correction when the entire original image is faded and color cast correction only in shadow areas, thereby providing a technology that is particularly suitable for automating color separation conditions. The fourth purpose is to

The fifth objective is to achieve the first objective and, in particular, to prevent the density expressivity of the converted image from being greatly affected by the density range of the original image! The purpose is to generate a normalized curve that can be used for 1 unit.

[Means to solve the problem]

A first configuration of the present invention is to create a normalization curve for converting the density value of each pixel of an image of an original image to be processed into a value within a normalized density range suitable for image processing in a predetermined image processing device. Generally applied when generating. In this method, (a) for the original image to be processed, a highlight density value and a shadow density value are determined from a range that the density value before conversion can take, and (b) according to a predetermined determination rule, the A dummy shadow density value is determined for the density value before conversion.

(C) And as the normalized curve, (c-1)
A first point that substantially passes through both a first point determined according to the highlight density value and a second point determined according to the shadow density value on the coordinate plane in which the normalized curve is to be defined. (c-2) In the density region near the highlight density value, both the first point and the third point on the coordinate plane determined according to the dummy shadow density value; The second condition is that it varies along a straight line passing through,
A curve that satisfies both is generated as a normalized curve.

In the second configuration of the present invention, when color-separating the color original image pixel by pixel using a predetermined color separation device,
The first configuration is included in the process of generating a normalization curve for each color component for converting the density value for each color component of each pixel into a value within a normalized density range suitable for color separation in the color separation device. The method according to is applied. Since the original image is a color original image, each of the highlight density value, shadow density value, color shadow density value, and normalization curve is determined for each color component.

If the step (b) requires (b-1) color cast correction of only the shadow part of the color original image,
A value deviated from the shadow density value by a value corresponding to the amount of color cast is obtained for each color component, and these are used as the dummy shadow density value for each color component, eliminating the need for color cast correction only for the shadow part. In this case, the dummy shadow density value is substantially the same as the shadow density value]. That is, this method is configured to perform color cast correction only on shadow portions when normalizing and converting the density values for each color component of the original image.

In the third configuration, in the second configuration, (d)
In order to generate a normalized curve, (d-1) having a highlight density value, a shadow density value, and a dummy shadow density value as parameters, and having the density before conversion as a variable; (d-2) the shadow When the density value and the dummy shadow density value are different values,
is nonlinear with respect to the variable, and (d-3)
A function that is linear with respect to the variable when the shadow density value and the dummy shadow density value are the same value is determined in advance.

Then, (e) the normalization curve is determined by (e-1) determining whether or not the color original image requires color cast correction only in the shadow area, and (e-2) color cast correction only in the shadow area. When the shadow density value and the dummy shadow density value are set to different values, the parameters of the function are specified, and (e-3
) When color cast correction of only the shadow portion is not required, the correction is made by setting the shadow density value and the dummy shadow value to the same value and specifying the parameter of the function.

On the other hand, the fourth configuration is applicable regardless of whether the original/image to be processed is a monochrome original or a color original, and in the method of the first configuration, the step (b) is (b-
1) It includes the step of adding a value proportional to the density range width of the standard original image and a value proportional to the highlight density value, and setting the obtained value as a dummy shadow density value. Second
．． Unlike the third configuration, this fourth configuration does not target color cast correction only in shadow areas.

In the fifth configuration, in the fourth configuration, the weighted average value of the shadow density value and the density range width of the standard original image is set to a value proportional to the highlight density value of the target original image determined in step (a). A dummy shadow density value is determined as the added value.

Note that "density" in this invention refers not only to the optical density but also to the signal level obtained by photoelectrically reading it, the Munsell value, and other quantities that express the gradation density of the image. It is a term.

[Effect]

In the first configuration of the present invention, the first to third points are specified on the normalized transformed coordinate plane according to the highlight density value, shadow density value, and dummy shadow density value. In the highlighted portion on this coordinate plane, the normalized curve has a shape along the straight line connecting the first and third points. In the shadow part, the burning curve deviates from this straight line and passes through the second point.

In the second configuration, the shadow density value and the dummy shadow density value are determined such that the difference between them corresponds to the amount of color cast only in the shadow portion of the original image. Therefore, the deviation of the normalized curve from the straight line has the effect of removing color cast only in the shadow portion. Since this shift substantially occurs only in or near the shadow portion on the coordinate plane, the gray balance in the low-medium density region is not disrupted.

In the third configuration of the present invention, a normalized curve is generated by specifying parameter values of a function that satisfy the conditions (d-1) to (d-3). This function can be used in both cases where color cast occurs only in the shadow area and when color cast does not occur.

In the fourth configuration, by using the value of the density range width of the standard original image in the process of determining the dummy shadow density value, even if the density range of the original image to be processed changes, the highlight density can be maintained in the highlight and intermediate density areas. Concentration values for which the difference in concentration from the two values is the same are converted into substantially the same or relatively close concentration values by iE normalization.

Furthermore, in the fifth configuration, by changing the weighting coefficient in the weighted average, it is possible to change the dependence of the dummy shadow density value on the value of the density range width of the standard original image.

〔Example〕

A. Overall structure and general operation of the first embodiment FIG. 1 is an external view of a color scanner for plate making to which the embodiment of the present invention is applied, and FIG. 2 is a block diagram showing its internal structure.

