JP2003108994A - Image creation method and system - Google Patents

Abstract

(57) [Summary] [Problem] To create a target created image by using a reduced model image obtained by reducing a model image, a lump is not generated on the created image, Provided are an image creating method and system capable of creating an image reproducing only expressed features. SOLUTION: As a basic method, a method of arranging masks on both a model image and a created image and determining pixels of the created image based on the similarity thereof is used, and the resolutions of the model image and the created image are reduced. In the case where the above-described basic method is applied to the reduced model image and the reduced created image, the pixels of the above-described basic method and the lower (lower resolution) reduced created image are used when creating the upper (higher resolution) reduced created image. , And by repeating this process, an intended created image is obtained.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for using a model image having a pattern to create an image having a similar pattern and having a size larger than that of the model image.

[0002]

2. Description of the Related Art Conventionally, various pattern images have been used to decorate the surface of building materials and the like. Such a pattern image is obtained by correcting a photographed image and creating it or artificially creating it by a computer operation. Generally, since the surface of a building material product is large, the pattern image also needs to have a large size. However, since it is very difficult to create a large-sized pattern image, a small pattern image is created, and this pattern image is repeatedly arranged vertically and horizontally to create a large pattern image.

In recent years, a method of using a pattern image of a small size as a model image to create a large size image having a similar pattern is used, and an image of a desired size is created based on a pattern that one likes. Is becoming available. This method prepares a mask of a predetermined shape, arranges it on the image to be created, and on the model image, finds the similarity between the two based on the pixel value at the position where the mask exists, and calculates the similarity. Pixels on the model image when is high are pixels on the image to be created. With this method, an image having a pattern similar to the pattern expressed on the model image can be obtained.

When a desired image is created using the model image as described above, attempts have been made to extract low frequency components in order to reproduce the rough features on the model image.
To extract low-frequency components, create a reduced model image with reduced model image resolution, and create a reduced created image with a lower resolution than the target created image based on each pixel on this reduced model image. Then, a target created image is created by restoring this reduced created image to the size of the target created image. As a result, the rough features of the model image are obtained on the created image.

[0005]

However, in the above-described conventional image creating method, when creating a created image having a large resolution from the created reduced image, a pixel is created for each pixel of the created reduced image. Allocate a plurality of pixels on the changed model image. Therefore, a lump-like object may be visible on the created image, which is a problem.

In view of the above points, according to the present invention, even when a target created image is created using a reduced model image obtained by reducing the model image, a lump or the like appears on the created image. An object of the present invention is to provide an image creating method and system capable of creating an image in which only the features expressed on a model image are reproduced without causing the above.

[0007]

In order to solve the above problems, according to the present invention, as a method for creating an image having a new pattern using a model image having a desired pattern, a step of inputting a model image, The resolution of the input model image is reduced, the reduction rate is made different, and a plurality of reduced model images having different resolutions are created, respectively. A step of creating a reduced created image, and a step of creating an upper reduced created image that is one step higher than the lowest reduced created image by using an upper reduced model image that is one step higher than the lowest reduced created image. In the step of creating the reduced model image, the pixel value of each pixel of the reduced model image is set to a predetermined number of higher-order reduced model images having higher resolution than the reduced model image. While determining based on the prime value, for each pixel of the reduced model image, the position of each pixel of the upper reduced model image that is the basis for determining the pixel value of the pixel is recorded, In the step of creating the reduced created image, the attribute information of each pixel of the lowest reduced created image is determined by arranging a mask of a predetermined shape on the area of the lowest reduced created image, and setting the same shape as the mask. A mask is arranged on the lowest reduced model image, and while the position of the mask on the area of the lowest reduced created image is fixed, the mask on the lowest reduced model image is moved by one pixel, The similarity between the masked area on the area of the lowest reduced created image and the masked area on the lowest reduced model image is obtained, and the similarity is determined to be the highest on the lowest reduced model image. The attribute information of a pixel whose positional relationship with the mask is the same as the positional relationship between the mask and the pixel to be determined on the area of the lowest reduced created image, This is performed by assigning it as pixel attribute information, and the step of creating the upper reduced created image is
The upper reduced created image to be created is classified into a pixel restoration area and a mask utilization area, and the pixels of the pixel restoration area are
Based on the positional relationship between each pixel of the recorded reduced model image and each pixel of the upper reduced model image that is the basis for determining the pixel value of the pixel, the attribute information thereof is determined, and the pixel of the mask use area is determined. Is performed by determining the attribute information by the same processing as the step of creating the lowermost reduced created image, and by repeating the step of creating the upper created image, the same as the model image is obtained. It is characterized in that a resolution creation image is created.

According to the present invention, the basic method is a method of arranging masks on both the model image and the created image and determining the pixels of the created image based on the degree of similarity between them. When applying the above basic method to the reduced model image and the reduced created image whose resolutions are respectively reduced, both the above basic method and the restoration from the pixels of the lower reduced created image are created when the upper reduced created image is created. Since it is used, it is possible to create an image in which only the features expressed on the model image are reproduced without causing lumps on the created image.

