JP3455397B2 - Image gradation interpolation method and apparatus, and recording medium recording a program for executing the processing - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for interpolating the gradation of a multi-gradation image.

[0002]

2. Description of the Related Art As one type of image processing using a computer, various gradation conversion is performed on a multi-gradation image. At this time, there is a problem that a gradation jump occurs due to the gradation conversion of the multi-gradation image. "Jump of gradation"
Means that the image level of the adjacent area changes stepwise,
A phenomenon in which at least one image level is missing.
When such a gradation jump is observed with the naked eye, it may appear as a pseudo contour. In other words, there is a problem that the original image does not have a contour, but the gradation jumps to make it look as if a contour were present.

[0003]

Conventionally, when such a gradation jump exists, the gradation jump is alleviated by averaging the pixel data. But,
When averaging the pixel data, the image is blurred.

The present invention has been made to solve the above-mentioned problems in the prior art, and an object thereof is to provide a technique for eliminating the gradation jump of a multi-gradation image without performing averaging processing. To do.

[0005]

In order to solve at least a part of the above-mentioned problems, a first invention is a method for interpolating gradation jump of a multi-gradation image. ) Generating the gradation-converted image data by converting the gradation of the multi-gradation image data representing the multi-gradation image, and (b) among the image levels included in the gradation-conversion image data, Identifying first and second image levels with intermediate tone jumps; and (c) processing the tone-converted image data to obtain regions having the first image level and the first image level. Determining first contour data representing an A-type contour line indicating a boundary with another region having an image level lower than the image level of (d) by processing the gradation-converted image data, With first image level And obtaining a second contour data representing the B type of contour line indicating a boundary between the other regions having area and said second image levels above image level that,
(E) determining the second inclusion in the area having the first image level by determining the mutual inclusion relationship between the A-type and B-type outlines based on the first and second outline data. A region adjacent to the image region having the image level of 10 is detected as a target region for gradation interpolation, and at least one set of contour lines indicating the contour of the target region is extracted from the A and B type contour lines. And (f) (N + 1) (N is 1 or more) between the A-type contour line and the B-type contour line in the one contour line set in the target region. An integer number of divided areas, and sequentially assigning the N intermediate image levels to N divided areas near the image area of the second image level among the (N + 1) divided areas, Equipped with.

According to the first aspect of the present invention, by specifying the first and second image levels in which the gradation jump exists among the image levels included in the gradation conversion data after the gradation conversion, The first and second image levels to be interpolated can be determined. Further, a target area for gradation interpolation is extracted from the A-type and B-type contour lines obtained in association with the first and second image levels, and the target area is (N + 1) segmented areas. , And N intermediate image levels are sequentially assigned. Therefore, the gradation of the target area can be interpolated without performing the averaging process of the pixel data.

In the first invention, the step (b)
Is a full level conversion configured by the converted data of each image level by converting all possible image levels of the multi-tone image data using the gradation conversion characteristic in the step (a). A step of generating data, detecting gradation jumps existing in the all-level converted data, and detecting lower and upper image levels sandwiching the gradation jumps from the first image level and the second image. It is preferable to include a step of determining each as a level.

In this way, it is possible to specify the first and second image levels to be subjected to gradation interpolation only by examining all level conversion data without examining the gradation conversion image data itself.

In the first invention, the step (f)
It is preferable that the method further includes the step of forming the (N + 1) segmented regions by determining the image level of each pixel in the target region according to Equation 1.

If the image level (ie, pixel data) of each pixel is determined according to Equation 1, the target area is substantially divided into (N + 1) divided areas having different image levels. That is, it is possible to realize gradation interpolation of the target area without setting a boundary line (intermediate contour line) between the (N + 1) number of divided areas.

Alternatively, in the above-mentioned first invention, the step (f) comprises: (f1) a shortest distance to the A type contour line in the one contour line set in the target region; It is preferable to include a step of forming the (N + 1) segmented areas so that N intermediate contour lines are created at positions having a predetermined relationship with the shortest distance to the B type contour line.

Here, "so that N intermediate contour lines are created" is not limited to the case where the intermediate contour lines are actually created, but (N + 1) divided areas are formed to create a virtual region. It also includes the case where a different intermediate contour line is formed.

In the position having the above-mentioned predetermined relationship, the shortest distance to the A-type contour line and the shortest distance to the B-type contour line in the one contour set are m.
It is preferable that the positions are in a relationship of pair n (m, n is an integer of 1 or more and N or less, and N combinations of m and n satisfy m + n = N + 1).

In this way, the relationship between the two types of shortest distances is m
The target region can be divided into (N + 1) divided regions by N intermediate contour lines at positions having a pair n relationship.

In the step (e), (e1) the A type contour line is defined as an A + type contour line whose inner region has an image level higher than the first image level, and the inner region is the first contour line. Classifying into an A-type contour having an image level less than one image level, and (e2) the B-type contour having an internal region having an image level greater than or equal to the second image level. B + type contour line,
Classifying the interior region into a B-type contour having an image level less than the second image level; and (e
3) At least one of an area sandwiched between at least the A + type outline and the B + type outline, and an area sandwiched between the B- type outline and the A- type outline. Is detected as the target area.

By classifying the contour lines in this way, it is possible to detect a target area which is "a region adjacent to an image region having a second image level in a region having a first image level". Is.

In the above first invention, further, (g)
It is preferable to include a step of extracting N intermediate contour lines existing at the boundaries of the (N + 1) divided areas, and (h) adding a fluctuation to the N intermediate contour lines.

Thus, the degree of fluctuation of the intermediate contour line is
It is possible to obtain a more natural multi-gradation image by matching the degree of fluctuation of the original A type contour line and the original B type contour line.

In the first invention, it is preferable that a step of enlarging the multi-tone image area is further provided before the step (c).

In this way, even if the target area is relatively small, the gradation can be smoothly interpolated.

A second invention is an apparatus for interpolating a gradation jump of a multi-gradation image, wherein the gradation conversion image data is converted by converting the gradation of the multi-gradation image data representing the multi-gradation image. And a gradation conversion unit that specifies first and second image levels having gradation jumps in the middle among the image levels included in the gradation conversion image data. By processing the gradation-converted image data, the region having the first image level and the first
The first converted image data by processing the gradation-converted image data while obtaining first outline data representing an A-type outline indicating a boundary with another region having an image level lower than the image level of Contour line detecting means for obtaining second contour data representing a B-type contour line indicating a boundary between a region having a level and another region having an image level equal to or higher than the second image level; By determining the mutual inclusion relationship between the A-type and B-type contour lines based on the second contour data, an image area having the second image level is determined in the area having the first image level. A target contour line that detects an adjacent region as a target region for gradation interpolation and extracts at least one set of contour lines indicating the contour of the target region from the A-type and B-type contour lines (N + 1) (N is an integer of 1 or more) between the output means and the outline of the type A and the outline of the type B in the set of outlines in the target area. An interpolating unit that divides the image into the divided areas and sequentially assigns the N intermediate image levels to the N divided areas near the image area of the second image level among the (N + 1) divided areas. .

A third aspect of the present invention is a computer-readable recording medium in which a computer program for interpolating a gradation jump of a multi-gradation image is recorded, which is a floor of multi-gradation image data representing a multi-gradation image. A tone conversion function that generates tone-converted image data by converting the tone,
Among the image levels included in the gradation conversion image data,
An interpolation gradation detection function for specifying first and second image levels in which a gradation jump exists in the middle, and a region having the first image level and the first and second image levels by processing the gradation conversion image data. The first contour data representing an A-type contour line indicating a boundary with another area having an image level less than one image level is obtained, and the gradation conversion image data is processed to obtain the first contour data. A contour line detecting function for obtaining second contour data representing a B-type contour line indicating a boundary between an area having an image level and another area having an image level higher than the second image level; By determining the mutual inclusion relation between the A-type and B-type contour lines based on the second contour data, the second contour is determined in the area having the first image level.
A region adjacent to the image region having the image level of 10 is detected as a target region for gradation interpolation, and at least one set of contour lines indicating the contour of the target region is extracted from the A and B type contour lines. (N + 1) (N is 1 or more) between the A-type contour line and the B-type contour line in the one set of contour line sets in the target area. An interpolating function for sequentially assigning the N intermediate image levels to N segmental regions near the image region of the second image level among the (N + 1) segmental regions. And a computer-readable recording medium in which a computer program for causing a computer to execute is recorded.

According to the second and third inventions, as in the first invention, the first and second image levels to be subjected to gradation interpolation can be determined, and the pixel data can be averaged. It is possible to interpolate the gradation of the target area without performing processing.

[0024]

Other Embodiments of the Invention The present invention also includes the following other embodiments. The first aspect is by being executed by a microprocessor of a computer system,
It is a program supply device for supplying a computer program for realizing each step or each means of the above invention through a communication path.