In FIG. 1, this plate-making scanner 100 is constructed by electrically connecting a scanner main body 200 and a color separation auxiliary device 300 via a transmission cable 101. The scanner main body 100 has an external configuration that is roughly divided into a human power section 201 and an output section 202, and the input section 201 includes an original image drum 2 and a pickup head 3. The output section 202 also includes the recording drum 11 and the recording head 1.
In the illustrated example, the operation panel 203 of the scanner main body 200 is provided on the input section 202 side.

The scanner main body 200 itself does not have a function of automatically setting color separation conditions, and this function is achieved by the color separation auxiliary device 300. Color separation auxiliary device 3
00 has a CRT 18 and a console 1 on top of the device main body 301.
It has an appearance with 7.

Refer now to FIG. Among the elements shown in FIG. 2, the elements other than those inside the broken line frame indicating the color separation auxiliary device 300 are components of the scanner body 200 of FIG. 1. A positive color original film 1 is wound around an original drum 2, and a pickup head 3 is provided opposite the drum 2. Main scanning and sub-scanning of the original image 1 are achieved by the rotation of the drum 2 in the α direction and the translation of the pickup head 3 in the β direction, and the image of the original image 1 is photoelectrically scanned pixel by pixel by the pickup head 3. be read.

The image information of the original picture is given to the input circuit 4 for each color component of blue (B), green (G), and red (R), and in this input circuit 4, the digital color density for each of B, G, and R is determined. Signal DB.

It is converted into D and DH.

At the time of pre-scanning, which will be described later, these signals and D
, DR are stored for each pixel in the frame memories 13B, 113G, 3G, and 13R. Information processing device 1
4 is a microcomputer, for example, which receives these signals from frame memories 13B, 13G, 13R, and D.
, DR is taken in and a normalized curve for each color component is generated. The details of this process will be described later, but the CPU 15 and memory 6 are used in this process. Further, the console 17 is used to manually input various commands and data to the information processing device 14 .

The normalized curve created by the information processing device 14 is
It is loaded into lookup table memories (LUTs) 5B, 5G, and 5R as numerical data for each color component.

Next, a main scan of the original image 1 is performed for image recording. The signals D, D, and DHGR obtained at this time undergo normalization conversion in LUTs 5B, 5G, and 5R, and the signals D
, D , D , and the next stage color calculation circuit NU
Given to NG NR 6. In the color calculation circuit 6, the signal DIliB'
Perform color calculations based on D, D, Y, M, C.

Each color plate signal for NG NR K is generated. These color signals are converted into Y, M, and C dot signals in a halftone conversion circuit 7, and are combined in a time-sharing manner by a selector 8 and applied to an output circuit 9.

The recording head 10 has a built-in laser light source, and modulates the laser beam by 0N10FF using a modulation signal given from the output circuit 9. The photosensitive film 12 is wound around the recording drum 11, and due to the combination of the α rotation of the drum 11 and the translation of the recording head 10 in the β direction, the exposure of the photosensitive film 12 by the laser beam is performed sequentially and sequentially in scanning lines. This is done pixel by pixel. In addition, below, Y, M, and C are printed on the photosensitive film 12.
.

Let us consider the case where each color image of K is recorded as a positive image.

B. Normalization curve generation operation in the first embodiment FIGS. 3A and 3B are flowcharts showing the operation of the color scanner 100 described above, and this operation is suitable for correcting color cast only in shadow areas. ing. Moreover, FIG. 4 is a diagram conceptually showing the process of normalization curve generation, and in this FIG. 4, Sl, 3101. . . . correspond to the step numbers in FIGS. 3A and 3B. Furthermore, the white arrows indicate data usage relationships.

Below, we will explain the process of normalization curve generation step by step.

(n-1) Identification of Average Color Density Value First, in step S1 of FIG. 3A, the entire portion of the original image 1 to be copied is blisscanned.

Then, the color density signals D , D , DBG for each pixel
R (FIG. 2) is stored in frame memories 13B, 13G, and 13R. Below are these signal DBs.

D6. The color density value indicated by DR is expressed by the same symbol D,
Expressed as D and DR. Furthermore, when the original image 1 has a large size, the color density values D and D are large.

This storage operation of DG DR is performed while thinning out pixels.

The next step 5100 is to obtain these color densities 1ii! D.B.

Highlight side average color density value DI from D c and D R
This is a subroutine SAB SAG 5AR-chin for determining IAB'D, D and the shadow side average color density value IIACII^R D, D, D. Although this subroutine is detailed in Japanese Patent Application No. 63-312684 previously filed by the applicant of the present invention, only an outline thereof will be described here.

In FIG. 3B showing the details of this step 5100, in the first step 5101, the average density value for each pixel is calculated from the density values D, Dc, and DR obtained in step S1: D - (D o + D c + D R) / 3
seek. Furthermore, this calculation is performed for all the pixels that have been Briscanned, and the average density value frequency histogram is created as shown in Figure 5, with the class indicating the range of the average density value DM on the horizontal axis and the number of pixels N, on the vertical axis. create. In FIG. 5, the median value of the class is indicated by DMI ([-1-n).

In step 5102, each pixel included in each class of the average density value frequency histogram is
, G, R contrast viD, D.

