JPH096959A - Spatial frequency characteristic setting method and digital filter setting method - Google Patents

Abstract

(57) [Abstract] [Purpose] The spatial frequency characteristic of the target image is characterized by the peak frequency, its intensity, and the cutoff frequency, and the target spatial frequency characteristic and the spatial frequency characteristic of the scanner are defined by mathematical expressions. Design a digital filter based on. [Constitution] A peak frequency, its intensity, and a cutoff frequency are determined, and these points are smoothly connected by a predetermined function to determine a spatial frequency characteristic of a target image. Then, the one-dimensional spatial frequency characteristic of the digital filter is determined based on the target spatial frequency characteristic and the mathematical expression that determines the spatial frequency characteristic of the scanner, and the two-dimensional spatial frequency characteristic is expanded by the McClellan transform, and the spatial frequency sampling method is used. The coefficient of the digital filter is obtained, and the digital filter thus determined is used to perform digital filtering on the image input by the scanner.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for setting a target spatial frequency characteristic at a certain stage during image processing, and a method for setting a digital filter used for image processing.

[0002]

2. Description of the Related Art Reading a document image with a scanner to obtain a digital multi-valued image and performing various processes on the digital multi-valued image is not limited to the field of printing and plate making, and in recent years, a personal computer has been used. Image processing systems have come to be used.

There are various types of processing such as contour enhancement, and there are a method of performing the processing with a scanner and a method of using a digital filter.

The following is a description of a case where contour enhancement is performed, as an example. First, as a method of performing contour enhancement processing by the scanner, there is a method of using a contour enhancement circuit provided in the scanner. That is, since a color scanner for printing plate making is usually provided with a contour emphasizing circuit as shown in FIG. 6, this is used.

In FIG. 6, a part of the scanning light transmitted through an original (not shown in FIG. 6) made of a negative film or the like is reflected by the half mirror 1 to be reflected by the main aperture 2 and the optical system 3.
It reaches the photodetector 4 through and is converted into an electric signal. A part of the scanning light passes through the half mirror 1, passes through an unsharp masking (hereinafter referred to as USM) aperture 5, an optical system 6, reaches a light receiver 7, and is converted into an electric signal.

The main image signal output from the light receiver 4 is supplied to the subtraction circuit 8 and the addition circuit 10. The sub-image signal output from the light receiver 7 is supplied to the subtraction circuit 8. The subtraction circuit 8 subtracts the sub image signal from the main image signal and outputs a difference signal.

The difference signal from the subtraction circuit 8 is multiplied by a predetermined coefficient α (hereinafter, this coefficient α is referred to as a sharpness parameter) in a coefficient circuit 9. This sharpness parameter α is preset by the operator, and the larger α is, the more the contour is emphasized.

The output of the coefficient circuit 9 is the addition circuit 10
At, the image signal is added to the main image signal to obtain an image signal whose contour is emphasized. This is the outline enhancement process performed by the scanner.

Next, there is a method using a digital filter. This is a method for performing edge enhancement by digitally filtering image data using a Laplacian filter as shown in FIG. 7, for example. Although FIG. 7 shows an example of a 3 × 3 digital filter, other 5 × 5 or 7 × 7 filters are also known.

[0010]

However, the quality of the image finally obtained in conventional image processing is not limited to edge enhancement, and smoothing processing or other processing is not the same. The quality of the image cannot be controlled using physically clear factors or criteria, because it is mostly determined by factors that are very difficult to quantify, such as the skill level or sense of the operator who performs the image processing or the operator who performs image processing. There was a problem.

For example, in the case where the contour enhancement circuit of the scanner shown in FIG. 6 is used to perform the contour enhancement, the degree of the contour enhancement can be adjusted by the sharpness parameter α, but the main aperture 2 is used. Since the spatial frequency characteristics of the image signal obtained by the light receiver 4 and the spatial frequency characteristics of the image signal obtained by the light receiver 7 via the USM aperture 5 are not known at all, When determining the sharpness parameter α, the operator relies on intuition and experience based on a very vague expression such as “contour is tight”, “slightly delicate and slightly sharp” requested by the client. It is actually there.

Of course, a skilled operator can obtain an image of a quality required by intuition and experience from these ambiguous expressions. However, the spatial frequency characteristic of the image input in this way can be determined. It goes without saying that it is still unclear what kind of thing it is.