[0025]

DETAILED DESCRIPTION OF THE INVENTION

A. Outline of gradation interpolation method: Next, an embodiment of the present invention will be described based on examples. FIG. 1 is an explanatory diagram showing an outline of a gradation interpolation method according to the present invention. FIG. 1A shows a multi-tone image and changes in the image level thereof. This multi-tone image is divided into three regions R1 to R3. In this example, the outline of each region is represented by a rectangle for convenience of illustration, but the outline may have an arbitrary shape in a general image. The following processing is the same when the contour has any shape other than a rectangle.

As shown in the lower part of FIG. 1A, the region R2
Is an A level region, and the region R3 is a B level region. There is a gradation jump between the A level and the B level. In this embodiment, first, the contour lines L1 to L3 representing the contours of the regions R1 to R3 as shown in FIG. 1B are detected from the multi-tone image as shown in FIG. Then, of these contour lines L1 to L3, a region R2 having an A level and adjacent to a B level region is recognized as a target region for gradation interpolation, and this target region R2 and other regions are recognized. The contour line set {L2, L3} indicating the boundary of is extracted. Contour line L2
Is a contour line (referred to as “A type contour line”) indicating a boundary between the area R2 of A level and the area R1 of image level lower than A level. Further, the contour line L3 is a contour line (referred to as “B type contour line”) indicating a boundary between the area R2 at the A level and the area R3 at the image level higher than the B level. Region R surrounded by two types of contour lines
2 is the target area that is the target of interpolation of the gradation level (image level).

Inside the target area R2, the shortest distance from the A type contour L2 and the B type contour L
The target region R2 is divided into two divided regions R2 so that the intermediate contour line Lint is formed at a position where the shortest distance from 3 is equal.
It is divided into a and R2b (FIG. 1 (b)). As will be described in detail later, in this embodiment, the intermediate contour line Lint is not actually formed, and the virtual intermediate point Lint is determined by determining the image level of each pixel in each divided area R2a, R2b. Only the contour Lint is formed.

Of the two divided areas R2a and R2b, the pixels of the divided area R2b on the side closer to the B level area R3 are, as shown in FIG. An intermediate image level (M level) of the levels is assigned. As can be seen by comparing with the original multi-tone image of FIG. 1A, the tone jump is alleviated in the multi-tone image of FIG. 1C. Since the processing shown in FIG. 1 does not include the averaging processing of pixel data (image data), it is possible to eliminate the gradation jump of a multi-gradation image without blurring the image.

By the way, in a general multi-gradation image, a contour line set {L that surrounds the target region R2 having the A level is set.
2, L3} is not always easy to extract. Therefore, in this embodiment, as will be described later, a plurality of contour lines of a multi-tone image are provided in a plurality of types (A +, A-, B +, B-).
Of each type) and by determining the inclusion relation of each contour line, at least one contour line set representing the boundary between the target region and another image region is extracted. Then, by performing the processing as shown in FIG. 1 on the target area surrounded by each set of contour lines,
The gradation is interpolated.

FIG. 2 is an explanatory diagram showing an outline of a method of interpolating a plurality of gradations. In FIG. 2A, there are a plurality of (specifically two) gradation jumps between the image level (A level) of the region R2 and the image level (B level) of the region R3. That is, B = A + 3. Therefore, FIG. 2 (b)
In the process of 2, the target region R2 which is the A level is surrounded by 2
The target region R2 is divided into three divided regions R2a to R2 so that the intermediate contour lines Lint1 and Lint2 are generated at the positions where the shortest distances from the two contour lines L2 and L3 are 1: 2 and 2: 1.
It is classified into 2c. As shown in FIG. 2 (c), B
Areas R2b and R2c on the side closer to the level area R3
To the intermediate image levels M1 and M2, respectively.

[0031] In general, N pieces between A level and B level (N is an integer of 1 or more) intermediate image level M ₁ ~M _N of
Is interpolated, the shortest distance to the A type contour line L2 and the shortest distance to the B type contour line L3 are included in the contour line set {L2, L3} that surrounds the target region R2 at the A level. It is possible to set N intermediate contour lines at a position of m: n (m, n is an integer of 1 or more and N or less, and N combinations of m and n satisfy m + n = N + 1). At this time, N intermediate image levels are assigned in order of size to N divided areas from the side closer to the B level area among (N + 1) divided areas divided by the N intermediate contour lines. . By doing so, as shown in FIG. 2C, it is possible to obtain a smooth image in which the gradation jump is eliminated.

Hereinafter, a case where the target area is divided into two divided areas as in the example of FIG. 1 will be described. However, it is easy to extend the following description when the target area is divided into (N + 1) divided areas.

In the following description, the data indicating the A level is called "undermarker Md", and the data indicating the B level is called "upper marker Mu". The undermarker Md is a marker indicating the lower level A of the two image levels (A level and B level) to be subjected to the gradation interpolation, and the upper marker Mu is a marker indicating the upper level B. You can

B. Device Configuration: FIG. 3 is a block diagram showing a computer system for performing tone interpolation of a multi-tone image by applying the embodiment of the present invention. This computer system includes a CPU 10 and a bus line 12. The bus line 12 has a ROM 14 and an RA
M16 is connected, and a keyboard 30, a mouse 32, a digitizer 34, a color CRT 36, and a magnetic disk 38 are connected via an input / output interface 40.
And are connected.

In the RAM 16, the gradation converting means 100 and
Interpolation gradation detection means 102, target area detection means 42,
An application program for implementing the gradation interpolating means 44, the fluctuation detecting means 46, and the fluctuation adding means 48 is stored.

FIG. 4 shows means 100, 102, 42, 44.
3 is a functional block diagram showing the internal configuration of FIG. The gradation conversion unit 100 receives the multi-gradation image data from the image input unit 50, converts the gradation of the multi-gradation image data, and generates the gradation conversion image data. The interpolation gradation detection unit 102 specifies the gradation jump included in the gradation conversion image data and the image levels below and above each gradation jump. The target area detection unit 42 detects a target area for gradation interpolation.
Specifically, a contour line set surrounding the target area including the gradation jump is extracted. The gradation interpolation unit 44 executes gradation interpolation for each target area. The gradation-interpolated multi-gradation image is output from the image output unit 52. Image input unit 5
As 0, for example, a reading scanner can be used. Further, as the image output unit 52, a color CRT 36, a color printer or the like can be used. Further, the input / output may be the magnetic disk 38.

The tone converting means 100 includes a histogram changing section 112, a histogram 114, and a lookup table 116. Interpolation gradation detection means 102
Is the missing gradation detection unit 120 and the gradation reference table 122.
And an upper marker 124 and an under marker 126. The target area detection unit 42 includes a gradation contour line detection unit 60, a gradation contour line data memory 62, a target contour line extraction unit 64, and a contour line loop buffer 66. Further, the gradation interpolating means 44 includes a distance calculating section 70,
And a pixel value calculation unit 72. Functions of these units will be described later.

The software program (application program) for realizing the functions of the above-mentioned means and units is transferred from a portable storage medium (portable storage medium) such as a floppy disk or a CD-ROM to the main memory of the computer or It is transferred to an external storage device. Alternatively,
You may make it supply from a program supply apparatus to a computer via a communication path.

In this specification, the computer means
The concept includes a hardware device and an operating system, and means a hardware device that operates under the control of the operating system. Further, when the operating system is unnecessary and the hardware device is operated by the application program alone, the hardware device itself corresponds to the computer. The hardware device includes at least a microprocessor such as a CPU and a unit for reading a computer program recorded in a recording medium. The computer program includes a program code that causes such a computer to realize the functions of the above-described means. Note that some of the functions described above may be realized by the operating system instead of the application program.

The "recording medium" in the present invention includes a flexible disk, a CD-ROM, a magneto-optical disk, an IC card, a ROM cartridge, a punched card, a printed matter on which codes such as a bar code are printed, and an internal storage of a computer. Device (memory such as RAM and ROM)
Also, various computer-readable media such as external storage devices can be used.

C. Processing contents of the embodiment: FIG. 5 is a flowchart showing a processing procedure in the embodiment. In step T1, the image input unit 50 performs input processing of multi-tone image data, and in step T2, the tone converting means 10 is executed.
0 (FIG. 4) performs gradation conversion processing.

FIG. 6 is an explanatory diagram showing the contents of contrast correction using a histogram as a kind of gradation conversion processing and the subsequent gradation interpolation. FIG. 6A shows a histogram of original multi-tone image data. This histogram is a histogram 114 in the gradation converting means 100.
Is stored as. The user is shown in FIG.
On the histogram of the original multi-tone image data shown in (a),
Convert the darkest point Vls in the target range of gradation conversion to Vld,
Specifies that the brightest point Vhs be converted to Vhd. The gradation conversion unit 100 creates the lookup table 116 for gradation conversion of the multi-gradation image data according to this designation. The gradation conversion characteristic represented by the look-up table 116 is given by the following mathematical expression 2, for example.