Cumulatively add DG DR. This process is performed independently for each class. After further performing this calculation, each class ft1
A cumulative density value histogram for each color component is created with D M, (+-t to n) as the horizontal axis and cumulative density values ΣD, ΣDo, ΣDR corresponding to pixels included in each class as the vertical axis. FIG. 6 shows an example of this cumulative density value histogram, showing X-B, G.

For each of R, such a cumulative density value histogram is obtained.

As an example, class value DH, -1, 0, class width 0.1 (7
) class (e, 95≦D<1.05) will be explained.

For convenience, the number of pixels included in this class is three, and the density data of each pixel is pixel 1: DB-1, 10, DG-0, 90DR-0.
, 95 (D, 4 0.98) pixel 2: DB-1
,00,Do-1,10DR-0,90(DM*1
．． 00) Pixel 3: DB-1,00, DG-0,9
5DR-0.95 (DM4 0.97).

When the cumulative density value histogram shown in FIG. 6 is X-B, the cumulative density value ΣDB in that class (e, 95≦DM<1.05) is ΣDB-1,10+ 1.00 +1.00-3 .10
becomes. Similarly, the other cumulative concentration values ΣD and ΣDR are respectively ΣDG −0,90+1.10+0.95
- 2.95XDR-0,95+ 0.90 +
It becomes 0.95-2.80.

Such processing is performed for each class, and each color component B, G
, R to complete each cumulative density value histogram. (See also the block of step 5102 in FIG. 4.
In the figure, each histogram is approximately drawn by a curved line. Further, step 5IOI, 5102 is actually processed in parallel. ) In step 8103, from the average density value frequency histogram shown in FIG. 5, the relative frequency RN(
%) as the vertical axis, a cumulative relative frequency histogram as shown in FIG. 7 is created. The histogram has a shape that varies from 0% to 100% within the minimum and maximum occurring density values DM□In'DMlax. Also, however,
In FIG. 7, this cumulative relative frequency histogram is approximated by a curve on the premise that the class width is sufficiently small.

In the next step 5104, predetermined cumulative density appearance rates RN and RN8 are applied to the cumulative relative frequency histogram shown in FIG. Find each. In addition, the cumulative m concentration appearance rate RN
The values of ++ and RNs are values obtained in advance as providing statistically optimal highlight points and shadow points by analyzing a large number of sample original images, and are, for example, values of about 1% and 98%. .

In step 5105, in the cumulative density value histogram for each color represented in FIG. 6, on the highlight side, as shown with diagonal lines in FIG. (DMsin≦DM
≦DH11'' On the shadow side, each area (DNS≦S D ≦D) with a temporary shadow average density value D or higher is reached [and M within that range is
The cumulative density values ΣD, ΣD, ΣDR of Ma+ax are calculated for each color component r3.
The pixels are averaged over G.

In other words, calculate the pixel average on the highlight side...〉11
, and the pixel average on the shadow side is written as 8. Then, the highlight side average color density values D , D , D obtained by this pixel average and the shadow side 11AB
II^G II^P average color density value D
, D , D are SAB SAG
5AR D -<ID> (X-B, G, R)
-(1) 11AX X 11 D -<ID> (X-B, G, R)
-(2) SAX X S.

For example, X in FIG. 8 is B, and the temporary highlight density D
When H1l is DNS, 0-(DMI+DM2
+DM3+DM4+DM5) 118B / (N p+ + N P2 + N P3 + N
114 + N 1)5). However, N P,
(1-1 to 5) are the numbers of pixels whose average density value D belongs to class DMl, and are determined from the histogram in FIG.

In this way, the highlight side and shadow side average color density values D, D for each color component shown in the center of FIG.
(X-1, R) is obtained. That's all for II A
X S A X The subroutine corresponding to step 5100 is completed and the process returns to the main routine of FIG. 3A.

(n-2) Calculation of reference value and correction amount In the next step S2 in the main routine, B, G
, the highlight reference value D110 and shadow reference value D8o common to each color component of R are determined by the following (4) and (5).
Obtain according to the formula.

D-FH (D, D, D)...(4
)II OII^11 11AG IIARD-
FS(D,D,D)...(5)S
OSAD SAG SAR However, the functions FH and FS are FH (D , D , D ) 11AB
II^G IIAR-r,,”MIN(D,
D , D ) 11AI3 II^G
II^ + (1-r,, ) D,, F-C6a)F
S(D,D,D)SAD SA
G5AR-r-MAX(D,D,D)S
SAB SAG SAR+ (
1-rs) Ds, defined as -(6b), MIN (...): Minimum value selection operation, MAX (...
): Maximum value selection calculation, D, D: Predetermined standard 11P SP highlight density value and shadow density value, r n , r s : 0 < r 11 <
A preselected constant in the range 1.0 < r s < 1. That is, the highlight reference value D110 is D
(X-B, G, R) (7) Minimum value and standard value Dn r
: Tonneau weighting 11^X is the average, and the shadow reference value D is D (X-B,
GSo SAX, R) maximum value and standard value D81. The weight i1; is average.

Nao, functions FH, FS1. t, D (X-+3.G
,I? ) and 11^X D (X-B, G, R), respectively, may be a function that calculates the average value SAX.

In the next step S3, the gray balance correction amount ΔD Ilx (x-n, c, l?) in the highlight part is
and the gray balance correction amount ΔDsx in the shadow area
(x-B, G, R) is determined using the following equations (7a) and (7b).