[0009]

DETAILED DESCRIPTION OF THE INVENTION Embodiments of the present invention will be described in detail below with reference to the drawings. FIG. 1 is a flowchart showing an outline of the image creating method according to the present invention. First, an image having a pattern as a model (hereinafter referred to as a model image) is input (step S1).

Next, parameters are set (step S2). As the parameter, the size (number of pixels) of the created image, which is the image to be created, is set.

Then, based on the input model image, a plurality of reduced model images with reduced resolution are created at a plurality of preset resolution reduction rates (step S).
3). The reduced model image is created by giving an average value of horizontal r × vertical r pixels on the model image as a pixel value of one pixel of the reduced model image, where r is the resolution reduction rate. At this time, each pixel of the reduced model image has the given average value as its pixel value, and which pixel r × r pixels of the model image the pixel is obtained as the pixel specifying information. Record. This is used later when restoring an image with higher resolution. In the reduced model image, the pixel value and the pixel specifying information will be collectively referred to as pixel attribute information. The number of kinds of reduced model images to be created and the resolution reduction rate of each reduced model image with respect to the original model image are set in advance. Among the reduced model images, the reduced model image having a higher resolution will be referred to as an upper reduced model image. It should be noted that in the present embodiment, with respect to the model image M, the three-stage reduced model images M1 and M2 are
-M3 is created, and the reduced model image M1 is r =
2, the reduced model image M2 was created based on the reduced model image M1 with r = 2, and the reduced model image M3 was created based on the reduced model image M2 with r = 2. I shall. In the present invention, the reduced created images G1, G2, and G3 are created in the process of obtaining the intended created image G, and these resolutions are respectively reduced model images M1, M2, and M.
It corresponds to 3.

Then, the lowest reduction model image corresponding to the lowest resolution model image having the lowest resolution is created. In this embodiment, the resolution reduction rate r = 8 as compared with the created image G.
The reduced created image G3 is created. When creating the reduced created image G3, first, an image area of a required size is secured in the image memory (step S4). Here, if the size of the created image is set to horizontal sx × vertical sy, and the resolution reduction ratio r is used, the horizontal direction r of the reduced created image area
The x and the vertical direction ry are calculated by the following [Formula 1].

[Formula 1] rx = sx / r ry = sy / r

Here, FIG. 2A shows the lowermost reduced image area prepared in the image memory as described above. Pixel values are given to the pixels located in the rightmost two columns and the lowermost two rows in this reduced created image area (step S
5). Specifically, in FIG. 2B, a pixel value determined by a random number or the like is given to the pixel located in the pixel addition portion, which is a shaded portion. The pixel value given to this pixel-added portion will be used later, assuming that the pixel value is in the position shown by hatching in FIG. Originally, Figure 2
Pixels should be generated outside the reduced created image area as shown in (c), but by generating pixels inside the reduced created image area as shown in FIG. And the efficiency of arithmetic processing is improved. That is,
This is because if pixels are generated outside the reduced created image area, an area of the image memory must be allocated in addition to the reduced created image area. Regarding the pixel-added portion, the pixel value at that position will be overwritten. Further, the reason why the pixel-added portion is set to two pixels is that it is necessary for the calculation of the degree of similarity using a mask having a shape described later. The state of the image memory is actually in a state as shown in FIG. 2B, but the program installed in the computer is assumed to be in a state as shown in FIG. To do. Therefore, the following processing will be described based on the state shown in FIG.

When the pixel value is given to the pixel in the pixel-added portion of the lowermost reduced created image area, the value of each pixel in the lowest reduced created image area is then determined (step S).
6). There are various conceivable orders for determining pixels in the lowest reduced image area, but in the present embodiment,
It will be decided from the pixel at the upper left corner shown by hatching in FIG. To determine the pixel at the upper left corner, a mask corresponding to 12 pixels as shown in the upper left corner is used. When such a mask is arranged at a position for determining the upper left pixel of the reduced created image area, two pixels are projected upward and leftward from the reduced created image area by two pixels as shown in FIG. 3A. become. It is for this reason that the attribute information for two pixels has been previously added to the surroundings as virtual pixels.

In order to determine the upper left pixel shown by hatching in FIG. 3 (a), specifically, a mask having the same shape as this mask is placed at positions corresponding to all the pixels of the reduced model image. And find the similarity of the masked area. Then, the pixel at the position corresponding to the mask on the reduced model image determined to have the highest degree of similarity is determined as the upper left pixel shown in FIG. 3A, and the attribute information is recorded. As a mask on the reduced model image, first, as shown in FIG.
It is arranged at the position shown in (b), and the degree of similarity between the masked area and the masked area shown in FIG. 3 (a) is obtained. Next, as shown in FIG. 3C, the mask for one pixel is shifted to the right, and the degree of similarity between the masked area and the masked area shown in FIG. 3A is obtained. In the same way, we perform processing to find the similarity for each pixel position,
When the mask reaches the right edge on the reduced model image,
Move the mask down one line to the left edge as shown in (d),
The degree of similarity between the masked area and the masked area shown in FIG. 3A is obtained. In this way, the degree of similarity between the masked area and the masked area shown in FIG. 3A is obtained for the positions corresponding to all the pixels on the reduced model image. For example, when it is determined that the similarity to the masked area at the position shown in FIG. 3E is the highest, the pixels shaded with diagonal lines in FIG.
The pixel is assigned as the upper left pixel shown in FIG. That is, the attribute information of the pixel shown by hatching in FIG. 3E is copied to the pixel at the upper left corner shown in FIG.