[0043]

[Equation 2]

Here, Vs is the pixel value before conversion, Vd is the pixel value after conversion, Vhs and Vls are the brightest and darkest points before conversion, and Vhd and Vld are the brightest and darkest points after conversion. It is a point.

Equation 2 shows a conversion characteristic for linearly interpolating between the brightest point and the darkest point. Of course, any conversion characteristic other than linear interpolation can be obtained by using the lookup table 116.
Can be represented by It should be noted that a means other than the look-up table can be used as the means for performing the gradation conversion. For example, a function representing the gradation conversion characteristic may be used instead of the look-up table. However, the use of the lookup table has an advantage that gradation conversion can be performed at high speed.

When such gradation conversion is performed on the original multi-gradation image data shown in FIG. 6A, gradation conversion image data as shown in FIG. 6B is obtained. There is a gradation jump in this gradation conversion image data. Therefore, FIG.
The means 102, 42, and 44 shown in FIG. 6 perform gradation interpolation to create multi-gradation image data in which gradation is interpolated as shown in FIG.

FIG. 7 is an explanatory diagram showing another example of contrast correction using a histogram. In this example, FIG.
In a part of the histogram of the original multi-gradation image data shown in (a), the brightest point Vh in the target range of contrast correction
s and the darkest point Vls are designated. Therefore, only the pixel value Vs between the brightest point Vhs and the darkest point Vls is converted by the look-up table. At this time, FIG. 7 (c)
As shown in, the gradation interpolation is performed at the brightest point Vhd after gradation conversion.
May be executed only between the darkest point and the darkest point Vls.

The gradation converting means 100 shown in FIG. 4 can perform processing called histogram smoothing in addition to the above-described contrast correction. FIG. 8 is an explanatory diagram showing the result of the histogram smoothing process performed by the histogram changing unit 112 (FIG. 4) and the contents of the subsequent gradation interpolation. What is histogram smoothing?
This is a process in which the allocation of pixel values at each image level is determined in advance and the pixel values are allocated to each pixel according to this allocation. There are several possible allocation methods,
For example, when pixels are filled in order from the upper (or lower) density and the number of allocations is exceeded, the method is carried over to the next density or borrowing the number of allocations of the next density, or dividing the overlying pixels randomly. Then, there is a method of matching the book ends of the allocation. If the edges of the book are not aligned, gradation gradation will occur. FIG. 8B shows gradation-converted image data obtained as a result of the smoothing process. Since gradation jumps also exist in this gradation conversion image data, FIG.
The gradation is interpolated as shown in. The conversion with the same number of allocations may be called histogram averaging.

Hereinafter, a case will be described in which the gradation jump of the gradation conversion image data obtained by the gradation interpolation using the lookup table 116 is interpolated. However,
Even when the histogram smoothing process performed by the histogram changing unit 112 is performed as the tone conversion process, the tone interpolation can be performed in substantially the same manner.

In step T3 of FIG. 5, the initialization processing of the interpolation gradation detecting means 102 is performed. In this initialization processing, the under marker Md and the upper marker Mu are each initialized to 0, and all elements of the array table [] of the gradation reference table 122 are also initialized to 0. In addition,
The gradation reference table 122 is a one-dimensional array whose address indicates a gradation value.

In step T4 of FIG. 5, the gradation reference table 122 is created. FIG. 9 shows the gradation reference table 12
It is explanatory drawing which shows the content of the production process of 2. In the process of creating the gradation reference table 122, all possible image levels of the multi-gradation image data (0 to 0 for 8-bit data)
255) as test data, lookup table 1
Enter in 16. Then, the lookup table 116
A flag (indicated by "*" in FIG. 9) indicating that the output is obtained is registered in the gradation reference table 122 at the array position of the level output from the. The image level described as "NULL" is an image level where the output from the lookup table 116 cannot exist. Therefore, this “NULL” indicates a gradation jump (missing gradation) in the gradation conversion image data, and the image levels above and below that are used as the A level and the B level in the gradation interpolation processing.

As can be understood from the contents of the processing for creating the gradation reference table 122 described above, the gradation reference table 122 corresponds to all level conversion data in the present invention.

In steps T6 to T10 of FIG. 5, the missing gradation detection unit 120 detects a gradation jump existing in the gradation reference table 122 and specifies the under marker Md and the upper marker Mu for the gradation jump. The processing procedure to be performed is shown. First, in step T6, it is determined whether the content of the gradation reference table 122 at the position indicated by the undermarker Md is not "NULL" and the content at the position following the undermarker Md is "NULL". To be done. This determination is a determination for determining the value of the undermarker Md so that the undermarker Md indicates the A level (the lower level of the gradation jump). As you can see from Figure 9,
The flag "*" is registered in the image level immediately below the gradation jump "NULL", and the array element of the missing gradation level above that image level should be "NULL". is there. Therefore, when the determination in step T6 is YES, the level of the undermarker Md indicates the image level (A level) immediately below the gradation jump. If the determination in step T6 is no, in step T7 the under marker Md and the upper marker Mu are
Is incremented by 1 and step T6 is executed again. On the other hand, if the determination in step T6 is yes, the processing in step T8 and subsequent steps is executed. Figure 9
In this example, when the undermarker Md is 2, step T
The determination result in 6 is YES, and the process proceeds to step T8.

At step T8 in FIG. 5, the upper marker Mu is incremented by one. In step T9,
The content of the position of the upper marker Mu in the gradation reference table 122 is not "NULL" and the upper marker Mu is 1
It is determined whether or not the content at the position before this is "NULL". This determination is a determination for determining the value of the upper marker Mu so that the upper marker Mu indicates the B level (upper level of the gradation jump). As can be seen from FIG. 9, the flag “*” is registered in the image level immediately above the gradation jump “NULL”, and the registered content of the missing gradation level below the image level is “ It should be "Null". Therefore, when the determination in step T9 is YES, the level of the upper marker Mu indicates the image level (B level) immediately above the gradation jump. If the determination in step T9 is no,
In step T10, it is determined whether or not the upper marker Mu is less than the maximum value Max (the maximum value that the image level can take). If the upper marker Mu is less than the maximum value Max, the process returns to step T8 and the upper marker Mu is returned.
Is incremented by 1. On the other hand, the upper marker M
If u is greater than or equal to the maximum value Max, all gradation interpolation processing has been completed, so the image data after gradation interpolation is output to the image output unit 52 (FIG. 4). If the determination in step T9 is yes, the current value of the upper marker Mu is adopted. In the next step T11, gradation interpolation processing, which will be described in detail later, is executed. In this gradation interpolation process, the value of the under marker Md specified by the judgment of step T6 and the value of the upper marker Mu specified by the judgment of step T9 are used.

In the example shown in FIG. 9, the first gradation interpolation processing is performed at Md = 2 and Mu = 4 according to the processing procedure of FIG. When this gradation interpolation process is completed,
In step T12, the undermarker Md is set equal to the value of the upper marker Mu. Then, returning to step T6, the detection of the next gradation jump and the gradation interpolation processing are performed. In the example of FIG. 9, Md = 4 and Mu =
In 6, the second gradation interpolation processing is performed. Thus, according to the procedure of FIG. 5, the gradation reference table 122
A set of markers {Md, Mu} is determined for each of the gradation jumps existing in 1), and gradation interpolation is executed using these markers.

D. Details of gradation interpolation processing: FIG.
5 is a flowchart showing a processing procedure of gradation interpolation in step T11 of FIG. In step S1, the gradation contour line detection unit 60 (FIG. 4) processes the gradation-converted multi-gradation image data (gradation conversion image data), so that the gradation contours included in the multi-gradation image are processed. Detect lines. FIG. 11 is an explanatory diagram showing the entire multi-gradation image to be processed in the embodiment and the gradation contour lines detected in step S1. The multi-tone image shown in FIG. 11B is divided into six regions R1 to R6. In this example, the outline of each region is represented by a rectangle for convenience of illustration, but the following process is the same even when the outline of each region has an arbitrary shape.

FIG. 11C shows the change in image level on a line passing through the vicinity of the center of this multi-tone image. In FIG. 11B, the same type of hatching is applied to regions having the same image level. As can be seen, the regions R1 and R4 have an A level, and the region R3 has a B level. B level is
The image level is two levels higher than the A level. That is, there is one gradation jump from the A level to the B level. The regions R2 and R5 have a level lower than the A level, and the region R6 has a C level. The C level is an image level higher than the B level.

As illustrated in FIG. 9 described above, A
The level is indicated by the undermarker Md, and the B level is indicated by the upper marker Mu.