ΔD −K φGH(D −D )11X
II II^x1] 0 (X-B, G, I
? ) −(7a) Δ D −K −GS(D −D
)SX S SAX
5O(X-B, G, R) - (7b) However, K and K are determined experimentally in advance and I
I S is a constant of i [, and the functions GH and GS are, for example, GH (D-D) II A ,,l ]1
1max l1m1n ... (8) GS (D -D8o) SAX -D-D ... (9) S
Defined by AX SO. However, D - MAX (D , D , D )I
I to a x If^13 II
＾GllAl? ...(I 0a) D - MIN (D , D , D ) I
lm in 1IAI3 11AG
IIΔR...(10b) A: Preselected positive constant, 1 m: Preselected positive constant, (for example “
3°).

As can be seen from equations (6a) to (9), the correction amount ΔD of the highlight part is (DD), and
II A
is (D-D), SX SAX S.

proportional to each other. (D-D) is Highrai II A
II O Reflects the amount of color cast in the to part, (Ds^-D
) reflects the amount of color cast in the shadow area.

By the way, if we pay attention to the gray balance correction amount ΔDIIX in the highlight part, (7a), (8), (
10a).

B and G in the highlighted part are determined by equation (10b).

Variation width ΔD of average density value of each color component of R ■
D-D...(11>IIs
The larger llmax 11m1n is, the smaller the gray balance correction amount ΔDIIX becomes. Therefore, when there are no highlight points in the original image 1, it is possible to prevent the amount of gray balance correction in the highlight portion from being greater than necessary. The constant A11 is a threshold value corresponding to the determination of the presence or absence of a highlight point, ΔD>A, ... (12) 1
1 steel, the above correction suppression effect becomes large.

(+3-3) Highlight density value, shadow density value, and in the next step S4, the reference values D and D are supplemented by +10
SO 1 [- amount ΔD, ΔD 8x (X-13, G, R), color composition + 1X highlight density per minute "IIX" and shadow density value D 8X (X-1, G, R) , next (7) (+3),
Calculate by 1 using equation (+4).

D - D + ΔD (X-n, C, R)
−(13) IIX 110 11X D −D + Δ D (X
-n, G, I? ) −(+4)sx so
sx In the next step S5, the operator determines whether color cast correction for only the shadow portion of the original image 1 is necessary. Generally, it is difficult to visually judge the color cast only in the shadow area, so it is determined whether or not there is color fading in the original image 1, and if it is not faded, it is assumed that the color cast is only in the shadow area, and in the next step S6, a dummy shadow is created. The density value DI)SX(X-n, G, I'?) is set to a common value using the shadow reference value D8o. That is, D -D -D -D...(1
5) 1) Sr3 DSG DSRSO.

On the other hand, if the original image 1 is completely discolored,
Dummy shadow density value D (X-11, G, I?)
DSX is the shadow density value D of the corresponding color component D8X(X-B
, G, R) respectively. That is, D
-D (X-B, G, R) -(1B
) DSX SX (step S7). In addition, in step S7, the dummy shadow density value D may not be completely the same as the shadow SX density value D, but may be set to a value x near D3X.

(n-4) Generation of normalized curve In the next step S8, the highlight density value D obtained as above, the shadow density value Dsx+1X' and the dummy shadow density fID (X-11,G
, R) as 1) SX to specify the iL normalization curve for each color component. This one
-1, a nonlinear transformation function f defined by the following equations (!7) to (I9) is determined in advance, and information representing this function is stored in the memory 16.

DN-f (D: u, v, w) ■p+ (p-q) Rag [g + g2] 10
1...(17) where p, q: constants set near the lower and upper limits of the normalized concentration range before masking, D = concentration value before conversion, DN: concentration value after conversion, u, v, w: Parameter, constants p and q are determined in advance for each color component. Then, as the values of parameters U, V, and W, u = D, v −D, W−D 5
Substituting x11X DSX (X-n, G, R) - (20) into equations (17) to (19) for each color component, the function f
Specify the conversion characteristics for each color component. “Ninth
The figure is a diagram illustrating the shape of the function f specified in this way on a density normalized transformation coordinate plane. However, p
, QX are the values of constants p and q for color component X.

Also, the concentration variable Dx.

A subscript "X" is attached to DNx to indicate that it corresponds to color component X. The characteristics of this function f will be described later, but examples of the function f that have been realized in this way are shown as Ca and C in FIG. 9 (the difference between C3 and Cb will also be described later).

When the function f is specified for each color component, the CPU 15 generates numerical data expressing the conversion characteristics of the function f, that is, data numerically expressing the curve C8 in FIG. 9.

Then, input that data into LUT5B in Figure 2 for each color component.
, 5G, 5R. With the above steps, the preparation process for normalization conversion is completed.

In step S9 of FIG. 3A, a main scan of the original image 1 is performed for the purpose of actual image reproduction. The operation during this main scan has already been described and will not be repeated here, but the normalized curves in the LUs 75B, 5G, and 5R generated as described above are used as signals.

Used for DG and DR conversion.

(11-5) Characteristics of normalized curve Referring again to equations (17) to (19) and FIG. 9 described above. Then, by equations (17) to (19), the function f
It can be seen that the normalized curve generated thereby has the following properties (i) to (iv).

