JPH10105681A - Image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent the quality degradation of a target image by performing pixel interpolation by using only the picture elements found in a source image included in the substantial area of the target image while referring the source image and both the cutting masks having the resolution of the target image. SOLUTION: A target cutting mask 5, 1st reference cutting mask 3 and 2nd reference cutting mask 4 are prepared from a source cutting mask 2. Besides, a target image 6 is prepared from a source image 1 of resolution A while referring to the 1st reference cutting mask 3 of the same resolution A and the 2nd reference cutting mask 4 of resolution B. The target cutting mask 5 is prepared by performing resolution conversion for the source cutting mask 2. At such a time, label values (s) and (t) are determined at the pixel positions of the target image and the source image and it is determined whether these pixel positions are included in the substantial area or a background area. Then, only the values of picture elements surrounding a prescribed point position in the substantial area of the source image are used for pixel interpolation. Thus, pixel values in the background area are not mixed into the pixel values in the substantial area.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a target image obtained by cutting out a predetermined area in an original image using a cutout mask corresponding to the original image (this image is referred to as a target image in this specification). The present invention relates to an image processing method for obtaining

[0002]

2. Description of the Related Art When dealing with digital images, scaling, rotation,
Performing image processing such as deformation is widely performed. The scaling process has the same meaning as the resolution conversion process described later.

[0003] Such image processing is generally described as follows.
It can be said that the pixel value at the position (x, y) of the original image is converted to the position (u, v) of the target image. Therefore, a function for converting (u, v) to (x, y) is represented by f
Then, such image conversion can be said to be a process of performing an operation of a function f ^{-1 on} an original image to obtain a target image. Here, the function f ⁻¹ is an inverse transform of the function f.

For example, if the function f is

[0005]

(Equation 1)

[0006] When the linear transformation is represented by the following equation, it represents a transformation for scaling the original image a times. Note that, in equation (1), a
> 1 indicates enlargement, and 0 <a <1 indicates reduction.

The function f is

[0008]

(Equation 2)

In the case of the linear transformation represented by the following equation, the transformation is rotated by the angle θ.

An interpolation method is generally used as a method for obtaining a target image from an original image using the function f ^-1 as described above. Interpolation means that, for each pixel of a target image, the position (u, v) (u, v is an integer value) is converted by a function f to obtain a corresponding position (x, y) on the original image. If the obtained position does not match the lattice point on the original image, that is, the position of the pixel of the original image, the position (x,
This is a method in which the value of the position is obtained from the values of the pixels around y), and the obtained value is used as the value of the pixel at the position (u, v) of the target image. A bilinear interpolation method, a cubic convolution interpolation method, and the like are known.

An example is as follows. In the following, the original image is assumed to be raster image data.

Now, the pixels of the original image are arranged in the (x, y) space as shown in FIG. 4A, and the pixels of the target image are arranged in the (u, v) space as shown in FIG. Assuming that the pixels are arrayed, the pixel Q at the position (u, v) of the target image is subjected to conversion by the function f and the position (x, y) on the original image
Is obtained, the position of P does not match the position of the pixel of the original image. Therefore, the value at the position P is obtained by interpolation using the values of the pixels around the position P, and the obtained value is used as the value of the pixel Q of the target image. In this case, when the bilinear interpolation method is used as the interpolation method, the values of the _four pixels P ₁ , P ₂ , P ₃ , and P ₄ surrounding the position indicated by P in the figure are used, and the cubic convolution interpolation method is used. In this case, the values of 16 pixels surrounding the position indicated by P are used.

Since the nearest neighbor interpolation method, bilinear interpolation method, and cubic convolution interpolation method are all well known, detailed description will be omitted.

In general, the pixel value may be any value as long as it forms a linear space on a real number field.
These are values such as density and color. When the pixel value is a color, a three-dimensional vector of (R, G, B) or (C, M,
(Y, K) of course.

[0015]

When an image is printed, the image may be cut out and used because of the necessity of the design. Image clipping refers to extracting an object of interest in an image and fitting it into white or monochrome, or fitting it into another image.