FIG. 11A shows the gradation contour lines L1 to L6 detected in step S1 of FIG.
FIG. 12 is an explanatory diagram showing a procedure for detecting the gradation contour lines L1 to L6 from the multi-gradation image. First, the multi-gradation image data shown in FIG.
The binarized image shown in (b-1) is obtained. FIG. 12 (b-1)
The binarized image of is the region R1, R3, R above the A level.
4, the sum area of R6 is 1 level, and the area below A level is 0
This is a level image. When the gradation contour lines are detected from this binarized image, three gradation contour lines L1, L2, L5 shown in FIG. 12B-2 are obtained. In this embodiment, the gradation contour lines L1, L2, L5 detected from the binarized image binarized at the A level are called "A type contour lines". The “A type contour line” is an A level image area,
In other words, it is a contour line that indicates a boundary with an image area having an image level lower than the A level.

The A type contour lines L1, L2 and L5 are further classified into "A + type contour lines" and "A- type contour lines". The "A + type contour line" is a contour line whose internal region has an image level of A level or higher,
That is, it is a contour line whose interior is filled in the binarized image. In the example of FIG. 12B-2, the contour line L1 is the “A + type contour line”. The "A-type contour line" is a contour line whose inner region has an image level lower than the A level, that is, a contour line whose inside is hollowed out in a binarized image. FIG. 12 (b-
In the example of 2), the contour lines L2 and L5 are “A-type contour lines”.

Regarding the method of detecting the contour line from the binarized image, Japanese Patent Laid-Open No. 5-2 disclosed by the present applicant.
Since it has been described in detail in Japanese Patent No. 42246, its details are omitted here.

In this embodiment, the A + type contour line L1 is used.
Is detected as a clockwise closed loop contour. On the other hand, the A-type contour lines L2 and L5 are detected as counterclockwise closed loop contour lines. Therefore, when the closed-loop contour lines L1, L2, L5 are detected, the rotation direction (clockwise or counterclockwise) is determined,
It is possible to distinguish whether each contour line is A + type or A- type.

The multi-gradation image data shown in FIG.
When binarized by level, the binarized image shown in FIG. 12 (c-1) is obtained. When a gradation contour line is detected from this binarized image, the B level 3 shown in FIG.
The gradation contours L3, L4, L6 of the book are obtained. These gradation contour lines L3, L4, L6 are also classified into B + type and B- type as shown in the figure. The “B + type contour line” is a contour line whose internal region has an image level of B level or higher, that is, a contour line whose interior is filled in a binarized image. The “B-type contour line” is a contour line whose inner region has an image level lower than the B level, that is, a contour line whose inside is hollowed out in a binarized image.

The gradation outline detected according to the method disclosed in Japanese Patent Laid-Open No. 5-24246 mentioned above is a boundary outline running along the boundary of pixels. FIG. 13 is an explanatory diagram showing the relationship between pixels and boundary contour lines. The boundary contour line passes through the boundary of pixels (circles with hatching). Therefore, there is an advantage that the boundary of the region having the width of one pixel can be accurately expressed. The position on the boundary contour is taken at the center of the circle described as "pixel representation of contour". Therefore, the position on the boundary line is displaced from the pixel position by 1/2 pixel in the vertical direction (main scanning direction) and the horizontal direction (sub scanning direction). In the interpolation process described later,
When obtaining the shortest distance between each pixel in the target area and the contour line, the distance from the position of the contour line to each pixel is calculated.

The gradation contour lines L1 to L6 thus detected
The contour data representing (FIG. 11A) is stored in the gradation contour data memory 62 (FIG. 4). In step S2 of FIG. 10, the target contour line extraction unit 64 detects the target region of the gradation interpolation processing, and also extracts the contour line set forming the boundary of the target region. FIG. 14 is an explanatory diagram showing an example of a region configured by various contour line sets. 14A to 14A and 14B-1 to 14A-1 to 14-14A to 14A-1 to 14A-1 to 14A-1 to 14A-4 to 14A-1 to 14A-1 to 14A-1 to 14A-1 to 14A-1 to 14A-1 to 14A-1 to 14A-1 to ...
There are 8 types of (b-4).

FIG. 14A-1 shows a region surrounded only by the A + type contour line. FIG. 14 (a-2)-
(A-4) shows a region surrounded by the A + type contour line and at least one of the A- type and B + type contour lines. In the four cases shown in FIGS. 14A-1 to 14A-4, gradation interpolation is required in FIG.
3) and (a-4). That is, A +
An A-level area inside the type contour line and outside the B + type contour line is detected as a target area for gradation interpolation. Note that, as shown in FIG. 14A-4, the A + type contour line may be included in the A + type contour line. Such an A-type contour line is also detected as a part of the contour line set forming the target area. In other words, in the case of FIG. 14 (a-3),
The A + type and B + type contours are extracted as a set of contour lines that form the boundary between the target area and another area. In the case of FIG. 14A-4, the A + type, B + type, and A− type contours are extracted as one set of contours that form the boundary between the target area and other areas.

FIG. 14 (b-1) shows a region surrounded only by a B-type contour line. FIG. 14 (b-2)-
(B-4) shows a region surrounded by a B-type contour line and at least one of a B + type contour line and an A-type contour line. Among the four cases of FIGS. 14B-1 to 14B-4, the target of the gradation interpolation in this embodiment is 2 of FIGS. 14B-3 and 14B-4. There are two cases. That is, the area at the A level, which is inside the B-type contour line and outside the A-type contour line, is detected as the target area for gradation interpolation. Note that FIG.
4 (b-4), B + in the B-type outline
There may be types of contours. This B + type contour line is also detected as a part of the contour line set forming the target area. In other words, in the case of FIG. 14B-3, the B-type and A-type contours are detected as one set of contours that form the boundary between the target area and another area. In the case of FIG. 14B-4, the B-type, A-type, and B + type contours are detected as a set of contour lines that form the boundary between the target area and another area. .

FIG. 15 is a flow chart showing the detailed procedure of step S2 in FIG. In step S11,
Contour data (contour) from the gradation contour data memory 62
Take out one. In step S12, the contour line forming the outer periphery of the target area is extracted by determining whether the extracted contour line is the A + type or the B- type. The A + type contour line may be a contour line that configures the outer periphery of the target area, as illustrated in FIG. 14A-3 or 14A-4. Further, the B-type contour line may also be a contour line forming the outer circumference of the target area, as shown in FIG. 14B-3 or 14B-4. Conversely, A-
Since the type and B + type contours do not form the outer periphery of the target area, in these cases, the process returns from step S12 to step S11, and the next contour data is extracted. On the other hand, if it is the A + type or the B- type, the contour line is registered as a parent loop (step S1).
3). Here, the “parent loop” means a contour line forming the outer periphery of the target area for gradation interpolation. In this embodiment, since each gradation contour line forms a closed loop, the contour line is simply referred to as a “loop”. The contour data of the parent loop is registered in the contour loop buffer 66 (step S14). In the example of the six contour lines L1 to L6 shown in FIG. 11A, since the first contour line L1 is the A + type, the contour line L1 is registered in the contour line loop buffer 66 as a parent loop.

In step S15, the contour line set forming the A-level target area is extracted. FIG. 16 is a flowchart showing the detailed procedure of step S15. In step S21, one piece of contour data is fetched from the gradation contour line data memory 62. However, in this case, the contour data registered as the current parent loop and the contour data already extracted as the contour line set forming the target area are excluded.

Steps S22 to S25 are a processing procedure for extracting a contour line (called a "child loop") included in the parent loop. In step S22, it is determined whether the type of the extracted contour line is the A-type or the B + type. The reason for this is that in the case of the regions of FIGS. 14A-3 and 14A-4 that are the targets of gradation interpolation, it is A that is combined with the parent loop (A + type contour line).
-Type and B + type contour lines, and similarly, FIG.
Also in the case of the areas (b-3) and (b-4), the combination with the parent loop (B-type contour line) is A-
This is because they are the outlines of type B and type B. If it is the A- or B + type, in step S23,
It is determined whether it is included in the parent loop.

FIG. 17 is an explanatory diagram showing the procedure for determining the inclusion relation of contour lines. In this embodiment, as shown in FIG.
The inclusion relation of the contour lines is determined by the two-step procedure of (b). First, in FIG. 17A, the inclusion relation is determined according to the range of the circumscribed rectangle of each contour line. That is, A
If the circumscribing rectangle of the − or B + type contour is included in the circumscribing rectangle of the parent loop, it is determined that it may be included. This determination can be easily made by comparing the coordinate ranges of the circumscribing rectangles. If there is a possibility of inclusion, inclusion determination is performed by the half-line method shown in FIG. 17 (b). The half line method draws a half line in any direction from one point on the contour line that may be included, and if this half line intersects the parent loop an odd number of times, it is determined that there is an inclusion relation, On the other hand, if there is an even number of intersections, it is a method of determining that there is no inclusion relationship. Of course, the inclusion relation with the parent loop may be determined by a method other than the method shown in FIG.