Here, in the present invention, a function for calculating the similarity between the masked area in the reduced created image area and the masked area in the reduced model image will be described. As this function, any function may be used as long as it can determine the similarity between the pixel existing at the mask position on the reduced created image area and the pixel existing at the mask position on the reduced model image. In the embodiment, the one shown in the following [Formula 2] is used.

[Formula 2] V = Σi (Rs ² + Gs ² + Bs ² ) where i = 1 to 12 (number of pixels)

In the above [Formula 2], V is an evaluation value for evaluating the degree of similarity, and Rs, Gs, and Bs are the pixels in the reduced image area when RGB color images are used. It is the difference between the pixel values of the respective color components of the pixels on the reduced model image. Therefore, the smaller the value of the evaluation value V, the closer the value is, and it is determined that the degree of similarity is high. Although i = 1 to 12 in the above example,
This is the case where a standard mask as shown in FIG. 3 is used, and it depends on the mask shape.

When the pixel at the upper left corner as shown in FIG. 3A is determined, as shown in FIG. 4A, the mask on the reduced image area is shifted to the right by one pixel and the same process is performed. To do. At this time, the pixels used to obtain the similarity of the masked area are 11 pixels that are virtual pixels and the pixel at the upper left end of the reduced created image area whose attribute information has already been determined (in FIG. 4A). The value of () is used.

Similarly, the pixels in the first row of the reduced image area are determined. FIG. 4B shows the position of the mask when determining the rightmost pixel in the first row. Figure 4
Also in (b), the values of the pixels indicated by the circles have already been determined. When the rightmost pixel in the first row is determined, the pixels in the second row are determined from the right edge instead of the left edge. At this time, as shown in FIG. 4C, the mask used has a shape which is left-right inverted from that used for determining the pixels in the first row. In addition, also in FIG.
The mark indicates a pixel whose value has already been determined. Figure 4 (c)
When the mask as shown in FIG. 4 is used, the mask having the same shape as that shown in FIG. 4C is naturally used as the mask on the reduced model image for obtaining the similarity. Using such a mask, the pixels on the second row are sequentially determined from the right side. For the third line of the reduced image area, see FIG.
Using the mask as shown in (a), it is decided from the leftmost pixel. That is, the pixels are determined in the left-right opposite directions by using the mask that is horizontally inverted for each row. Although it is possible to determine all the rows from one direction at all times, in such a case, there may occur a phenomenon in which an unintended flow like a diagonal flow appears on the created image. . In order to eliminate such a problem, it is desirable to determine pixels in a zigzag manner as in the example of FIG.

Although it is possible to create all the rows of the reduced created image area by using the masks shown in FIGS. 3 and 4, in the present invention, the lower two rows of the reduced created image area are created. For the minute, seamless processing is performed by using masks of different shapes. As mentioned above, the images created in the present invention are often used on the surface of building material products. Printed materials used for building materials are often printed using rotary presses, and for this reason, the original image needs to be seamless above and below. If the upper and lower sides are not seamless, printed matter with a seam is produced for each circumferential unit of the plate cylinder used for rotary printing, which is not preferable. In order to solve such a problem at the same time, in the present embodiment, the following processing is performed for the lower two lines of the reduced created image area.

For the first row to the third row from the bottom of the reduced created image area, the pixels are determined using a mask of 12 pixels as shown in FIGS. 3 and 4, but the reduced created For the two lines from the bottom of the image area, masks having different shapes as shown in FIG. 5 are used. In FIG. 5, the shaded areas are the pixels to be determined, and the circled ones are the already determined pixels. A mask as shown in FIG. 5A is used to determine the pixels in the second row from the bottom. Of the 17 pixels that make up the mask, the bottom 5
Individual pixels are shown by dotted lines because no pixel actually exists at this position. When the same created image is joined vertically, the uppermost part of the created image is located below the lowermost part of the created image. Therefore, instead of the part indicated by the dotted line, Uses one pixel. That is, the values of the pixels in the uppermost row are reflected in the determination of the pixels in the second row from the bottom, and the seamlessness is achieved.

To determine the bottom row of pixels, refer to FIG.
As shown in (b), the pixel at the left end, which is hatched with diagonal lines, is determined using a mask composed of 22 pixels. Of the 22 pixels that make up the mask, the lower 2
As for 10 rows, pixels are not actually present at this position, so that they are indicated by dotted lines. As described above, when the images are joined vertically, the uppermost part of the created image is located below the created image. Therefore, instead of the portion indicated by the dotted line, the first line of the reduced created image is displayed. 1 shown in the second line
Use 0 pixels. As a result, the values of the pixels in the first and second rows from the top are reflected in the pixels in the first row from the bottom, and the seamlessness is achieved.