When an image is clipped, data describing area information for indicating which part of the image is to be extracted is required. This data is called a clipping mask, and its data format takes a value of 0 or 1 in an area to be extracted in the image (hereinafter, referred to as a substantial area),
In areas that are not extracted (hereinafter referred to as background areas), a raster binary image format that takes 1 or 0, a run-length compressed form, or a boundary line between the real area and the background area is expressed by a broken line or curve There are vector formats.

The cutout mask can be generally referred to as area data, and has a function of extracting an area label value corresponding to an arbitrary position by referring to the area data. It is only necessary to have it. Therefore, the region data need not be binary, and may be a multi-valued image in which the image is divided into regions and labels with different values are assigned to the respective regions.

In the following, for easy understanding, the cutout mask is assumed to be in a raster binary image data format, but it is obvious that other formats may be used. Then, it is assumed that the pixels in the real area have a value of 0 and the pixels in the background area have a value of 1. Various methods are known for the method of forming the cutout mask.

FIG. 5 shows an example of an original image and a cutout mask corresponding to the original image. FIG. 5A shows the original image, with a triangle and
Circles and rectangles are drawn. FIG. 5B shows an example of a cutout mask. According to this example, it is specified to extract a triangular pattern from an original image. In this cutout mask, the values of all the pixels in the triangle serving as the real region are 0, and the values of all the pixels outside the triangle are 1.

When an image is cut out using a clipping mask, the resolution of the original image may differ from the resolution of the clipping mask. In such a case, conventionally, first, a resolution conversion process, that is, a scaling process is performed on each of the original image and the clipping mask, and an image having the same resolution as the resolution of the target image and the clipping mask are formed. In general, a pixel of the first converted image corresponding to a pixel having a value of 0 in the clipping mask is extracted.

However, in such a method, the image quality may be noticeably deteriorated at the outline of the extracted image. This is for the following reasons.

Now, the pixels of the target image are arranged in the (u, v) space as shown in FIG. 6A, and the pixels of the original image are arranged in the (x, y) space as shown in FIG. 6B. And Further, when the cutout mask is converted to the same resolution as the target image, a part of the cutout line is located at a position indicated by L in FIG. 6A, and a pixel on the cutout line L and a pixel on the left side thereof are extracted. Suppose

It is assumed that the pixel Q of the target image is located at the position indicated by P when converted into the original image, and the cutout line L is located at the position indicated by L 'in FIG. 6B when converted into the original image.

At this time, as described above, the value of the position of P in the original image is obtained from the values of the pixels around it, and the value is used as the value of the pixel Q in the target image. When the value of the position of P is obtained, the value of the pixel indicated by P _B1 in FIG. 6B is used. However, since the pixel indicated by P _B1 is on the right side of the cutout line L ′, Pixels in the area.

That is, the pixel Q in the substantial area of the target image
Also reflects the value of the pixel P _{B1 in} the background area of the original image. As a result, when the value of the pixel is a color, the color of the pixel P _{B1 in} the background area of the original image oozes out in the color of the pixel Q, and the quality of the clipped image deteriorates. It will be.

The same applies to the case where the cubic convolution interpolation method is used as the interpolation method.

As described above, performing the pixel interpolation to convert the resolution means that the pixel values of the target image are created by mixing the values of some pixels of the original image at an appropriate mixing ratio. This makes it possible to maintain the smoothness of the image and to suppress the occurrence of graininess at the contour portion, but when performing image clipping using a clipping mask, the above-described inconvenience may occur. It is. In particular, when the pixel value is a color, the color of the real region and the color of the background region often differ significantly. Is created, and the image quality is degraded.

An object of the present invention is to solve the above-mentioned problems, and an object of the present invention is to provide an image processing method which does not degrade the quality of a target image when performing image clipping.