That is, in the past, at the stage of being input by the scanner, the image quality is only expressed by a very ambiguous word, and is managed based on a clear element or standard that anyone can understand. It does not exist.

This causes the following problems when the image data input by the scanner is filtered using a digital filter.

First, when processing is performed using a digital filter, since the spatial frequency characteristics of the digital filter are not always clearly understood, what results is the spatial frequency characteristics of the image as a result of filtering. There is a problem that I don't know if.

That is, even if the operator who inputs the scanner performs the digital filtering process,
There is a problem in that it is not possible to clearly understand what quality of image is obtained by the digital filtering, that is, what is the spatial frequency characteristic of the obtained image.

Since the above problems occur even when one operator performs the processing from image input to digital filtering as described above, this is even more the case when the operator performing image input and digital filtering are different. .

That is, in such a case, even if the operator who performs digital filtering does not know what kind of processing and how much processing was performed at the time of scanner input, even if he knew, the contour was made tight, or slightly delicate, and somewhat delicate. It is only a very vague expression such as sharpening, and since the spatial frequency characteristic of the digital filter is not always clear as described above, what kind of image will be finally obtained as a result of digital filtering? It is not possible to grasp clearly.

Specifically, a desired image cannot be obtained even if the contour is tried to be emphasized, the contour is emphasized more than necessary, or unnecessary roughness is generated in the image. There was an occasion.

As described above, in the related art, when an image is processed by using a scanner or a digital filter, not only an image having a desired quality cannot be obtained but also the image quality may be remarkably impaired. In that case, the image had to be input again from the scanner.

Therefore, it is an object of the present invention to provide a spatial frequency characteristic setting method for making it possible to clearly grasp the quality of an image.

Another object of the present invention is to provide a method of setting a digital filter for obtaining an image having a desired spatial frequency characteristic.

[0023]

In order to achieve the above object, the spatial frequency characteristic setting method according to claim 1 defines a peak frequency, its intensity, and a cutoff frequency,
The feature is that the target spatial frequency characteristic is set by smoothly connecting these by a predetermined function.

A method of setting a digital filter according to a second aspect is a method of setting a digital filter for processing image data read by an input device for reading an image into image data having a target two-dimensional spatial frequency characteristic. Then, the one-dimensional spatial frequency characteristic of the digital filter is obtained based on the mathematical expression that determines the one-dimensional spatial frequency characteristic of the input device that reads the image and the mathematical expression that determines the target one-dimensional spatial frequency characteristic. It is characterized in that a two-dimensional spatial frequency characteristic is obtained from the original spatial frequency characteristic, and each coefficient of the digital filter having a desired size is determined based on the two-dimensional spatial frequency characteristic.

[0025]

According to the first aspect of the present invention, first, the peak frequency, its intensity, and the cutoff frequency are determined, and then these are smoothly connected by a predetermined function. Thereby, the target spatial frequency characteristic can be set.

Therefore, for example, the spatial frequency characteristic of the final target image can be set by this method, so that the quality of the image can be controlled by using physically clear elements and criteria. Therefore, anyone can clearly understand the image quality. As a result, it is possible to prevent the image quality from being significantly impaired as in the conventional case.

Claim 2 is as follows. According to this digital filter setting method, first, a mathematical formula that determines the one-dimensional spatial frequency characteristic of an input device that reads an image and a mathematical formula that determines the target one-dimensional spatial frequency characteristic are determined.

Next, the one-dimensional spatial frequency characteristic of the digital filter is obtained from these mathematical expressions, and the two-dimensional spatial frequency characteristic is obtained from the one-dimensional spatial frequency characteristic. Then, each coefficient of the digital filter having a desired size is determined based on the two-dimensional spatial frequency characteristic.

As described above, the digital filter for obtaining the image data having the target spatial frequency characteristic can be set, and the image data inputted from the input device is digitally filtered using this digital filter. Thus, an image having a desired target spatial frequency characteristic can be obtained.

As described above, it becomes possible to manage the image by the clear element and the standard of the spatial frequency characteristic, and it is possible to avoid the image quality being significantly impaired as in the conventional case.

[0031]

Embodiments will be described below with reference to the drawings.
FIG. 1 is a diagram for explaining steps when the spatial frequency characteristic setting method and the digital filter setting method according to the present invention are applied to image processing for printing plate making.