As a matter of course, the masks on the model image for obtaining the similarity with the masks shown in FIGS. 5A and 5B have the same shape. In addition, in FIGS. 5A and 5B, the case of determining from the pixel on the left side has been described, but in the case of determining from the pixel on the right side, FIG.
A mask having a symmetrical shape with the masks shown in (a) and (b) is used. In addition, the determination of the similarity is performed by the above [Formula 2].
It can be performed by calculating an evaluation value such as. However, in [Equation 2], i = 1 to 17 in the mask shown in FIG. 5A, and i = 1 to 22 in the mask shown in FIG. 5B.

When the reduced created image is obtained as described above, it is judged whether or not this is the intended created image (step S7). Specifically, it is determined whether or not the set number of steps (the same as the number of steps of creating the reduced model image) have been created. If the created image has not been obtained, the processes of steps S4 to S6 are repeated.

When the reduced created image G3 is obtained by deciding each pixel in the reduced created image area, the created image has not been obtained yet. Therefore, the steps S4 to S6 are repeated to reduce the created image G3. A reduced created image G2 having a higher resolution by one step is created. The reduced created image G2 is created using the reduced created image G3 and the reduced model image M3 used to create the reduced created image G3. Specifically, as shown in FIG. 6A, the reduced created image G2 is divided into a pixel restoration region and a mask use region, and pixel attribute information is given by different methods.

First, in the pixel restoration area, pixels are restored using the pixels of the reduced created image G3. For example, in FIG.
As shown in, when one pixel on the reduced created image G3 is restored to 4 pixels on the reduced created image G2, the reduced created image G3
For the pixels above 3, as described in step S6 above,
Which pixel on the reduced model image M3 is assigned is recorded as pixel specifying information, and
For the pixel on the reduced model image M3, the above step S
As described in Section 3, since which 4 pixels on the reduced model image M2 were created based on the attribute information is recorded, the 4 pixels on the reduced model image M2 correspond to the 4 pixels on the reduced created image G2. By assigning
The attribute information of the four shaded pixels on the reduced created image G2 shown in FIG. 6A is determined. It should be noted that in the reduced created image G3 shown in FIG. 6B, the pixels in the area marked “used” are used for pixel restoration, and the pixels in the area marked “not used” are not used. become.

After the pixels of the pixel restoration area are determined, the pixels of the mask utilization area are determined. Specifically, the masks shown in FIGS. 3 to 5 are arranged in both the reduced model image M2 and the mask use area, and the pixels on the reduced model image M2 are allocated by the same processing as in step S6. Go.

In creating the reduced created image G2,
An image area is secured in the image memory (step S4), and virtual pixels are generated (step S5). However, in step S5, of the portion that the mask protrudes when determining the pixels in the mask use area, the pixel provided in the pixel restoration area can be used in the portion that is adjacent to the pixel restoration area. It suffices to generate virtual pixels only for the non-existing portions. After that, pixels are allocated to the pixel restoration area and the mask area as described above (step S6).

The reduced created image G2 is obtained by the above processing.
Is obtained, next, using the reduced created image G2 and the reduced model image M1, 1 is obtained from the reduced created image G2.
A reduced created image G1 having a high step resolution is created. The creation of the reduced created image G1 is similar to the creation of the reduced created image G2.
This is performed by dividing the pixel restoration area and the mask usage area and giving the attribute information of the pixels by different methods. The positional relationship between the pixel restoration area and the mask usage area is shown in FIG.
The result is as shown in (a). That is, when the reduced creation image G2 is created, the area that was the pixel restoration area is directly used as the pixel restoration area, and the area that was the mask utilization area is further divided into the pixel restoration area and the mask area. Corresponding to this, the area of pixels used and the area of pixels not used in the reduced created image G2 are as shown in FIG. 7B.

Regarding the pixels in the pixel restoration area of the reduced created image G1, the pixels on the reduced model image M2 are specified based on the attribute information of the pixels of the reduced created image G2, and based on the attribute information of the pixels of the reduced model image M2. And reduced model image M1
This is performed by specifying the upper 4 pixels and assigning the specified 4 pixels. Regarding the mask use area of the reduced created image G1, the masks as shown in FIGS. 3 to 5 are arranged in both the reduced model image M1 and the mask use area, and the reduced model image M1 is processed by the same processing as in step S6. This is done by allocating the upper pixels.

After the reduced created image G1 is obtained, the reduced created image G1 and the model image M are used to
A target created image G is created. The creation of the created image G is performed by dividing it into a pixel restoration area and a mask use area and giving attribute information of pixels by different methods as in the creation of the reduced created images G2 and G1. The positional relationship between the mask utilization area and the mask utilization area is as shown in FIG. That is, when the reduced creation image G1 is created, the area that was the pixel restoration area is directly used as the pixel restoration area, and the area that was the mask utilization area is further added.
It is divided into a pixel restoration area and a mask area. Corresponding to this, the area of pixels used and the area of pixels not used in the reduced created image G1 are as shown in FIG. 8B.