If it is determined that the loop is included in the parent loop, it is registered in the contour loop buffer 66 as a child loop in step S24 of FIG. After this,
The process returns from step S25 to step S21, and the processes of steps S21 to S24 are repeated for all contour lines. Thus, all child loops contained in the current parent loop are registered.

In step S26, the inclusion check between the child loops registered for the current parent loop is performed according to the method shown in FIG. A child loop included in the child loop (also referred to as “grandchild loop”) does not form a boundary between the target area and another area, and is excluded from the child loop. In this way, the contour data of the child loop left at the end is registered in the contour line loop buffer 66 (step S2).
7).

FIG. 18 is an explanatory diagram showing two contour line sets registered as a result of the processing of FIG. FIG.
(B) shows a first contour line set having the contour line L1 as a parent loop and the two contour lines L2 and L3 directly included therein as child loops. Further, FIG. 18C shows a second contour line set in which the contour line L4 is a parent loop and the two contour lines L5 and L6 directly included therein are child loops. When the first contour line set is detected, the A loop buffer 6 in the contour line loop buffer 66 is detected.
6a includes contour data D of the A + type parent loop L1.
_L1 and contour data D _{L2 of} the A-type child loop L2 are registered. Further, the contour data D _L3 of the B + type child loop _L3 is registered in the B loop buffer 66b. On the other hand, when the second contour line set is detected, the A loop buffer 6 in the contour line loop buffer 66 is
6a includes contour data D of the A-type child loop L5.
_L5 is registered. In the B loop buffer 66b, the contour data D _L4 and B of the B-type parent loop L4 are stored.
The contour data D _{L6 of the} + type child loop L6 is registered. As will be described later, when the distance calculation is performed in the target area for gradation interpolation, A type (A + type and A− type) is calculated at each pixel position in the target area.
And the shortest distance from the B type (B + type and B− type) contour lines.
In other words, the distance calculation does not distinguish between the A + type and the A− type, nor does it distinguish between the B + type and the B− type. Therefore, in the contour line loop buffer 66, the contour data is classified and registered into A type and B type.

As described above, A + type and A- type
The type distinction and the B + type distinction and the B- type distinction play an important role in obtaining a contour line set that constitutes the boundary between the target region and another region.

When the contour line set forming the target area for gradation interpolation is extracted in this way, the process proceeds to step S16 in FIG. In step S16, the area outside the A-type child loop in the inner area of the parent loop is filled with one level. In step S17, the area inside the B + type child loop is removed from the area thus filled. FIG. 19 is an explanatory diagram showing the processing contents of steps S16 and S17. Here, the case of the first contour line set shown in FIG. 18B is shown. First,
When the area inside the parent loop L1 (A + type) and outside the A- type child loop L2 is filled with one level, the bitmap image of FIG. 19B is obtained. Next, when the area inside the B + type child loop L3 is removed (set to 0 level), the bitmap image of FIG. 19C is obtained. By doing so, a binary bitmap image in which only the target area for gradation interpolation is filled can be obtained. The bitmap image of the target area thus obtained is temporarily stored in a memory area (not shown) in the RAM 16. In this way, the detection process of the target area in step S2 of FIG. 10 is completed.

In steps S3 to S8 of FIG. 10, the gradation interpolation means 44 (FIG. 4) executes gradation interpolation for each target area. FIG. 20 is a functional block diagram showing the function of the gradation interpolation means 44. The gradation interpolating means 44,
In addition to the distance calculation unit 70 and the pixel value calculation unit 72, it has a contour line / point sequence conversion unit 74, a target pixel buffer 76, an A loop contour point sequence buffer 78a, and a B loop contour point sequence buffer 78b. ing.

In step S3 of FIG. 10, the coordinates of each pixel in the target area are extracted from the bitmap image of FIG. 19C and serially stored in the target pixel buffer 76 (FIG. 20). In step S4, the A loop buffer 6
6a and the contour data (see FIGS. 18 (b) and 18 (c)) stored in the B loop buffer 66b are converted into the coordinates of the point sequence by the contour line / point sequence conversion unit 74, and the A loop contour point sequence It is stored in the buffer 78a and the B loop contour point sequence buffer 78b, respectively. "Contour line sequence"
Means the point sequence described as "pixel representation of contour line" in FIG. 13 described above.

In step S5 of FIG. 10, the distance calculator 7
0 retrieves the coordinates of one pixel from the target pixel buffer 76 and calculates the shortest distance D (Md) from the point sequence of the A loop and the shortest distance D (Mu) from the point sequence of the B loop. Here, D (Md) means the shortest distance from the point sequence of the undermarker Md (indicating A level), and D (Mu) is the point sequence of the upper marker Mu (indicating B level). Means the shortest distance from.

In step S6 of FIG. 10, the pixel value calculation unit 72 calculates the pixel value val (p) at the target pixel position according to the following mathematical formula 3.

[0081]

[Equation 3]

Here, val (Md) is the undermarker M
It is a value (= A level) indicated by d, and val (Mu)
Is a value (= B level) indicated by the upper marker Mu.

Generally, B level is N + 1 higher than A level.
Equation 3 can be rewritten as Equation 4 below.

[0084]

[Equation 4]

In this embodiment, since B = A + 2, N = 1
Is. The second term on the right side of Equation 4 is the first after being converted to an integer.
It is added to the A level of the term. Therefore, the shortest distance D (Md) to the A loop is the sum of the shortest distances (D (Md) + D (M
At pixel positions less than 1/2 of u)), the second term on the right side becomes 0, and the pixel value val (p) is equal to the A level. Also,
The shortest distance D (Md) to the A loop is the sum (D
At a pixel position of 1/2 or more of (Md) + D (Mu)),
The second term on the right side becomes 1, and the pixel value val (p) becomes A + 1 level.

When the A loop and the B loop partially overlap each other as shown in FIG. 21 (a), the point sequences of the overlapping portions as shown in FIG. 21 (b) cancel each other out and do not overlap each other. Only the partial point sequence is referenced in the distance calculation. By doing so, it is possible to shorten the processing time for distance calculation.

FIG. 22 is an explanatory diagram showing distributions of pixel values val (p) obtained for the first and second contour line sets (FIGS. 18B and 18C). As can be seen from FIG. 22, the target area at the A level is A loop and B
It is divided into two divided areas by a virtual contour line Lint located at a position where the shortest distance to the loop is equal. Of these two divided areas, the divided area closer to the B level area is set to the A + 1 level and the other divided area is A
To be kept at a level. As can be seen from the above description, in this embodiment, the contour data representing the intermediate contour line that defines the boundary between the A-level segment area and the A + 1-level segment area is not generated, but a virtual intermediate contour line is generated. Lint divides the target area into A-level and A + 1-level partition areas. Generally, N between A level and B level
When interpolating the number of intermediate levels, the target area is N
Divide them into +1 segmented areas, and divide them into A to (A + N)
It is possible to sequentially paint with N + 1 levels up to the level. In this case, the virtual intermediate contour line Lint is
The shortest distance to the A loop and the shortest distance to the B loop are m
Pairs are generated at positions having a relationship of pair n (m, n is an integer of 1 or more and N or less, and N combinations of m and n satisfy m + n = N + 1).

Of course, contour data representing N intermediate contour lines that divide the target area into N + 1 divided areas are actually generated, and respective levels are assigned to the divided areas defined by these intermediate contour lines. It is also possible to do so. In the present invention, "forming the intermediate contour line" is not limited to the case where the contour data representing the intermediate contour line is actually generated, and the equation 3 or the equation 4 is used as in the above embodiment.
As a result of the calculation of the pixel value by, there is a meaning including the case of generating a segmented region segmented by a virtual intermediate contour line.

Thus, the target pixel buffer 76 (see FIG.
By repeatedly executing steps S5 and S6 of FIG. 10 until 0) becomes empty, interpolated multi-tone image data is generated for the target region whose boundary is constituted by one set of contour lines. Then, the process returns from step S8 to step S3, and the process for the next set of contour lines is repeated. The multi-tone image data representing the interpolated target area is overwritten on the target area of the original multi-tone image data.

FIG. 23 is an explanatory diagram showing the result of gradation interpolation of another multi-gradation image. 23A shows a multi-tone image before interpolation, and FIG. 23B shows a multi-tone image after interpolation. As in the multi-tone image shown in FIG. 23A, the target area at the A level may include a plurality of areas at the B level. Also in this case, FIG.
By interpolating the gradation according to the procedure shown in FIG.
As shown in FIG. 3B, an intermediate gradation (that is, A + 1 level) area is formed so as to include all of the plurality of B level areas. As a result, there is an advantage that more natural gradation interpolation can be performed as compared with the conventional method. this is,
Generally, the same is true when gradation interpolation is performed so as to interpolate N intermediate image levels between the A level and the B level.