[0029]

The above-mentioned conventional problem is caused by using the values of the pixels in the background area of the original image when forming the pixels in the substantial area of the target image. I understand. Therefore, the above-described problem can be solved by using only the values of the pixels in the substantial area of the original image when creating the pixels in the substantial area of the target image.
Referring to FIG. 6, assuming that a bilinear interpolation method is used as an interpolation method, conventionally, when a value of a position indicated by P is obtained, four pixels P _R1 , P
_Although the values of _R2 , P _R3 , and P _B1 are used, in the present invention, when the value of the position indicated by P is obtained by pixel interpolation, it is a pixel surrounding the position of P and the actual area of the original image. _Are used only the values of the three pixels P _R1 , P _R2 , and P _R3 . The same applies when an interpolation method other than the bilinear interpolation method is used.

Therefore, in order to achieve the above object, the image processing method of the present invention provides an object having a predetermined resolution based on an original image and a cutout mask corresponding to the original image and having a predetermined resolution. An image processing method for creating an image, comprising: creating a clipping mask having a resolution of an original image and a clipping mask having a resolution of a target image; and referencing both of these masks to create an entity of the target image. It is characterized in that a pixel in the original image that enters the region is obtained, and pixel interpolation is performed using only the obtained pixel to obtain a target image.

[0031]

Embodiments of the present invention will be described below with reference to the drawings. Here, the resolution of the original image is A, the resolution of the original clipping mask is A ', the resolution of the target image is B, and the resolution of the target clipping mask is B'. Here, it is natural that A = A 'and B = B'.

A function for converting a target image to an original image and a function for converting a target clipping mask to an original clipping mask are both f, and an inverse function of the function f is represented by f ^-1 . Here, f may be an identity map.

First, as shown in FIG.
And an original cutout mask 2 corresponding thereto. The original cutout mask may be created by an appropriate method.

First, three types of cutout masks, ie, the target cutout mask 5, the first reference cutout mask 3, and the second reference cutout mask 4, are created from the original cutout mask 2. Further, a target image 6 is created from the original image 1 while referring to the first reference clipping mask 3 and the second reference clipping mask 4. The resolution of the first reference cutout mask 3 is A
, Ie, a clipping mask having the same resolution as that of the original image 1, and the second reference clipping mask 4 is a clipping mask having a resolution B, that is, a clipping mask having the same resolution as that of the target image 6.

The target cutout mask 5 can be created by subjecting the original cutout mask 2 to inverse conversion of f ^-1 and then to resolution conversion. This is described as “(resolution conversion) （f ⁻¹ ” in FIG. That is, for each pixel of the target clipping mask 5, the coordinates (u,
v) is subjected to conversion f to obtain a position (x, y) on the original clipping mask, and a value of pixels around the position (x, y) is calculated at a position (x, y) by an appropriate interpolation method. Is calculated, and a value obtained by binarizing the value with an appropriate threshold value, for example, 0.5, is used as a value at the coordinates (u, v) of the target cutout mask. it can. The second reference cutout mask 5 can be created in a similar manner. The creation of the target image 6 from the original image 1 may be performed in the same manner. On the other hand, the first reference cutout mask 3 can be created only by subjecting the original cutout mask 2 to resolution conversion processing.

Next, the processing shown in FIG. 2 is performed using the first reference clipping mask 3, the second reference clipping mask 4, and the original image 1.

First, in step S1, one pixel is selected from the second reference cutout mask 4. Let the position of that pixel be (u, v). Since the position of the pixel is on a grid point, u and v are both integer values.

Next, at step S2, the position (u,
v) is transformed f and the position (x,
y). x and y are real values.

In step S3, a label value s is determined for the pixel position (u, v) selected in step S1.
Here, the label value s at the position (u, v) is
It is assumed that the value is the value of the pixel of the second reference cutout mask 4 selected in Step 1.

The label value s indicates whether the pixel at the position (u, v) in the target image is a pixel in the real area or a pixel in the background area. In this case, if s = 0, the pixel located at this position (u, v) of the target image is in the substantial area, and if s = 1, this pixel (u, v) of the target image
Will be in the background area.