First, an image is input using a scanner (step S1), but the input spatial frequency characteristic of the scanner is clearly grasped in order to finally obtain an image of the target spatial frequency characteristic. There is a need. If the input spatial frequency characteristic of the scanner is not clear, it is not known what the spatial frequency characteristic of the final image will be, even if the spatial frequency of the digital filter that performs filtering afterwards is clearly understood. This is because the image quality will be significantly impaired as in the conventional case.

Therefore, the spatial frequency characteristic of the scanner is approximated by a mathematical formula as in step S11.

The aperture of the scanner has a circular aperture and a rectangular aperture. The present inventors have shown that the spatial frequency characteristic of the circular aperture is expressed by the following equation (1), and the spatial frequency characteristic of the rectangular aperture is the following ( 2) It was found that it is expressed by the equation (Note that regarding this point, the Proceedings of the 92nd Spring Meeting of the Printing Society of Japan, pp.13-16 Kudo, Takita, and Okubo "Spatial frequency of plate-making color scanners" Characteristics ").

[0035]

[Equation 1]

In the equation (1), R is the radius (mm) of the aperture, and J ₁ is the Bessel function of the first kind.
In the equation (2), L represents the length of one side of the square aperture, and f represents the spatial frequency. Furthermore, L
(F) is a function showing the low-pass effect of the optical system,
D is a circular aperture radius conversion value (mm) of the low pass effect by the optical system.

Therefore, assuming that a scanner with a circular aperture is used, the spatial frequency characteristic F _O (f) of the image data output from the contour emphasizing circuit shown in FIG. 6 is given by equation (3).

[0038]

[Equation 2]

Here, α is the sharpness parameter, F _MO (f) and F _UO (f) represent the spatial frequency characteristic of the main aperture and the spatial frequency characteristic of the USM aperture, respectively, and R _M and R _U are Main aperture radius (mm) and USM aperture radius (mm) respectively
Is shown. Further, L (f) is a function showing the low-pass effect of the optical system, and D is a circular aperture radius conversion value (mm) of the low-pass effect of the optical system.

That is, the above equation (3) can be uniquely determined by analyzing the structure of the scanner and the characteristics of the optical system used, whereby the spatial frequency characteristics of the scanner can be approximated by mathematical expressions. It will be.

Although the scanner using the circular aperture has been described above, it is obvious that the scanner using the rectangular aperture can be similarly approximated by mathematical expressions.
In the following, a circular aperture is used.

From the above, by substituting the values used in the actual scanner for each parameter of the equations (3) to (8), a mathematical expression approximating the spatial frequency characteristic of the image input by the scanner is obtained. be able to.

Actually, it is desirable to set the sharpness parameter α = 0 so that the contour enhancement circuit of the scanner does not operate when an image is input. This is because if the contour is emphasized at the image input stage, the image is likely to be deteriorated when a process such as digital filtering or scaling is performed later.

When the spatial frequency characteristic of the image data input by the scanner is determined as described above, the target spatial frequency characteristic of the image is then determined (step S12).
This spatial frequency characteristic defines what quality of image is obtained by the digital filtering process in step S2. The spatial frequency characteristic selected in this step is a one-dimensional spatial frequency characteristic in which the horizontal axis represents the spatial frequency and the vertical axis represents the amplitude intensity.

The operator can arbitrarily set the spatial frequency characteristic of the target image, but it is desirable to express the spatial frequency characteristic by a mathematical expression so that anyone can clearly understand the target quality. become.

By the way, when the target one-dimensional spatial frequency characteristic is determined as shown in FIG. 2, such a curve can be generally expressed by combining curves of appropriate functions. In FIG. 2, the horizontal axis represents the spatial frequency and the vertical axis represents the amplitude intensity.

However, it is not always clear to the operator what image quality the spatial frequency characteristic realizes by simply expressing the target spatial frequency characteristic by a mathematical expression.

Therefore, as a characteristic of the target spatial frequency characteristic, a peak point P and a cutoff point Q are set as shown in FIG. As a result, the frequency p of the peak point P, its intensity q, and the frequency of the cutoff point Q, that is, the cutoff frequency c are naturally determined.

As described above, the spatial frequency characteristic characterized by the peak point P and the cutoff point Q is not only for the expert who has been involved in image input and image processing by the scanner for many years, but also for those who are not, the spatial frequency characteristic. It has been confirmed that it is possible to intuitively understand what kind of characteristic the person has, and it matches the subjective feeling.