FIG. 24 is an explanatory diagram showing the result of gradation interpolation of another multi-gradation image. In this example, a considerable “fluctuation” (irregularity) is found in the gradation contour lines (contour lines of A type and B type) referred to during interpolation, but the gradation contour lines (intermediate There is not much fluctuation in the contour line. In such a case, the fluctuation detecting means 46 and the fluctuation adding means 48 (FIG. 3) can add an appropriate fluctuation to the intermediate contour line.

In this embodiment, fractalization processing is used as a method of adding fluctuation to the intermediate contour line. When the fluctuation is added by the fractalization process, the fluctuation detecting means 46 first measures the fractal dimension D of the gradation contour line (A loop and B loop) referred to during gradation interpolation.

The fractal dimension D is an index for quantitatively determining the complexity of an arbitrary curve. The fractal dimension D is
For example, it can be measured by the box counting method shown in FIG. When obtaining the fractal dimension for an arbitrary curve L as shown in FIG. 25A, first, a square having a width d on one side is arranged at the starting point Y0. Then, the intersection Y1 of this square and the curve L is obtained. Next, the intersection Y1 is reset to the center of the square, and the intersection Y2 between the square and the curve L is obtained again. In this way, by sequentially covering the curve L with squares, the number N (d) of squares required to cover the entire curve L is obtained. This number N (d)
Is a function of the width d of one side of the square. Then, the width d of one side is changed to obtain the number N (d) of squares required to cover the curve L, respectively. FIG. 25B is a logarithmic plot of the relationship between the number of squares N (d) and the width d of one side. As shown in FIG. 25 (B), the fractal dimension D is defined by the slope of this graph.

The fractal dimension does not depend on the size of the curve L. For example, if the curve L is simply enlarged by M times, the curve of FIG. 25 (B) is only shifted up by lnM. Therefore, the fractal dimension does not change by simple scaling. As described above, the fractal dimension is an index indicating the complexity of the curve without depending on the size of the curve.

The term "fractal" means a state in which even if a part of a curve (graphic) is enlarged, the same kind of structure is embedded in the inside. However, as described with reference to FIG. 25, the “fractal dimension” is not an index for only the fractal curve as described above, but the fractal dimension can be measured for any curve.

Generally, if the fractal dimension of a curve is increased, a natural fluctuation can be given when the curve is enlarged or the resolution of the curve is increased.

By the way, the fractalization process is usually performed recursively a plurality of times. The number of repetitions n of the fractalization process may be used by the user, or a preset value may be used. The curve becomes more complex as the number of iterations n of the fractalization process increases. Usually, it is preferable to set the number of repetitions n to a value of about 2 to 3.

When the fractalization dimension D of the contour line referred to in the gradation interpolation is measured, the fluctuation adding means 48
Fractalization processing of the intermediate contour line is executed. As described above, the contour data indicating the intermediate contour line is not created in the gradation interpolation processing shown in FIG. Therefore, prior to the fractalization processing (fluctuation addition processing) of the intermediate contour line, the contour data of the intermediate contour line is formed by a procedure similar to the procedure shown in FIG.

FIG. 26 is an explanatory diagram showing an outline of the fractalization processing (fluctuation addition processing) by the midpoint displacement method. Here, a case where the straight line L12 between the two points V1 and V2 is fractalized will be described. In the first conversion shown in FIG. 26 (A), first, the coordinates (x _m , y _m ) of the midpoint MV of the straight line L12 are obtained according to the following formula 5.

[0100]

[Equation 5]

Then, from this middle point MV, a point V3 is set in the direction perpendicular to the straight line L12. Here, the direction orthogonal to the straight line L12 (that is, the direction from the midpoint MV to the point V3)
The unit vector of is given by Equation 6 below.

[0102]

[Equation 6]

In the midpoint displacement method, the coordinates (x ₃ , y of the point V3 are
₃ ) is given by the following Equation 7.

[0104]

[Equation 7]

Here, Rg has an average value of 0 and a standard deviation of 1
Gaussian random number (random distribution whose value frequency distribution is Gaussian)
And D is the fractal dimension. As can be seen from the form of Expression 7, the second term on the right side of Expression 7 is from the midpoint MV to the point V.
This corresponds to the difference in coordinates up to 3. Hereinafter, this difference will be referred to as "midpoint displacement". It can be seen that the displacement of the midpoint depends on the fractal dimension D.

In the second conversion shown in FIG. 26B, the straight line between the points V1 and V3 and the straight line between the points V3 and V2 are fractalized. That is,
A new point V4 is set between the points V1 and V3,
A new point V5 is set between 3 and the point V2. However,
The positions of the points V4 and V5 are set to be opposite to each other along the traveling direction of a straight line passing through the three points V1, V3, V2 (for example, the direction from the point V1 to the point V2). Figure 2
6 (C) shows the third conversion.

Generally, the coordinates (x _{k + 1} , y _{k + 1} ) of the point generated in the k-th division are given by the following formula 8.

[0108]

[Equation 8]

Where (x _s , y _s ), (x _e , y _e ).
Are coordinates of both end points of the straight line portion before the k-th division is performed (see the example of FIG. 26C). Since the second term on the right side of Equation 8 (the term indicating the displacement of the middle point) includes 2- ^kD , the displacement at the middle point becomes smaller (that is, the curve becomes finer) as the number of divisions k increases. I understand. This division is repeated until the division section becomes smaller than a predetermined limit value (ε> 0) or a predetermined number of times.

As described above, the fractal dimension can be increased by recursively executing the fractalization processing by the midpoint displacement method. That is, it is possible to convert a relatively simple intermediate contour line into a more complicated curve and give a fluctuation close to the gradation contour line referred to in the gradation interpolation. In this specification, the term "curve" is used as a broad term including a straight line.

Note that the intermediate contour line obtained by the procedure shown in FIG. 12 is a boundary contour line running on the boundary of pixels, and the intermediate contour line shown in FIG.
As shown in (a), it becomes a stepwise contour line. The fractalization process by the midpoint displacement method is not very suitable for stepwise curves. Therefore, it is preferable to smooth the intermediate contour line as shown in FIG. 27B before fractalizing the intermediate contour line. Here, the smoothing process is performed according to the following Expression 9.

[0112]

[Equation 9]

Expression 9 means a calculation for changing the coordinates of the central contour point from the coordinates of the three contour points shown in FIG. 27A, for example. The left side of Expression 9 is the coordinates of the i-th contour point after smoothing. The constant a is 0 <
It is a value in the range of a <1. A value of about 0.2 is preferable as the constant a.

When the fluctuation is added to the intermediate contour line by the fractalization process of another kind suitable for the process of the stepwise curve instead of the midpoint displacement method, the smoothing process shown in FIG. 27 is unnecessary. Further, as a processing method for giving fluctuation to the intermediate contour line, a method other than the fractalization processing can be adopted. For example, it is possible to extract the high-frequency component by subjecting the intermediate contour line to a fast Fourier transform to give 1 / f fluctuation.

As described above, according to this embodiment, the look-up table 116 showing the gradation conversion characteristic is shown.
In FIG. 4, all possible image levels of the multi-tone image data are input as test data, and the tone reference table 122 is generated by the output. Then, since the gradation jump included in the gradation conversion image data is detected from the gradation jump existing in the gradation reference table 122, it is included in the gradation conversion image data without checking all the gradation conversion image data. It is possible to detect a gradation jump that occurs. Further, in the above-described embodiment, for each detected gradation jump, the contour lines of the image level regions on both sides of the gradation jump are detected. By investigating the inclusion relation of these contour lines, the contour line set that constitutes the boundary between the target region for gradation interpolation and another region can be easily extracted. Further, the pixel value of each pixel in the target area is determined according to the shortest distance to the A loop and the B loop of the contour line set, and the intermediate image level is assigned in the target area. Therefore, it is possible to smoothly perform the gradation interpolation without performing the averaging process of the pixel values.

E. Other Embodiments: FIG. 28 is an explanatory diagram showing a gradation conversion method for multi-gradation image data according to another embodiment of the present invention. This gradation conversion is gradation correction using a look-up table. FIG. 28A shows a gradation correction characteristic (gradation conversion characteristic) for correcting the input image data Vin into the output image data Vout. FIG. 28 (b)
As shown in, when the input image data Vin is input to the lookup table, the output image data Vout is output.

FIG. 29 shows an example of gradation correction according to various gradation correction characteristics. 29 (a-1), (b-1),
FIG. 29 (a-) shows a gradation correction curve.
2), (b-2), and (c-2) respectively show gradation images whose gradations have been corrected by these gradation correction curves. FIG. 29A-1 shows the input image data Vi.
n and the output image data Vout show the same characteristic, and therefore FIG. 29A-2 is the same as the original image before gradation correction.

In FIG. 29 (b-2), the image data Vou
There is a grayscale jump between the values of t of 0 and 3. Also,
In FIG. 29C-2, a gradation jump exists between the values of the image data Vout of 4 and 7. In this way, a gradation jump may occur as a result of the gradation correction.