Next, at step S4, the position of a pixel on the original image 1 used for obtaining the value of the position (x, y) on the original image 1 obtained at step S2 is obtained. The position of such a pixel on the original image 1 differs depending on what interpolation method is used to determine the value of the position (x, y), but can be easily determined because it is uniquely determined by the interpolation method.
For example, according to FIG. 4, it is assumed that the position indicated by P in FIG. 4A is obtained in step S2, and the value of the position indicated by P is obtained by the bilinear interpolation method. In step S4, P ₁ in FIG. , P ₂ , P ₃ , and P ₄ , the positions of four pixels surrounding the position P are determined. Then, the position of the obtained pixel is represented by (x ′, y ′)
And The values of x 'and y' are integer values.

Next, in step S5, one pixel position is selected from the pixel positions (x ', y') on the original image 1 obtained in step S4, and the first reference cutout mask 2 is selected.
The value of the pixel at the same position (x ', y') is read, and the value is set as the label value t at the pixel position (x ', y') on the original image 1.

Thus, the label value t indicates whether the pixel at the coordinates (x ', y') on the original image 1 is a pixel in the real area or a pixel in the background area. You can see that. In this case, if t = 0, the pixel located at these coordinates (x ′, y ′) of the original image 1 is in the real area, and t =
If it is 1, the pixel located at the coordinates (x ', y') of the original image 1 is in the background area.

Next, in step S6, the value of the label value s determined in step S3 is compared with the value of the label value t determined in step S5. If s = t = 0, the corresponding position on the original image 1 ( The flag of the pixel at the position (x ', y') on the original image 1 is cleared (step S7).
8).

Then, the processing of steps S5 to S8 is performed for all the pixel positions on the original image 1 obtained in step S4. This is the determination in step S9.

By the processing up to this point, the pixels of the original image 1 to be used for forming the pixels that are to be the real regions in the target image are selected, and only those pixels are flagged.

The following is a description, for example, with reference to FIG. FIG. 6A shows the second reference cutout mask 4, in which the line L is the boundary of the cutout line, and the value of the pixel on the line L and the pixel on the left side thereof is 0, and the value of the pixel on the right side of the line L is 1 And FIG. 6B shows the original image 1, and the line L ′ is the boundary of the cutout line in the first reference cutout mask 2. In the first reference cutout mask 2, the value of the pixel on the left side of the line L ′ is 0, The value of the pixel to the right of line L 'is
Let it be 1. Further, it is assumed that a bilinear interpolation method is used as the interpolation method.

Now, assuming that the pixel indicated by Q in FIG. 6A is selected in step S1 and that the position on the original image 1 obtained in step S2 is as indicated by P, in step S3 the pixel of Q The value of the position label value s is set to 0. This is because the pixel at the position of Q becomes a pixel of the substantial area in the target image.

In step S4, the positions of four pixels P _R1 , P _R2 , P _R3 and P _B1 surrounding the position P in FIG. 6B are obtained. Then, for these four pixel positions, step S
5 to S8 are performed. The label value t at the position of the three pixels P _R1 , P _R2 , and P _R3 is 0, and the label value t at the position of the remaining pixels P _B1 is 1. .

Therefore, three pixels P _R1 , P _R on the original image 1
_R2, flagged in step S7 for P _R3, so that the flag is dropped in step S8 to the pixel P _B1 on the original image 1.

The above is the processing up to step S9.
In this way, when the pixel position on the original image 1 for obtaining the value of the pixel in the substantial area on the target image is obtained,
Next, the process of step S10 in FIG. 3 is executed.

In step S10, the combination ratio is determined for the pixels for which the flag has been set in step S7. This composition ratio can be determined, for example, as follows.
Here, it is assumed that a normal interpolation method such as a bilinear interpolation method or a cubic convolution interpolation method is used.