(1) Highlight density value DHX is shadow density value D
SX is much smaller than the dummy shadow density value D
Since DSX becomes a small value, (2
0) In formula, u << v w
... (21
) holds true. Furthermore, since the difference between the shadow density value D8x and the dummy shadow density value D is relatively small, DSX V and W have approximately the same value. Therefore, the quantity appearing on the right side of equation (19): (w-u) / (v-u) ... (22
) is almost “1”, and the defining formula of the quantity g2 ((19)
The second term on the right side of the equation) is approximately "0.lo".

Therefore, by subtracting the first term and the second term on the right side, the quantity g
The absolute value of 2 is quite close to 0.

On the other hand, when the density value D (-DX) is the same as the value of the parameter U (that is, the highlight density value DIIX), (
18) The exponent on the right side of the equation becomes 0", and gl becomes "1".

For these reasons, the sum (g
The main term of + g 2 ) is gl, and ■ g ”g2- 1 ...(23)
becomes. As a result, D “p(-px)...(24)
is obtained.

That is, by the function f, D" DIIX is effectively D
The normalized curves C, C, in FIG. Pass.

(In the equations (17) to (19), D-D
(Shadow density value) (25) sx. Since the value of the parameter W is given by the shadow density value D8X according to the formula (20), the quantity g1 of the formula (18)
and the second term on the right side of equation (19) are completely the same. As a result, the value in [...] of the logarithmic function in equation (17) is “
0.1'', and D-q(-qx)...(2B).In other words, D ``'' D sx is converted to DN = qX by the function f, and the normal The curves C, C, are two-dimensional coordinates (D8X.

qX). However, the first
Of the two points Q-QQSSa'S ``''sb shown in the figure, the former is SX when the gray balance correction amount ΔD in the shadow part is a positive value ΔD
This is the second point Q8 in the case of SXa. The latter is the second point Qssx when the correction amount ΔD is a negative value ΔD
It is sxb. Density value D- before conversion
DD represents the shadow density value in these two cases. The normalized curves in these cases are shown as curves C and C, respectively.

(Ill) When the density value is not the same as the value of the parameter U (-highlight density value D11x) but is relatively close to it, that is, in the highlight part, the absolute value of the exponent on the right side of equation (18) is " Close to 0". Therefore, g
The value of l is close to "1" and is much larger than the value of g2, Iog [g 10 g2] 10 100 million 1og10 [gt] = -(D-u)/(v-u) -( 2
7). Therefore, the function f in the highlighted part is DN ” p-(p-q)(D-u)/(v-u)
...(28) is approximated as follows.

By the way, the formula: % formula %() (29) is the first point Q determined according to the highlight density, x(-u), and the third point Q determined according to the dummy shadow density DDSX"1 V). Point: Q = (D, QX) ... (
30) This is the equation of the straight line L (see Figure 9) passing through D DSX .

Therefore, in the highlighted part HLP of FIG. 9,
This means that the normalized curves C and C change along a straight line.

On the other hand, as explained in (■1), the normalized curve C
, C pass through the second point Q8, 1b. In the illustrated example, the position of the second point Q8 is different from the third point QD. Therefore, in the shadow portion SDP, the normalized curves C and Cb deviate from straight lines, and the amount of deviation becomes larger as the density value increases. Further, although the amount of deviation in the medium density area MDP is larger than that in the highlight portion HLP, the amount of deviation in the shadow portion SDP is not very large.

Since the amount of deviation in the shadow portion SDP is set to a value corresponding to the amount of color blurring only in the shadow portion of the original image 1, the curve C in FIG.

The normalization curve of this example illustrated by C5 is a curve that can remove color cast only in shadow areas without disturbing the gray balance in the low-medium density range.

(1v) As mentioned above, when the color of only the shadow part is not a fake, for example, the entire original image 1 has faded, the shadow density value Dsx and the dummy shadow density value D
are assumed to have the same value.

DSX The normalized curve at this time is v−
By substituting w, it can be expressed by the same equation as equation (29). That is, the normalized curve at this time is a straight line passing through the first point QI+ and the second point Q8 (-Q or Q3.). When the second point Q8 is QSa, this straight line is shown in Figure 9. It is shown closed. This straight line. The normalization conversion characteristic shown by is the highlight part HLP,
Both the medium density area MDP and the shadow area SDP function to suppress the density of specific color components and ensure gray balance between each color component. In other words, the fading of the entire original image 1 is corrected.

Therefore, the nonlinear function f used in this example is an effective function for both the color cast correction of only the shadow part and the fading of the entire original image 1, and seven systems are used for these various corrections. -Can be used in a variety of ways.

Even if there is no color cast or fading in the original image 1, the shadow density value D, the dummy shadow density value D, and sx
By setting osx to be the same, a straight line equation can be obtained from the function f, so the function f is also effective in such a case.

C1 Second Embodiment The second embodiment of the present invention can be implemented independently of color cast correction 1E, and the basic configuration of a color scanner for plate making to which this second embodiment is applied is shown in FIGS. This embodiment is the same as the scanner 100 shown in FIG. 2, and differs from the first embodiment only in the normalization curve generated by the information processing device 14.