FIG. 30 shows each means 100, 102, 42, 4 in the case of performing gradation correction by a look-up table.
4 is a functional block diagram showing the internal configuration of FIG. The configuration of FIG. 30 is obtained by replacing the histogram changing unit 112 and the histogram 114 in the configuration of FIG. 4 described above with a lookup table changing unit 118. The look-up table changing unit 118 changes the contents of the look-up table 116 according to the designation of the gradation correction curve by the user. The gradation correction curve is designated by the user by, for example, referring to FIG.
This is performed by displaying a gradation correction curve as shown in 9 (a-1) to (c-1) on the color CRT 36 and changing the shape of the curve using coordinate input means such as the mouse 32.

FIG. 31 is an explanatory diagram showing interpolation gradations detected for the image of FIG. 29 (c-2). The detection of the interpolation gradation is performed by the missing gradation detection unit 120 according to the procedure of steps T3 to T10 in FIG. As a result, the value of the under marker Md is determined to be 4, and the value of the upper marker Mu is determined to be 7. The gradation interpolation method using these markers is the same as that described with reference to FIG.

As described above, even when the gradation of the multi-gradation image is converted by the gradation correction using the look-up table, the gradation jump existing in the image after the gradation correction is detected and the gradation is corrected. Interpolation can be performed.

FIG. 32 is an explanatory diagram showing a gradation conversion method of multi-gradation image data by bit padding in still another embodiment of the present invention. In the bit padding, all zero bits are added to the lower side of the effective bit to expand the gradation resolution (dynamic range) of the image data. In the example of FIG. 32, 8 bits (256 gradations) are expanded to 16 bits (64,000 gradations). The bit padding is used when the gradation resolution of image data read by an input device such as a scanner or a digital camera is increased and a computer system corrects the image to have a higher quality.

FIG. 33 is an explanatory diagram showing the contents of gradation correction by bit padding and subsequent gradation interpolation. Figure 3
3A shows a histogram of original multi-tone image data, and FIG. 33B shows a histogram after 1-bit padding. Bit padding,
In this way, gradation jumps occur periodically. When the gradation interpolation of the present invention is performed on the multi-gradation image after bit padding, as shown in FIG. 33C, there is no gradation jump,
An image having a smooth gradation can be obtained.

FIG. 34 is a block diagram showing the internal structure of each means 100, 102, 42, 44 in the case of performing bit padding. In the configuration of FIG. 34, the histogram changing unit 112 and the histogram 114 of the configuration of FIG. 4 described above are omitted. The lookup table 116 contains
The gradation conversion characteristics corresponding to bit padding are registered.

FIG. 35 is an explanatory diagram showing interpolated gradations detected for an image subjected to 1-bit padding.
The detection of the interpolation gradation is performed by the missing gradation detection unit 120 according to the procedure of steps T3 to T10 in FIG. The value of the undermarker Md of the first set is 2, and the value of the upper marker Mu is 4. The value of the under marker Md of the second set is 4, and the value of the upper marker Mu is 6. Although omitted for the third and subsequent sets, similarly, the values of the under marker Md and the upper marker Mu for each gradation jump are determined. The gradation interpolation method using these markers is the same as that described with reference to FIG.

As described above, even when the gradation of a multi-gradation image is converted by bit padding, gradation jumps existing in the image after gradation correction can be detected and gradation interpolation can be performed. Is. The gradation conversion method is not limited to the method described above, and various other methods can be applied.

The present invention is not limited to the above embodiments and embodiments, but can be implemented in various modes without departing from the scope of the invention.
For example, the following modifications are possible.

(1) In the above-described embodiment, the gradation reference table 122 is created by inputting all possible image levels of the image data as test data into the look-up table representing the gradation conversion characteristics. A gradation jump (undermarker Md and upper marker Mu) is detected using the gradation reference table 122. However, the present invention is not limited to the case where the lookup table is used, but can be applied to the case where the gradation conversion characteristic is represented by another means such as a function. Even in this case,
It is preferable to generate all level conversion data composed of the converted data of each image level by converting all the image levels that a multi-tone image can take using the gradation conversion characteristic. This makes it possible to detect the gradation jump included in the gradation conversion image data without checking the gradation conversion image data itself.

As described above, instead of detecting the gradation jump included in the gradation conversion image data by using the gradation conversion characteristic, the gradation jump is detected by examining the gradation conversion image data itself. You may do it. For example, a histogram of gradation-converted image data may be created, and gradation jumps existing in the histogram may be detected.

(2) In the above embodiment, the entire multi-gradation image is targeted for the gradation interpolation processing, but a part of the multi-gradation image is specified and the gradation interpolation processing is performed only on that part. You may do it. As a method of specifying a part of the multi-tone image, a method in which the user specifies on the screen of the color CRT 36, a method of analyzing the multi-tone image data and automatically recognizing a region having a tone jump, and the like are available. Conceivable.

Particularly, when there is no gradation jump in the entire multi-gradation image, but there is a gradation jump in a partial area, only a part of the multi-gradation image is targeted for gradation interpolation. By doing so, there is an advantage that only the gradation jump of that portion can be interpolated well.

(3) The image area to be subjected to the gradation interpolation processing may be enlarged before the gradation interpolation. Fig. 36
FIG. 6 is an explanatory diagram showing an example of a case where gradation interpolation processing is performed after enlarging an image area. First, in FIG. 36A, assuming that the A level is 6 and the B level is 10, the target area (the area of A = 6 level) sandwiched between the A type contour line L10 and the B type contour line L20. Consider interpolating an intermediate image level (eg 8). However, since this target area has a width of one pixel, gradation interpolation cannot be performed. On the other hand, if this multi-tone image is enlarged twice, and the target region of A = 6 level is subjected to tone interpolation, an 8-level intermediate region can be formed as shown in FIG.
As a result, gradation interpolation is achieved and the gradation becomes smooth.

FIG. 36 shows an example of an extremely simplified case. Generally, the gradation is smoothed by executing the gradation interpolation after enlarging the target image area before the gradation interpolation. Can be In addition, after the gradation interpolation, the multi-gradation image after the gradation interpolation may be reduced at a desired magnification.

[Brief description of drawings]

FIG. 1 is an explanatory diagram showing an outline of a gradation interpolation method according to the present invention.

FIG. 2 is an explanatory diagram showing an outline of a method of interpolating a plurality of gradations.

FIG. 3 is a block diagram showing a computer system for performing gradation interpolation of a multi-gradation image by applying the embodiment of the present invention.

FIG. 4 is a functional block diagram showing an internal configuration of means 100, 102, 42, 44.

FIG. 5 is a flowchart showing a processing procedure in the embodiment.

FIG. 6 is an explanatory diagram showing the contents of contrast correction using a histogram as one type of gradation conversion processing and subsequent gradation interpolation.

FIG. 7 is an explanatory diagram showing another example of contrast correction using a histogram and the content of gradation interpolation after that.

FIG. 8 is an explanatory diagram showing the result of histogram smoothing processing and the content of subsequent gradation interpolation.

FIG. 9 is an explanatory diagram showing the contents of a process of creating a gradation reference table 122.

FIG. 10 is a flowchart showing a processing procedure of gradation interpolation in the embodiment.

FIG. 11 is an explanatory diagram showing the entire multi-gradation image to be processed in the embodiment and the gradation contour lines detected in step S1.

FIG. 12 is an explanatory diagram showing a procedure for detecting gradation contour lines L1 to L6 from a multi-gradation image.

FIG. 13 is an explanatory diagram showing a relationship between a pixel and a boundary contour line.

FIG. 14 is an explanatory diagram showing examples of areas indicated by various contour line sets.

FIG. 15 is a flowchart showing a detailed procedure of step S2.

FIG. 16 is a flowchart showing the detailed procedure of step S15.

FIG. 17 is an explanatory diagram showing a procedure for determining the inclusion relationship of contour lines.

FIG. 18 is an explanatory diagram showing two sets of extracted contour lines.

FIG. 19 is an explanatory diagram showing the processing contents of steps S16 and S17.

FIG. 20 is a functional block diagram showing the function of the gradation interpolating means 44.

FIG. 21 is an explanatory diagram showing a contour line portion referred to in distance calculation.

FIG. 22 is an explanatory diagram showing distributions of pixel values val (p) obtained for the first and second contour line sets.

FIG. 23 is an explanatory diagram showing a result of gradation interpolation of another multi-gradation image.

FIG. 24 is an explanatory diagram showing a result of gradation interpolation of another multi-gradation image.

FIG. 25 is an explanatory diagram showing a method for measuring fractal dimension.

FIG. 26 is an explanatory diagram showing an outline of fractalization processing (fluctuation addition processing) by the midpoint displacement method.

FIG. 27 is an explanatory diagram showing a method of smoothing an intermediate contour line.

FIG. 28 is an explanatory diagram showing a gradation conversion method for multi-gradation image data by gradation correction using a look-up table according to another embodiment of the present invention.