Now, it is assumed that n pixels on the original image 1 are used to obtain the value of the pixel to be interpolated, the values of those pixels are set to P _i (i = 1, 2,..., N), When the conventional synthesis ratio corresponding to a pixel value of P _i defined by interpolation method used for the a _i, the value P _c of the pixel to be interpolated is generally _{P c = a 1 P 1 +} a 2 P 2 + ... + A _j P _j +... + A _n P _n (3) Here, a ₁ + a ₂ +... + A _j +... + A _n = 1 (4). In the case of the bilinear interpolation method, n = 4, and in the case of the cubic convolution interpolation method, n = 16.
Needless to say,

However, since the number of pixels on the original image 1 used for interpolation by the above-described processing is n or less, _ai cannot be used as a synthesis ratio as it is.

[0055] Thus, by introducing a function called l _i, shall be given a combination ratio b _i to determine in step S10 by the following equation.

B _i = a _i l _i / (a ₁ l ₁ +... + A _j l _j +... + A _n l _n ) (5) where l _i is the value of the pixel flagged in step S 7. It is 1 and 0 for pixels for which no flag was set in step S8.

Therefore, taking as an example the description with reference to FIG. 6 assuming that the bilinear interpolation method is used as the interpolation method, in order to obtain the value of the position of P in FIG. _{R1, P R2, P R3,} P B1 is selected, for these pixels, the synthetic ratio a ₁ by bilinear interpolation method, _{respectively, a 2, a 3, a} 4 becomes that is determined, when the above-mentioned Becomes l ₁ = l ₂ = l ₃ = 1, l ₄ = 0, so that the synthesis ratios b ₁ , b ₂ , b for the pixels P _R1 , P _R2 , P _R3 used in the interpolation determined in step S10.
₃ is as follows.

B ₁ = a ₁ / (a ₁ + a ₂ + a ₃ ) (6) b ₂ = a ₂ / (a ₁ + a ₂ + a ₃ ) (7) b ₃ = a ₃ / (a ₁ + a ₂₎ + A ₃ ) (8) Here, it is clear that the sum of the synthesis ratios (b ₁ + b ₂ + b ₃ ) becomes 1. The same applies to the case where the cubic convolution interpolation method is used.

When the combination ratio is obtained as described above, next, in step S11 in FIG. 3, the value of the position (x, y) on the original image 1 is interpolated using the combination ratio obtained in step S10. . To explain according to the above-described example, assuming that P _R1 , P _R2 , and P _R3 in FIG. 6B indicate the values of the respective pixels as they are, the value of the position indicated by P in FIG. 6B is b ₁ P _R1 + b ₂ P _R2 + b ₃ P _R3 (9)

Next, the value obtained in step S11 is written to the position (u, v) of the target image (step S12). Then, the above processing is performed for all pixel positions of the second reference cutout mask 4. The determination process for that is step S13.

When the creation of the target image has been completed as described above, the target image is output, and the target clipping mask 5 that is generated first may be output.

As described above, according to the present invention, the values of the pixels in the real area of the target image are interpolated only from the values of the pixels in the real area of the original image 1, so that the real The value of the pixel in the background area of the original image 1 is not mixed with the value of the pixel in the area, and deterioration of the image quality of the target image can be prevented. Note that the background area of the target image is not used finally, so any value may be used. For example, all
Just fill in with 0 or 1.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating an image processing method according to the present invention.

FIG. 2 is a flowchart illustrating an image processing method according to the present invention.

FIG. 3 is a flowchart illustrating an image processing method according to the present invention.

FIG. 4 is a diagram for explaining an interpolation method.

FIG. 5 is a diagram showing an example of an original image and a cutout mask corresponding to the original image.

FIG. 6 is a diagram for explaining a conventional problem.

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Takehiko Sato 1-1-1 Ichigaya Kagacho, Shinjuku-ku, Tokyo Inside Dai Nippon Printing Co., Ltd.

Claims (1)

[Claims] An image processing method for creating a target image having a predetermined resolution based on an original image and a clipping mask corresponding to the original image and having a predetermined resolution, comprising: Create a clipping mask having a target image resolution and a clipping mask having the resolution of the target image, refer to both of these clipping masks, determine the pixels in the original image that fall within the substantial area of the target image, and use only the determined pixels. An image processing method for obtaining a target image by performing pixel interpolation.