There is another important factor in determining the target spatial frequency characteristic. It has a spatial frequency of 0
What is the strength at that time? This target spatial frequency characteristic is used to determine the spatial frequency characteristic of the digital filter for performing digital filtering, as will be seen later, but the selection of strength when this spatial frequency is zero. If erroneous, the DC component of the image will be changed by the digital filtering using the digital filter.

Therefore, as shown by a point R in FIG. 3, when selecting the target spatial frequency characteristic, the spatial frequency is
The intensity at 0 is 1. As a result, the DC component of the image does not change before and after the digital filtering.

In this way, the three points P, Q, and R that characterize the target spatial frequency characteristic are determined, and if these three points are smoothly connected by a curve, the target spatial frequency characteristic can be determined by a mathematical expression. What kind of function is used to connect the three points of P, Q and R is arbitrary, but it is also possible to use a quadratic polynomial or a cubic polynomial.

Whatever function is used, the section from the point R, which is the direct current of the spatial frequency, to the cutoff point Q is divided into the section from the point R to the point P and the section from the point P to the point Q. By dividing into two sections, a curve passing through points R and P is defined in the former section, a curve passing through points P and Q is defined in the latter section, and these two curves are continuous at point P. Good.

For example, when a quadratic polynomial is used, in the section from point R to point P, a parabola passing through the points R and P and having a differential coefficient of 0 at the point P is determined, and the point P to the point is determined. In the section up to Q, a parabola passing through the points P and Q and having a differential coefficient of 0 at the point P may be defined. It is clear that these two curves connect smoothly at point P.

When a third-order polynomial is used, in the section from point R to point P, a cubic curve passing through points R and P and having differential coefficients of 0 at points R and P, respectively. In the section from point P to point Q, passing point P and point Q,
Moreover, it is only necessary to determine a cubic curve having a differential coefficient of 0 at each of the points P and Q. It is clear that these two curves connect smoothly at point P.

The following expression (9) is one general expression when the target spatial frequency characteristic is expressed by a cubic polynomial. In equation (9), f is the spatial frequency, A is the amplitude intensity,
p is the spatial frequency of the peak point, q is the intensity of the peak point, and c is the cutoff frequency. The spatial frequency is normalized to 1 in this equation.

[0057]

(Equation 3)

Specifically, in equation (9), p = 0.6, q
= 3 and c = 0.8, the target spatial frequency characteristic is as shown in Fig. 4a, and p = 0.6 in Eq. (9).
, Q = 8 and c = 0.8, the target spatial frequency characteristic is as shown in FIG. 4b.

Generally, the higher the peak, the sharper the image becomes. On the contrary, the lower the peak, the more blurred the image becomes. Further, when the spatial frequency of the peak is low, the edge becomes thick and the feeling is strong, and when the spatial frequency of the peak is high, a delicate feeling can be expressed. Further, it has been found that the image has a blurred look when the cutoff frequency is low.

The target spatial frequency characteristic can be selected as described above. However, it is needless to say that such a target spatial frequency characteristic can be selected whenever the need arises. However, it is preferable that a large number of spatial frequency characteristics are created in advance and stored in a database so that the database can be selected according to the purpose.

In that case, for each spatial frequency characteristic, for example, this spatial frequency characteristic is an approximation of the characteristic when the sharpness intensity a and the aperture number b are input in the scanner of the model A, or It is preferable to add a comment using a term used in a normal work, such as that the spatial frequency characteristic is skin-like and is suitable for making it slightly delicate and slightly sharp. This is because the operator can easily select a desired spatial frequency characteristic by referring to these comments.

Such comments include terms expressing the type of image such as mechanical and skin, terms expressing the degree of emphasis such as strengthening, weakening, and standard, and terms expressing the nature of emphasis such as delicate and powerful. It has been confirmed that they can be created in combination.

When the target spatial frequency characteristic is selected as described above, the digital filter is next designed (step S13). Now, the spatial frequency characteristic of the scanner obtained in step S11 of FIG. 1 is F _IN (f), step S
The target spatial frequency characteristic selected in 12 is F
_OUT (f), the spatial frequency characteristic of the digital filter is F _{DF
Assuming that} (f), F _OUT (f) = F _IN (f) F _DF (f) (10), the spatial frequency characteristic F of the digital filter is
It is obvious that _DF (f) can be obtained by F _DF (f) = F _OUT (f) / F _IN (f) (11).