As for the pixels in the pixel restoration area of the created image G, the pixels on the reduced model image M1 are specified based on the attribute information of the pixels of the reduced created image G1, and the reduced model image M1 is selected.
This is performed by specifying 4 pixels on the model image M based on the attribute information of the pixels and assigning the specified 4 pixels. Regarding the mask use area of the created image G, the masks as shown in FIGS. 3 to 5 are arranged in both the model image M and the mask use area, and the pixels on the model image M are processed by the same processing as in step S6. It is done by assigning. The created image G is obtained as in steps S1 to S7. In the obtained created image G, 8 × 8 64 pixels are reduced and created in the portion that is the pixel restoration region from the beginning on the lower left and the upper right. Since it corresponds to one pixel of the image G3 and is arranged in units of 64 pixels, a streak-like object due to the unit of 64 pixels may be seen. However, as described above, since the entire created image G is not the pixel restoration area based on the reduced created image G3, no streaks occur throughout the entire image. When the created image G is created in the above example, depending on which reduced created image is used to start the restoration, there is a portion in which a lump of 16 pixel units and a lump of 4 pixel units are arranged. The portion that was the mask utilization area is not arranged in a lump. That is, by creating a created image by the method according to the present invention, it is possible to prevent repetition of a lump of the same size over the entire image, and it becomes difficult for streaks to occur in the entire image.

The basic processing of the image creating method according to the present invention has been described above. However, in the present invention, many contrivances are made in order to faithfully reproduce the characteristics of the model image. In the following, an effective application example that makes such ingenuity will be described.

(Application Example of Pixel Restoration) In the process of step S6, when the pixels of the pixel restoration area of the upper reduced created image are obtained from the lower reduced created image, the lower reduced model from the lower reduced created image is obtained. Through the image, the corresponding pixel on the upper reduced model image is obtained, and it is assigned as it is to the upper reduced created image. At this time,
When the resolution reduction rate is small (r = 2 to 3), there is not much problem, but when the resolution reduction rate is large, there is a lump like each pixel number corresponding to the resolution reduction rate. A visible phenomenon will occur. For example, if the resolution reduction rate r = 40, blocks consisting of 40 × 40 pixels are arranged on the upper reduced created image. At this time, in the portion corresponding to the joint of 40 × 40 pixels, the continuity is not taken into consideration in the arrangement of the adjacent pixels, so that the joint becomes conspicuous. As a result, the entire high-order reduced created image looks like an aggregate of blocks of 40 × 40 pixels.

Therefore, when the value of the resolution reduction rate r is large, the above problem is solved by performing the following process in the process of step S6. When the resolution reduction ratio r = 40, originally, 40 × 40 pixels should be assigned to the higher reduced created image, but 46 × 46 pixels in the reduced model image are assigned to the reduced created image. In other words, the upper, lower, left, and right 3 pixels are extracted and allocated in excess. The extra extracted pixels are combined with other pixels assigned on the reduced created image. The synthesis rate at this time is set to a value that becomes smaller as it gets closer to the end of the block of 46 × 46 pixels, and the synthesis rate is changed using a random number. The synthesis rate α is calculated by the following [Formula 3].

[Equation 3] α = DI (X) + RAND

In the above [Formula 3], DI (X) is
Is a function that increases with the distance from the edge of the block,
It takes a value from 0 to 1. Further, RAND is -0.5 to 0.
It is a random number that takes a value of 5. After the calculation, α is corrected to have a value of 0 to 1. In addition, here, 3
The extra pixels are extracted and assigned, but the number of extra pixels is also changed according to the value of the resolution reduction rate r.

(Application Example of Pixel Determining Using Mask) If the pixel determining method using a mask as described in step S7 is used as it is, the pattern of the entire reduced model image becomes a reduced created image (pixel restoration region). However, the phenomenon that some patterns of the reduced model image repeatedly appear on the reduced created image may occur. This is because the number of times the pixels on the reduced model image to be used are extracted when determining the pixels on the reduced created image is biased depending on the pixels. That is, a certain pixel is selected many times, but a certain pixel is not selected once, so that the characteristics of the entire reduced model image are not reflected on the reduced created image.

Therefore, a device for solving such a problem will be described. For this purpose, it is sufficient to count how many times each pixel of the reduced model image has been selected, and limit the number of selection times so that it does not exceed a predetermined number of times. In the present invention, the following [Formula 4] is used for this limitation.

[Formula 4] IC _{(x, y)} <K × Count / (mx × my) + C

In the above [Formula 4], IC _{(x, y)} is
This is a variable indicating how many times the pixel at the coordinates (x, y) on the reduced model image has been selected so far, and is incremented by 1 each time it is determined as a pixel in the reduced created image. K is a constant, and an essential requirement is that it be a value greater than 1. If possible, K is preferably a numerical value less than 2. Count is a variable indicating the number of pixels that have been determined so far in the reduced created image area when the pixel is determined. mx × my is a constant indicating the total number of pixels of the reduced model image. C is a constant set from the outside. As described above, by comparing the masked area on the reduced created image with the masked area on the reduced model image, the pixel on the reduced model image when the similarity is highest is determined. IC _{(x, y)} is the number of times selected so far
The above [Formula 4] is used as Then, only when this [Equation 4] is satisfied, the pixel is selected as a pixel on the reduced created image. When selected, only 1 is added to IC _{(x, y )} indicating the number of selections of this pixel.