In the second embodiment, after obtaining the highlight density value DIIX and the shadow density value Dsx (X-r3.G, R) for the original image 1 to be copied in steps 81 to S4 in FIG. 3A, steps 85 to S7 are performed. Instead, step S50 in FIG. 10 is executed.

In order to execute this step S50, a highlight of the standard original image is selected in advance.
x and the shadow density value D are specified in advance. Just F
X However, a standard original image refers to an original image that has color tones and gradations that are empirically standard, and when using a photograph as an original image, an original image that has appropriate exposure conditions. is pointing to. Then, the highlight density value and shadow density value for each color component that have been empirically found for such a standard original image are expressed as D and D, respectively.
That is to say.

II I)X S P It is determined by the following equation (31) l.

D − to −D, x+ to −ΔDFxSX + (] −1() ・D
(X-B, G, lυSl'X...(31) However, ΔDI''X is calculated using the following formula (32): % formula % (32
), and corresponds to the density range width for each color component in a standard original image (hereinafter referred to as "standard density range width"). Also, K is 0≦≦1
... is a constant that satisfies (33) and is determined in advance.

Equation (3I) is a value proportional to the standard concentration range width ΔDFX: -ΔD,,x+ (1-I() ・Ds[, x...
・(34) and the highlight density value DI of the original image 1 to be copied
This means that the dummy shadow density value D is determined by adding the values proportional to IX: K - D , , x - (35). Also, from another point of view, the standard concentration range width ΔD,...! : The standard shadow density value D is weighted averaged using the weighting coefficient K, and the value of equation (34) obtained by FX is added to the value of equation (35) to obtain the dummy shadow density value D. become. In this way, the dummy shadow density value D
The advantages of determining SX will be discussed later.

When step S50 in FIG. 10 is completed, a normalized curve is generated in step S8 in FIG. 3A, and in step S9, the original image 1 to be copied is scanned while normalized density conversion is performed using the normalized curve. The principle of halftone image recording itself is the same as in the first embodiment.

FIG. 11 shows two normalized curves C and C2 determined by this second embodiment. Of these, the first
The normalized curve CI has a small highlight density value D and shadow density value D +1XI
The second normalized curve c2 shows the case where the highlight density value D and the shadow density value D are II X 2 3 X 2
Indicates a large case. That is, these two normalized curves C and C2 are curves for two types of original images to be copied having different density ranges.

As in the first embodiment, by using the nonlinear conversion function f defined by equations (17) to (I9), the first normalized curve C1 is
P has a shape along the straight line L1 connecting the highlight point Q111 and the dummy shadow point QDI. Then, the curve C1 passes through the shadow point Qs1. Similarly, the second normalized curve c2 changes along the straight line L2 connecting the highlight point Q112 and the dummy shadow point QD2 in the highlight area HLP and the medium density area MDP, and changes from the straight line L2 in the shadow area SDP. That's shadow point QS2
pass through.

Here, two highlight density values D, DllXI
If the difference between llX 2 is ΔHX, the following equation (36) holds true.

ΔH-D-D...(86)X
llX2 11XI On the other hand, (31
) is applied to the first and second normalized curves C1°C2, respectively, to obtain the following equation (37).

(38) holds true.

D - to - D + - ΔDFx1) SXI
IIXI+(1-K) ・DSP
X'・(37)D - to -D +K
・ΔDFx1)SX2 11X2+(
1-K) ・DSEX ... (38) Therefore, two dummy shadow density values DDSXI'D
If the difference between is written as ΔD, then from equations (37) and (38) 1) SX2 DSX, the following equation (39) is obtained.

ΔD Mi D -D 1)SX I)SX2 DSXI-K・ (
D-D)...(39)+1X2
11XI Therefore, by combining equations (36) and (39), a proportional relationship equation such as the following equation (40) can be obtained.

ΔD− to ΔHx−(45) DSX The following can be understood from equation (40). That is,
Older dummy shadow points Q and Q in Figure 11
D2 loaf distance ΔD is the two highlight points Q, QDSX
II+ is twice the mutual distance ΔHx of 112, and especially when it is -1, these mutual distances ΔD and ΔDIIX are equal to each other DSX. Therefore, when it is -1, it is a straight line L +.

L2 are mutually parallel. This consists of two shadow points D
, D is the value of the mutual distance ΔSx of no SXI
This is a property that holds true in the SX2 relationship.

Here, consider an arbitrary positive value ΔDr that is not very large. Then, the density value obtained by converting the density value ΔD larger than the first highlight density value DIIX (see FIG. 11) using the first 1E normalization curve C1 is converted to DN
Let it be l. At this time, since the straight lines and L2 are parallel to each other as described above, the density value D2, which is the same value ΔD1x2 larger than the second highlight density value D, is converted by the second 1E normalized curve C2. The density value obtained is also DN□.

Therefore, if the normalization conversion according to this embodiment is performed on each of a plurality of original images to be reproduced having different density ranges, a proper reproduced image can always be obtained regardless of the difference in density range. Shadow Department SDP
Now, the regular north side tic.

C defines each unique shadow point QQ as 2
Since the condition that S1' passes through S2 is also satisfied, normalization transformation with favorable properties is performed over the entire range of density values.

By the way, when the constant K is smaller than "1", the straight lines L and L in FIG. 11 are not completely parallel, but approximately parallel. Therefore, at this time, the 11th
The density values D 1 and D 2 in the figure are converted into values that are relatively close, although not completely identical, through normalization conversion.