FIG. 29 is an explanatory diagram showing an example of gradation correction according to various gradation correction characteristics.

FIG. 30 is a functional block diagram showing the internal configuration of each means 100, 102, 42, and 44 when performing gradation correction using a lookup table.

31 is an explanatory diagram showing interpolation gradations detected for the image in FIG. 29 (c-2).

FIG. 32 is an explanatory view showing a gradation conversion method for multi-gradation image data by bit padding in still another embodiment of the present invention.

FIG. 33 is an explanatory diagram showing the contents of gradation correction by bit padding and subsequent gradation interpolation.

FIG. 34: Means 100 for Bit Inflating,
The functional block diagram which shows the internal structure of 102,42,44.

FIG. 35 is an explanatory diagram showing interpolation gradations detected for an image in which 1-bit padding has been performed.

FIG. 36 is an explanatory diagram showing a case where gradation interpolation processing is performed after the image area is enlarged.

[Explanation of symbols]

10 ... CPU 12 ... Bus line 14 ... ROM 16 ... RAM 30 ... Keyboard 32 ... Mouse 34 ... Digitizer 36 ... Color CRT 38 ... Magnetic disk 40 ... Input / output interface 42 ... Target area detecting means 44 ... Gradation interpolation means 46. Fluctuation detecting means 48. Fluctuation adding means 50 ... Image input section 52 ... Image output unit 60 ... Gradation contour detection section 62 ... Gradation contour data memory 64 ... Target contour line extraction unit 66 ... Contour loop buffer 70 ... Distance calculation unit 72 ... Pixel value calculator 74 ... Contour line / point sequence conversion unit 76 ... Target pixel buffer 78a ... A loop contour point sequence buffer 78b ... B loop contour point sequence buffer 100 ... Gradation conversion means 102 ... Interpolation gradation detection means 112 ... Histogram change section 114 ... Histogram 116 ... Look-up table (LUT) 118 ... Lookup table change section 120 ... Missing gradation detection unit 120 122 ... gradation reference table 124 ... Upper marker Mu 126 ... Undermarker Md

─────────────────────────────────────────────────── ─── Continuation of the front page (56) Reference JP-A-9-60156 (JP, A) JP-A-9-114972 (JP, A) JP-A-2-248159 (JP, A) JP-A-1- 61878 (JP, A) JP 62-292074 (JP, A) JP 61-91661 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) H04N 1/40-1 / 409

Claims (11)

(57) [Claims] 1. A method of interpolating a gradation jump of a multi-gradation image, comprising: (a) converting the gradation of multi-gradation image data representing a multi-gradation image to obtain gradation conversion image data. A step of generating, (b) a step of specifying first and second image levels having a gradation jump in the middle among the image levels included in the gradation conversion image data, and (c) the floor Processing the tonal transformed image data to represent a first type A contour line delineating a boundary between an area having the first image level and another area having an image level less than the first image level. A step of obtaining contour data,
(D) By processing the gradation-converted image data, a B-type contour line indicating a boundary between an area having the first image level and another area having an image level equal to or higher than the second image level And (e) determining the mutual inclusion relation between the A-type and B-type contour lines based on the first and second contour data. Area adjacent to the image area having the second image level is detected as a target area for gradation interpolation, and at least one contour line set indicating the contour of the target area is set. Extracting from the A-type and B-type contours, (f) in the target area, the A-type contour and the B-type contour in the one set of contours Between (N + 1 (N is an integer greater than or equal to 1) segmented regions, and the N intermediate images are divided into N segmented regions near the image region of the second image level among the (N + 1) segmented regions. And a step of sequentially assigning levels. 2. The gradation interpolation method according to claim 1, wherein the step (b) uses all of the gradation conversion characteristics obtained in the step (a) to obtain all possible multi-gradation image data. A step of generating all-level conversion data composed of data after conversion of each image level by converting the image level; detecting a gradation jump existing in the all-level conversion data; The lower and upper image levels sandwiching between the first image level and the second image level are
And a step of determining each as an image level of
Tone interpolation method. 3. The gradation interpolation method according to claim 1, wherein in the step (b), the first and second gradation jumps for a plurality of gradation jumps existing in the gradation conversion image data are included. A gradation interpolation method, comprising: determining each set of image levels, wherein steps (c) to (f) are performed for each set of the first and second image levels. 4. The gradation interpolation method according to claim 1, wherein in the step (f), the image level of each pixel in the target area is determined according to Expression 1. , A gradation interpolation method, which comprises the step of forming the (N + 1) number of divided areas. [Equation 1] Here, val (p) is the image level of each pixel in the target area, val (Md) is the first image level, and va
l (Mu) is the second image level, D (Md) is the shortest distance to the A type contour, and D (Mu) is the shortest distance to the B type contour. 5. The gradation interpolation method according to claim 1, wherein the step (f) includes (f1) in the target region,
N intermediate contour lines are created at positions where the shortest distance to the A type contour line and the shortest distance to the B type contour line in the one contour set have a predetermined relationship. Gradation interpolation method, comprising the step of forming the (N + 1) segment areas. 6. The gradation interpolation method according to claim 5, wherein the position having the predetermined relationship is the shortest distance to the A-type contour line in the one contour set, and B
The shortest distance to the contour of the type is m to n (m and n are integers from 1 to N, and N combinations of m and n are m + n).
= N + 1), the gradation interpolation method. 7. The gradation interpolation method according to claim 1, wherein in the step (e), (e1) the A-type contour line,
Classifying into an A + type contour line whose internal region has an image level equal to or higher than the first image level and an A- type contour line whose internal region has an image level lower than the first image level. And (e2) the B type contour line, the B + type contour line whose inner region has an image level equal to or higher than the second image level, and the inner region whose image level is lower than the second image level. With B-
(E3) a region sandwiched between at least the A + type contour line and the B + type contour line, the B− type contour line, and the A−
And a step of detecting at least one of a region sandwiched by the contour line of the type as the target region. 8. The gradation interpolation method according to claim 1, further comprising: (g) extracting N intermediate contour lines existing at the boundaries of the (N + 1) segment areas. And (h) adding fluctuations to the N intermediate contour lines. 9. The gradation interpolation method according to claim 1, further comprising the step of enlarging the multi-tone image area before the step (c). . 10. A device for interpolating a gradation jump of a multi-gradation image, wherein the gradation conversion image data is generated by converting the gradation of multi-gradation image data representing the multi-gradation image. A tone conversion means, and among the image levels included in the tone conversion image data,
Interpolation gradation detection means for specifying first and second image levels having gradation jumps in the middle; and processing the gradation converted image data to obtain a region having the first image level and the area having the first image level. By obtaining first contour data representing an A type contour line indicating a boundary with another region having an image level less than one image level, and processing the gradation-converted image data,
Contour line detecting means for obtaining second contour data representing a B type contour line indicating a boundary between an area having the first image level and another area having an image level equal to or higher than the second image level, By determining the mutual inclusion relationship between the A-type and B-type contour lines based on the first and second contour data, the second image level in the region having the first image level is determined. A target contour for detecting a region adjacent to the image region having the following as a target region for gradation interpolation, and extracting at least one set of contour lines indicating the contour of the target region from the A and B type contour lines. Line extraction means, and (N + 1) (N is an integer of 1 or more) between the A type contour line and the B type contour line in the one set of contour line sets in the target region. In the division area of Of the (N + 1) divided areas, the interpolation means sequentially assigns the N intermediate image levels to the N divided areas close to the image area of the second image level. . 11. A computer-readable recording medium recording a computer program for interpolating a gradation jump of a multi-gradation image, wherein the gradation of multi-gradation image data representing the multi-gradation image is converted. By this, a gradation conversion function for generating gradation-converted image data, and among the image levels included in the gradation-converted image data,
An interpolation gradation detection function for specifying first and second image levels having a gradation jump in the middle; and a region having the first image level and the first and second areas by processing the gradation conversion image data. By obtaining first contour data representing an A type contour line indicating a boundary with another region having an image level less than one image level, and processing the gradation-converted image data,
A contour line detecting function for obtaining second contour data representing a B type contour line indicating a boundary between an area having the first image level and another area having an image level equal to or higher than the second image level, By determining the mutual inclusion relationship between the A-type and B-type contour lines based on the first and second contour data, the second image level in the area having the first image level is determined. A target contour for detecting a region adjacent to the image region having the following as a target region for gradation interpolation, and extracting at least one set of contour lines indicating the contour of the target region from the A and B type contour lines. Line extraction function, and (N + 1) (N is an integer of 1 or more) between the A type contour line and the B type contour line in the one set of contour line sets in the target area. In the division area of In order to realize the computer, an interpolation function of sequentially allocating the N intermediate image levels to the N divided areas close to the image area of the second image level among the (N + 1) divided areas A computer-readable recording medium recording the computer program of.