By the way, in the step of designing the digital filter in step S13, each coefficient of the digital filter having the spatial frequency characteristic represented by the equation (11) is obtained. It is necessary to set the number.

It is needless to say that the number of taps can be arbitrarily set. However, the larger the number of taps, the better the spatial frequency characteristic defined by the equation (11) can be realized, but the calculation time increases. Therefore, there is a property that the time required for digital filtering becomes long. Therefore, it is desirable to use a larger number of taps within the allowable processing time range.

When calculating the coefficients of the digital filter, the spatial frequency V in the vertical direction of the image, the spatial frequency H in the horizontal direction of the image, and the amplitude intensity A as shown in FIG.
It is necessary to know the two-dimensional spatial frequency characteristics expressed in the axial space.

However, the one-dimensional spatial frequency characteristic is obtained by the above equation (11). Therefore, it is necessary to extend the one-dimensional spatial frequency characteristic of the equation (11) to the two-dimensional spatial frequency characteristic. For that purpose, the McClellan transform may be used.
Note that it is well known to extend the one-dimensional spatial frequency characteristic to the two-dimensional spatial frequency characteristic by using the McClellan transform, and the description thereof will be omitted.

Then, the two-dimensional spatial frequency characteristics thus obtained are sampled according to the number of taps determined, and the discrete Fourier transform is performed to obtain each coefficient of the digital filter. Since this is a process that is widely performed as the spatial frequency sampling method, detailed description is omitted.

The digital filter can be determined as described above, and the digital filter is used to perform digital filtering on the image data input from the scanner (step S2). It is obvious that the image data having the spatial frequency characteristic selected in step S12 of FIG. 1 can be obtained by this digital filtering process.

After performing the digital filtering process in step S2, a predetermined process such as outputting with a film output machine may be performed.

As described above, according to the present invention, the spatial frequency characteristics at each processing stage are clear, and therefore the image quality is managed at any stage by a clear element or standard called the spatial frequency characteristics. This makes it unnecessary to rely on the experience or intuition of the operator as in the conventional case.

Although the embodiments of the present invention have been described above, the present invention is not limited to the above embodiments and various modifications can be made. For example, in the above-described embodiments, the vertical and horizontal spatial frequency bands of the image are described as being equal, but it is obvious to those skilled in the art that the present invention can be applied to the case where the vertical and horizontal spatial frequency bands of the image are different. is there.

Further, step S11 and step 1 in FIG.
The steps of 2 may be reversed, and may be performed simultaneously in parallel. That is, it is important to approximate the spatial frequency characteristic of the scanner and the spatial frequency characteristic that determines the target image quality by mathematical expressions, and it does not matter in what order these steps are performed.

[Brief description of drawings]

FIG. 1 is a diagram for explaining steps when a spatial frequency characteristic setting method and a digital filter setting method according to the present invention are applied to image processing for printing plate making.

FIG. 2 is a diagram for explaining a process of selecting a target spatial frequency characteristic in step 12 of FIG.

FIG. 3 is a diagram for explaining a process of selecting a target spatial frequency characteristic in step 12 of FIG.

FIG. 4 is a diagram showing a specific example of target spatial frequency characteristics defined by equation (9).

FIG. 5 is a diagram for explaining a two-dimensional spatial frequency characteristic.

FIG. 6 is a diagram showing a configuration example of a contour emphasis circuit provided in the scanner.

FIG. 7 is a diagram showing a configuration example of a Laplacian filter.

[Explanation of symbols]

1 ... Half mirror, 2 ... Main aperture, 3 ... Optical system, 4 ... Photo receiver, 5 ... USM aperture, 6 ... Optical system,
7 ... Light receiver, 8 ... Subtraction circuit, 9 ... Counting circuit, 10 ... Addition circuit.

Claims (2)

[Claims] 1. A method for setting a spatial frequency characteristic, wherein a frequency of a peak, its intensity, and a cutoff frequency are determined, and these are smoothly connected by a predetermined function to set a target spatial frequency characteristic. 2. A method of setting a digital filter for processing image data read by an input device for reading an image into image data having a target two-dimensional spatial frequency characteristic, the method comprising: The one-dimensional spatial frequency characteristic is determined based on the one-dimensional spatial frequency characteristic-determining equation and the target one-dimensional spatial frequency characteristic-determining mathematical equation. And setting each coefficient of the digital filter of a desired size based on the two-dimensional spatial frequency characteristic.