FIG. 12 is a graph showing by a solid line SX how the dummy shadow density value D changes when the value of the constant is changed, and is drawn using equation (31). When K-0, the dummy shadow density value DD8X matches the standard shadow density value D. Since this FX standard shadow density value D is a constant value that is unrelated to the image FX of the original image 1 to be reproduced, a common dummy shadow density value D 2 is set for all the original images 1 to be reproduced. As the constant K increases, the dummy shadow density value D also increases linearly, and in 5x K-1, as mentioned above, the highlight density value D of the original image to be copied 1 and the standard density range width ΔDFX! It becomes the sum of 1X. Therefore, when the coefficient K is set to "1" or a value close to it (for example, 0.5 or more), the dummy shadow density value D changes relatively largely according to the high SX light density value D11X of the original image 1, and the coefficient K For example, when the value is set to 0.5 or less, the dependence of the dummy shadow density value D on the highlight density value DIIX of the original image 1 is suppressed, and the density value D of the dummy shadow becomes close to the standard shadow 1) SX density value D . Therefore, by arbitrarily specifying the value of the coefficient FXK, the normalized curve Cl,
It is also possible to change the properties of C2.

By doing so, it is possible to automatically generate a normalization curve that is considered desirable based on empirical rules using a small number of parameters.

D. Modification [1] FIG. 13 shows a color scanner for plate making 4[]O in which a scanner main body and a color separation auxiliary device are integrated. This scanner 400 is also roughly divided into a human power section 401 and an output section 402, and an operation panel 403 is provided above the human power section 401. Further, each circuit in the color separation auxiliary device 300 shown in FIG. 1 is housed in the main body of the human power section 401 in FIG. 1'.

The CRT 418 and the operation panel 417 are each
This corresponds to the CRT 18 and console 17 in the figure. The functions and operations of this scanner 400 are substantially the same as those of the scanner 00 of FIG. 1, but the scanner 400 is relatively combinant because it does not require a separate color separation auxiliary device.

[2] In generating the normalized curve, nonlinear functions other than the function f described above can also be used.

In general, let the function expressing the straight line be f, (D), and f
(D,,X)-0...(41)t (D
)+f (D)=q...
(42) l SX 2
SX Monotonically increasing or tll, key decreasing nonlinear function f2 (D) that substantially satisfies X
Using r (D) + t2 (D)
...(43) ■ If a function is created, it will be a function that complies with the requirements of this invention.

Note that the function f in equation (17) is equal to f in equation (43),
(D) = p (p q) (Du) / (v-
This corresponds to the case where u)...(44) +2 (D) -(p-q)i)og [1+g2/gI]0...(45). However, g1g2 is a function defined by equations (18) and (+9).

[3] In the above embodiment, normalization conversion is performed using LUT5B, 5G
, 5R, but it may also be performed analogously using a function generator.

[4] It is preferable to determine the highlight density value DIIX and the shadow density value D8x through a statistical method using a histogram or the like as in the above embodiment.
If highlight points and shadow points are found in
The operator specifies them, measures their color density, and
The value determined by A-1 may be adopted as DIIX'D8X.

[5] When generating the normalization curve according to the present invention [2], correction may be performed depending on the subject matter of the original image to be processed. Such amendments are described, for example, in Japanese Patent Application No. 1-308485 filed by the applicant of the present invention. For example, the highlight density correction amount Δdx (X-13, G, l?) is determined in advance according to the subject matter of the original image to be processed, and the highlight density value, DI
IX may be shifted by this correction amount Δdx to obtain a new highlight density value D. In this way, the normalized curve C1o in the case of 1cX without correction becomes the curve C1□ after correction.

[6] Although the first embodiment is applied only to color original images, the second embodiment can be applied to both color original images and monochrome original images. In the case of a monochrome original image, one highlight density value, one shadow density value, and one dummy shadow density value are determined, and one normalization curve is also determined.

[8] This invention is not limited to color scanners for plate making;
The present invention is applicable to all image processing devices that perform color separation and other processing on original images.

〔Effect of the invention〕

As explained above, according to the first configuration of the present invention,
In addition to the highlight density value and the shadow density value, a dummy shadow density value is determined, and the shape of the normalization curve in the highlight and middle density areas is determined by the highlight density value and the dummy shadow density value. The shadow density value is reflected only in the shape. Therefore, the shape of the normalization curve in each concentration range can be uniformly managed using a relatively small number of parameters, and a skilled operator is not required. Therefore, it is a method suitable for automation (corresponding to the first objective).

In the second configuration of the present invention, by providing a highlight density value and a shadow density value for each color component, and providing a dummy shadow density value corresponding to the amount of color cast only in the shadow part, the color of only the shadow part is A normalization curve suitable for fog removal can be systematically generated. Since the normalization curve is generated based on values that can be objectively grasped in this way, it is possible to quantitatively manage the amount of color cast correction (corresponding to the second objective).

Also, color cast correction is done during normalization conversion, so
Processing characteristics in other circuits within the color separation device are set for the signal after color cast correction. Therefore, regardless of whether or not the original image has a color cast, these circuits can determine the processing characteristics assuming that the original image has no color cast (corresponding to the third objective).