In the above [Formula 4], Count / on the right side
Since (mx × my) indicates how many times the number of pixels determined in the reduced created image area is the total number of pixels of the reduced model image, the variable IC _{(x, y)} is By setting the number to be smaller, the number of times each pixel is selected on the reduced model image is averaged. However, with this alone, Count is changed to (mx × m
As the remainder when divided by y) approaches 0, the number of selectable pixels will decrease significantly. For example, when the remainder is 1, there is only one room for selection. Therefore, the pixel selected at this time has no correlation with the neighborhood,
It has an unnatural texture. The constant K is used to avoid this. In this way, even if the above-mentioned surplus is reduced, the number of selectable pixels does not decrease, and an unnatural texture does not appear. As a result, it is possible to select pixels from the reduced model image almost on average. Further, C on the right side of the above [Formula 4] is a constant for setting an allowable range, and an integer of 0 or more is set. As the value of the constant C increases, the variation in the number of times each pixel is selected on the model image increases, and C =
When set to 0, each pixel is selected almost uniformly.

Further, in the present invention, in order to reproduce the rough characteristics of the model image and to reduce the amount of calculation, the reduced image having the reduced resolution is produced and then the produced image is produced. However, this causes a problem as shown in FIG. FIG. 9A is an enlarged view of a part (three pixels) of the lower reduced image, and FIG.
[Fig. 4] is a partially enlarged view of a high-order reduced created image. Note that FIG. 9 shows a case where the resolution reduction rate r = 3. r
= 3, one pixel on the lower reduced created image is restored to 9 (3 × 3) pixels on the upper reduced created image. In this case, the array of 9 pixels is It is restored in the same state as on the reduced model image. Therefore, FIG.
As shown in (a), when the same pixel on the lower reduced created image is selected three times in succession, when restored, the pattern repeats in a narrow range on the upper reduced created image. This is because a specific pixel is repeated, as can be seen from the pixel shaded in FIG. 9B.

Therefore, in the present invention, the above-mentioned problem does not occur by not selecting the same pixel as a neighboring pixel. FIG. 10 is a partially enlarged view of the lower reduced image. In particular, FIG.
When determining pixels from left to right, FIG.
Indicates a case where pixels are determined from right to left. When determining pixels from the left to the right, FIG.
When determining the pixels indicated by hatching in (a),
Make sure that it is not the same as the four pixels marked with a circle.
In determining pixels indicated by hatching, the similarity between the masked area on the reduced created image and the masked area on the reduced model image is determined in the same manner as described in the above example. The pixel having the highest degree of similarity is selected, but here, the pixel on the selected reduced model image is not indicated as a pixel shaded with diagonal lines but is indicated by a circle. It checks to see if it is not the same as the four pixels. As a result, the same pixel as a neighboring pixel is not selected, and the above problem is solved. In addition, in the case of determining pixels from right to left, FIG.
The positional relationship as shown in (b), that is, FIG. 10 (a)
A process for preventing overlapping is performed for pixels that are symmetrical with respect to the case.

Next, the above two measures are taken in step S of FIG.
A specific application in the determination of each pixel in 6 will be described in more detail with reference to the flowchart shown in FIG. First, for a certain pixel on the reduced model image, the number of selections of that pixel IC _{(x, y)} (where x =
1-mx, y = 1-my). Then, it is determined whether or not the number of selections IC _{(x, y)} satisfies the above [Formula 4] (step S11). When it is determined that the condition of [Formula 4] is not satisfied, the pixel is not selected and the process proceeds to the determination of the next pixel (step S1).
2). When it is determined that the condition of [Formula 4] is satisfied, the evaluation value is calculated using the mask corresponding to the pixel (step S13). For the calculation of the evaluation value, the one shown in the above [Formula 2] is used. The calculated evaluation value is the coordinate value (x, y) of the pixel on the reduced model image.
A predetermined number is recorded together with {however, x = 1 to mx, y = 1 to my}. Specifically, the number of pixels in the vicinity for which duplicate extraction is prohibited + 1 is recorded from the smallest evaluation value (those with a high similarity). For example, as shown in FIG. 10, when four adjacent pixels as adjacent pixels are to be subjected to duplicate extraction prohibition, five evaluation values are recorded from the smallest evaluation value.