Furthermore, according to the third configuration of the present invention, by simply changing the parameters of a predetermined function, removal of color cast only in shadow portions and correction of overall fading of the original image can be handled in a unified manner. Therefore, this invention is a method particularly suitable for automating color separation conditions (corresponding to the fourth objective).
.

Further, according to the fourth and fifth configurations of the present invention, since the dummy shadow value is determined while reflecting the value of the density range width of the standard original image, even if the density range of the original image to be processed changes, the highlight part In the middle density range, density values that have the same difference from the highlight density value can be converted into density values that are substantially the same or relatively close to each other by normalization. The methods according to the fourth and fifth configurations are applicable to both color original images and monochrome original images, regardless of whether or not they are for color cast correction, and are applicable to the density range of the original image to be processed. Regardless, it is possible to always ensure proper density expression in the image after normalization conversion (corresponding to the fifth objective).

In particular, in the fifth configuration, the dependence of the dummy shadow density value on the width of the standard density range can be arbitrarily changed by changing the weighting coefficient in the weighted average, so its versatility is also high.

[Brief explanation of drawings]

FIG. 1 is an external view of a color scanner for plate making to which an embodiment of the present invention can be applied, FIG. 2 is a block diagram of a color scanner for plate making according to the embodiment, and FIGS. 3A and 3B are Flowchart showing the operation of the embodiment; FIG. 4 is a diagram conceptually showing the data processing procedure in the first embodiment; FIG. 5 is a diagram showing an average density value frequency histogram; FIG. 6 is a diagram showing color components. 7 is a diagram showing a cumulative relative frequency histogram, and FIG. 8 is an explanatory diagram of the process of calculating the average color density value from a part of the histogram in FIG. 5. 10 is a flowchart showing a part of the process of the second embodiment of the present invention. FIG. 11 is a graph showing an example of the normalized curve generated by the first embodiment. FIG. 12 is a graph showing the dependence of the dummy shadow density value on the weighting coefficient of the weighted average, and FIG. 13 is a graph showing an example of the normalized curve generated by the embodiment of FIG. FIG. 14 is an external view showing another example of a color scanner for plate making, and is a diagram illustrating correction of highlight density values according to the subject matter of the original image to be processed. Dx...Density value for each color component, DNx...Normalized density value for each color component, DIIX...
...Highlight density value for each color component, D8X...Shadow density value for each color component, D...Dummy shadow density value for each color component, SX Q...First point, Q8... Second point, 1 Q,...Third point, c,,c,c,c2...Normalized curve ID...Standard highlight density value, 1FXD...Standard shadow density value , FX ΔDPX...Standard concentration range width, K...Coefficient in weighted average

Claims (5)

[Claims] (1) A method for generating a normalization curve for converting the density value of each pixel of an image of an original image to be processed into a value within a normalized density range suitable for image processing in a predetermined image processing device, , (a) for the original image to be processed, determine a highlight density value and a shadow density value from the range that the density value before conversion can take; A dummy shadow density value is determined for the density value, and (c) as the normalized curve, (c-1) a first shadow density value determined according to the highlight density value on the coordinate plane in which the normalized curve is to be defined. and a second point determined according to the shadow density value; and (c-2) in the density region near the highlight density value, 1 and a third point on the coordinate plane determined according to the dummy shadow density value.
A normalized curve generation method characterized by generating a curve that satisfies both the conditions of and. (2) When color-separating a color original image pixel by pixel using a predetermined color separation device, the density value of each color component of each pixel is set within the normalized density range suitable for color separation in the color separation device. The method according to claim 1 is applied to the process of generating a normalization curve for each color component for conversion into values, and each of the highlight density value, shadow density value, dummy shadow density value, and normalization curve is a color component. If the color cast is determined for each component and the step (b) requires (b-1) color cast correction only for the shadow part of the color original image, a value corresponding to the amount of color cast is added from the shadow density value. The deviated values are obtained for each color component and used as the dummy shadow density value for each color component, and when color cast correction only for the shadow portion is not required, a value that is substantially the same as the shadow density value is determined. A normalization curve generation method, comprising the step of setting a value as the dummy shadow density value. (3) In the method according to claim 2, (d) for generating the normalized curve, (d-1) having a highlight density value, a shadow density value, and a dummy shadow density value as parameters; as a variable, (d-2) when the shadow density value and the dummy shadow density value are different values, it is nonlinear with respect to the variable, and (d-3) the shadow density value and the dummy shadow A function that is linear with respect to the variables is determined in advance when the density value and the density value are the same, and (e) the normalization curve is calculated as follows: (e-1) If the color original image requires color cast correction only in the shadow area (e-2) If the color cast correction of only the shadow part is required, the shadow density value and the dummy shadow density value are set to different values and the function is (e-3) When the color cast correction of only the shadow portion is not required, the shadow density value and the dummy shadow value are set to the same value, and the parameter of the function is set. The normalization curve generation method determined by specifying. (4) In the method according to claim 1, the step (b) comprises: (b-1) a value obtained by adding a value proportional to the density range width of the standard original image and a value proportional to the highlight density value; A normalization curve generation method including the step of using the obtained value as a dummy shadow density value. (5) In the method of claim 4, the dummy shadow density value is a weighted average value of the shadow density value and density range width of the standard original image, which is proportional to the highlight density value of the target original image determined in step (a). Normalization curve generation method determined as the value added to the value.