By performing the processes of steps S11 to S13 on all pixels (mx × my) on the reduced model image (steps S14 and S15), mx × my pixels on the reduced model image are obtained. Of these, five pixels having a small evaluation value are obtained. Next, among the five pixels, the pixels having the smallest evaluation value are compared in order from the smallest pixel to see whether or not the pixel is the same as the pixel already selected in the vicinity on the reduced created image (step S16). For example, consider a case where the coordinate value of the reduced model image corresponding to the pixel (10, 10) on the reduced created image is being specified. In this case, the pixels pm1, pm2, pm3.
It is assumed that pm4 and pm5 are recorded. Pixel pg (10,10) to be decided on the reduced created image
Is a row determined from the left, the pixel pg1 (9,9) at the upper left of the pixel pg (10,10) to be determined.
9), the pixel pg2 (10, 9) immediately above, and the pixel pg at the upper right
3 (11, 9) and the left pixel pg4 (9, 10) are subject to duplicate extraction prohibition. In this case, the pixel pg to be determined,
The relationship with each pixel pg1 to pg4 that is the object of prohibition of duplicate extraction corresponds to the relationship between the pixel shaded in FIG. 10A and the four pixels marked with a circle.

In such a situation, the pixel pm1 having the smallest evaluation value and being the first candidate to be assigned as the pixel pg on the reduced created image is already the pixel pg1 on the reduced created image.
~ Check if it is assigned to pixel pg4. When the pixel pm1 is already assigned to the pixels pg1 to pg4, the pixel pm1 is excluded from the candidates of pixels to be selected as the pixel pg. Similarly, the pixels pm2, pm3, ...
If there is a pixel that is not assigned to any of g4, the pixel is determined as a pixel to be assigned to the pixel S on the created image. Since five pixels pm1 to pm5 are recorded as the candidates of the pixels to be assigned to the pixel S, even if the candidate pixels pm1 to pm4 are already assigned to all the four pixels pg1 to pg4, The pixel pm5 can be assigned to the pixel pg. This is the reason why one more candidate pixel is recorded than the overlap extraction prohibition target pixel. When the pixel is determined, 1 is added to the number of selections IC _{(x, y)} of the pixel on the selected reduced model image, and the number of pixels determined so far on the reduced created image is added. Also for the variable Count that is shown, 1
Is added. The pixels are determined in step S6 in the flowchart of FIG. 1 as in steps S11 to S16.

(System Configuration) Next, a system configuration for executing the above-described image creating method will be described. Figure 1
FIG. 2 is a configuration diagram showing an embodiment of an image creating system according to the present invention. In FIG. 12, the image input means 1 is
It is for executing the process of step S1 in FIG. 1, and has a function of inputting a model image as digital data. As the model image input means 1, when the model image is already prepared as digital data, it can be realized by a reading device capable of reading an electronic recording medium recording the digital data, and the model image is prepared as a paper medium. If so, it can be realized by an optical reading device such as a scanner.

The parameter setting means 2 executes the processing of step S2 of FIG. 1, specifically, the setting of the size of the created image, the number of steps of resolution reduction, and the processing of setting the resolution reduction rate r of each step. It is for the purpose and can be realized with a mouse, a keyboard, or the like. The parameter setting means 2 is also used to set various necessary parameters such as setting random numbers.

The mask shape storage means 3 stores the reduced created image (or created image) and the reduced model image (or model image) for pixel determination as described in step S6 of FIG. 1 and step S13 of FIG. ) Is to store the shape of the mask to be placed on both sides.

The arithmetic and control unit 4 plays a central role in the image forming system according to the present invention, and has a pixel determining unit 4a and a resolution converting unit 4b. The pixel determining unit 4a has a function of executing the processes of steps S4 to S6 of FIG. 1, and secures an area for creating a reduced created image on the image memory 5 to determine pixels. The resolution conversion unit 4b has a function of executing the process of step S3 in FIG. 1, and creates a lower-order reduced model image from the model image according to the set resolution reduction rate. The arithmetic and control unit 4 is realized by a CPU and a memory of a computer, and the pixel determining means 4a and the resolution converting means 4b which are specific functions are realized by a dedicated software program installed in the computer.

The image memory 5 is a storage area for creating a reduced created image, an area for the image size set by the parameter setting means 2 is secured, and the attribute information of the pixel determined by the pixel determining means 4a. Will be recorded.

Although the preferred embodiment of the present invention has been described above, the present invention is not limited to the above embodiment and various modifications can be made. For example, in the above-described embodiment, three reduced model images M1 to M3 are created and the created images are obtained through the three reduced created images G1 to G3, but at least two of these are created. Any number is possible as long as it is above each. As the number of different reduced images is increased, the amount of calculation increases, but the size of the cluster at the time of restoration varies, so that a step such as a step is less noticeable on the created image.

[0056]

As described above, according to the present invention,
A basic method is to place a mask on both the model image and the created image and determine the pixels of the created image based on the similarity between them, and a reduced model image in which the resolutions of the model image and the created image are reduced. When applying the above-mentioned basic method to the reduced created image, both the basic method and the restoration from the pixel of the lower reduced created image are used when creating the upper reduced created image. This has the effect of making it possible to create an image that reproduces only the features expressed on the model image without producing lumps on the top.

[Brief description of drawings]

FIG. 1 is a flowchart of an image creating method according to the present invention.

FIG. 2 is a diagram showing a reduced created image area secured on an image memory.

FIG. 3 is a diagram showing a relationship between a reduced created image area and a reduced model image, and a mask arranged thereon.

FIG. 4 is a diagram showing a relationship between a reduced created image area and a mask when the order of pixel determination is changed left and right alternately.

FIG. 5 is a diagram showing a relationship between a reduced created image area and a mask when performing a seamless process of a lower portion of the reduced created image area.

FIG. 6 is a reduced created image G2 and a (bottom) reduced created image G.
It is a figure which shows the relationship of 3.

FIG. 7 is a diagram showing a relationship between a reduced created image G1 and a reduced created image G2.

FIG. 8 is a diagram showing a relationship between a created image G and a reduced created image G1.

FIG. 9 is a diagram showing a pixel relationship when the same pixel is continuously selected on the reduced created image.

FIG. 10 is a diagram showing a positional relationship between a pixel to be determined and a pixel whose duplication is prohibited.

FIG. 11 is a flowchart showing an application example in determining each pixel in the reduced created image area.

FIG. 12 is a configuration diagram showing an embodiment of an image creating system according to the present invention.

[Explanation of symbols]

1. Image input means 2. Parameter setting means 3 ... Mask shape storage means 4 ... Arithmetic control device 4a ... Pixel determining means 4b ... Resolution conversion means 5 ... Image memory

Claims (3)

[Claims] 1. A method of creating an image having a new pattern by using a model image having a desired pattern, comprising the steps of inputting a model image and reducing the resolution of the input model image. A step of creating a plurality of reduced model images having different resolutions with different reduction rates, a step of creating a lowest reduced created image using the lowest reduced model image having the lowest resolution, Using an upper reduced model image that is one step higher than the lowest reduced created image to create an upper reduced created image that is one step higher than the lowest reduced created image. In the step of creating, the pixel value of each pixel of the reduced model image is determined based on a predetermined number of pixel values of the higher-order reduced model image having higher resolution than the reduced model image. For each pixel of the reduced model image, the position of each pixel of the upper reduced model image that is the basis for determining the pixel value of the pixel is recorded, and the step of creating the lowest reduced created image is Determining the attribute information of each pixel of the lowest reduced created image, placing a mask of a predetermined shape on the area of the lowest reduced created image,
A mask having the same shape as the mask is arranged on the lowest reduction model image, and the mask on the lowest reduction model image is fixed by one pixel while the position of the mask on the area of the lowest reduction creation image is fixed. While moving each of them, the similarity between the masked area on the area of the lowest reduced image and the masked area on the lowest reduced model image is obtained, and the lowest rank determined to have the highest similarity is calculated. The attribute information of a pixel whose positional relationship with the mask on the reduced model image is the same as the positional relationship between the mask and the pixel to be determined on the area of the lowest reduction created image is set to the lowest reduction. This is performed by assigning it as attribute information of pixels in the created image area. In the step of creating the upper reduced created image, the upper reduced created image to be created is set in the pixel restoration area and the mask region. For the pixels of the pixel restoration region, the pixels of the pixel restoration region are classified based on the positional relationship between each pixel of the recorded reduced model image and each pixel of the upper reduced model image that is the basis for determining the pixel value of the pixel. , The attribute information is determined, and for the pixels of the mask use area, the attribute information is determined by the same processing as the step of creating the lowest reduced created image. An image creating method characterized in that a created image having the same resolution as the model image is created by repeatedly executing the step of creating. 2. An image input means for inputting a model image having a desired pattern, a setting means for setting a size of a created image which is an image to be created, and a resolution of the model image according to a preset resolution. A resolution conversion unit that creates a plurality of reduced model images and a pixel determination unit that prepares a created image region for allocating pixels of the created image and determines each pixel on the created image region are provided. Then, the determination of each pixel by the pixel determining means defines a pixel restoration region and a mask use region in the created image region, and each pixel of the pixel restored region is defined in each of the created lower reduced model image. The attribute information is determined based on the positional relationship between the pixel and each pixel of the upper reduced model image. A mask having a predetermined shape is placed on the mask use area, a mask having the same shape as the mask is placed on the model image, and the position of the mask on the mask use area is fixed, and the model is changed. While moving the mask on the image pixel by pixel, the similarity between the masked area on the mask utilization area and the masked area on the model image is calculated, and the model determined to have the highest similarity. The attribute information of a pixel having a positional relationship with the mask on the image that is the same as the positional relationship between the mask on the mask use area and the pixel to be determined is attribute information of the pixel on the mask use area. An image creating system characterized by being assigned as 3. A step of creating a plurality of reduced model images having different resolutions by reducing the resolution of the model image and making the reduction rate different, and using the lowest reduced model image having the lowest resolution. And creating the lowest reduced created image, 1 from the lowest reduced created image
A step of creating an upper reduced created image that is one step higher than the lowest reduced created image using the upper reduced model image that is the upper order, and the step of creating the upper created image is repeated. By doing so, a program for creating a created image with the same resolution as the model image.