JPH06348834A - Image processing device - Google Patents

Abstract

PURPOSE:To perform processing at high speed without generating the deterioration of picture quality due to moire, etc., in an image processing device capable of performing the magnification-reduction/resolution conversion processing of a binary image. CONSTITUTION:This device is provided with a magnification conversion means 15 which converts a magnification to N/M (where, H, M: natural number) times that is a value to approximate the magnification to magnify/reduce the binary image, an integer ratio magnification/reduction conversion value storage means 16 which stores a table to output the binary values or multivalues of N picture elements that is the magnification/reduction images in accordance with the input of M picture elements of the binary image, an integer ratio magnifying/reducing means 11 which forms the image of binary or multivalues by multiplying by N/M a targeted binary image by using the table stored in the integer ratio magnification/reduction conversion value storage means 16, a real number ratio magnifying/reducing means 12 which magnifies/reduces the binary or multivalues formed by the integer ratio magnifying/ reducing means 11 by magnification XN/M as preserving density by interpolation, and a binarizing means 13 which binarizes the image magnified/reduced by the real number ratio magnifying/reducing means 12.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus for an image, and more particularly to an image processing apparatus capable of enlarging / reducing an image.

[0002]

2. Description of the Related Art Generally, an image processing apparatus or a DTP (Desk Top Publishing) capable of inputting an image
In a system or the like, it is necessary to perform scaling processing on the image.
If the resolution at the time of image input and the resolution of the display / output device are different, it is necessary to perform resolution conversion processing in order to make the sizes of the input document and the output document match. The processing content of this resolution conversion processing is the same as that of the scaling processing.

Conventionally, a thinning-out method has been widely used in such image enlargement / reduction processing. However, although the thinning method has a high processing speed, it has a drawback that the image quality after scaling is poor. In particular, when a natural image expressed in black and white binary by pseudo halftone processing or the like is enlarged or reduced by the thinning method, moire may occur and the image quality may be significantly deteriorated.

Therefore, as a method for preventing such deterioration,
As described in Japanese Patent Application Laid-Open No. 2-239380, a scaling method for preserving a local density distribution before and after processing called a projection method, and an output multivalued by a projection method called an error diffusion method are converted into a local density distribution. A scaling method has been proposed which combines pseudo halftone processing for returning to a binary image while saving.

However, the projection method needs to calculate a weighted sum corresponding to the area occupied by each pixel before conversion when each pixel after conversion is mapped on the image before conversion, and therefore the calculation method is performed. It has the drawback of being extremely time-consuming (Fig. 2
reference).

Therefore, as a method of shortening this calculation time, in Japanese Patent Laid-Open No. 4-299470, the projection method is N / N.
There has been proposed a method of simplifying the processing by expanding it into a combination of an integer ratio scaling represented by M and a real number scaling having a linear interpolation function, a cosine interpolation function, a Gaussian interpolation function and other interpolation functions.

[0007]

However, even in the above processing method, a considerable processing time is required for each of the integer ratio scaling and the real number scaling.

The present invention has been made in view of the above points, and in an image processing apparatus capable of performing scaling / resolution conversion processing of a binary image, high-speed processing is performed without causing image quality deterioration due to moire or the like. An object of the present invention is to provide an image processing device capable of performing the above.

[0009]

An image processing apparatus according to the present invention is a magnification conversion for converting a binary image into N / M (where N and M are natural numbers) times which is a value approximating the magnification. Means, an integer ratio enlargement / reduction conversion value storage means for storing a table for outputting binary or multi-valued N pixels which are enlargement / reduction images corresponding to the input of M pixels of the binary image, and the integer ratio. An integer ratio enlarging / reducing means for forming a binary or multivalued image by multiplying a target binary image by N / M using a table stored in the enlarging / reducing conversion value storage means, and the integer. A real number ratio enlarging / reducing means for enlarging / reducing a binary or multi-valued image formed by the ratio enlarging / reducing means while storing density by interpolation at a magnification of M / N times, and the real number ratio enlarging / reducing means. Binarization method for binarizing images enlarged / reduced by With the door.

Further, the integer ratio enlarging / reducing means and the real number ratio enlarging / reducing means are respectively operated in the main scanning direction and the sub-scanning direction of the target image.

[0011]

The present invention prepares a table for collectively reading a plurality of pixels on an original image and outputting the result obtained by scaling the pixel pattern to an integer ratio, and using this table, a binary or multi-pixel for a plurality of pixels is batched. It is configured so that the integer ratio scaling result of values can be obtained at high speed. According to the present invention, since a plurality of pixels on an original image can be collectively subjected to integer ratio scaling, it is possible to process the integer ratio scaling processing portion of the resolution conversion processing that prevents image quality deterioration at high speed.

Further, when the integer ratio / real number ratio scaling is performed, the scaling process in the main scanning direction and the subscanning direction of the image are executed independently, and the integer ratio scaling and the real number ratio scaling are performed in the main image. Since the processing is performed independently in the scanning direction and the sub-scanning direction, it is possible to simplify each scaling operation.

[0013]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of an image processing apparatus according to the present invention. In the figure, 10 is a binary image storing means for storing an original image, 11 is an integer ratio enlarging / reducing means for enlarging / reducing an image at a magnification represented by an integer ratio, and 12 is an image scaling at a magnification represented by a real number ratio. A real number ratio enlarging / reducing means, 13 is a binarizing means for returning the multi-valued image data by the enlarging / reducing processing to binary data again, 14
Is a binary image display means for displaying the binary image data scaled up and down by these means, and 15 is a magnification conversion means for separating the overall magnification into a portion represented by an integer ratio and a remaining portion represented by a real number ratio. , 16 have a magnification conversion table for collectively converting a plurality of input pixels, and an integer ratio enlargement / reduction conversion value for generating the contents of this table based on the integer ratio enlargement / reduction ratio separated by the magnification conversion means 15. It is a storage means.

Here, consider the case where the image is reduced. Figure 2
In, the pixels separated by broken lines indicate the image before conversion, and the pixels separated by the solid line indicate the corresponding converted image. When the scaling of an arbitrary magnification is processed by the normal projection method, it is also possible to compare the pixel with the diagonal line with the dotted side and the pixel with the dotted line in FIG. As can be seen, the number of original pixels (hereinafter abbreviated as reference pixels) that are referred to in order to obtain the value of the converted pixels (hereinafter abbreviated as converted pixels) differs depending on the converted pixels. For this reason, the processing considering these cases becomes very complicated, which causes a decrease in processing speed and an increase in device scale.

Therefore, in the present embodiment, the processing for obtaining an integer ratio close to the specified magnification by the magnification conversion means 15 is performed, and the integer ratio enlarging / reducing unit 11 performs enlarging / reducing the integer ratio in advance, thereby remaining the remainder. The real number ratio enlargement / reduction processing is simplified.
The magnification conversion unit 15 will be described below with reference to the flowchart of FIG.
An example of the processing in 1 will be described. N and M are variables.

The number of main (sub) scan pixels of the input image is set to N, and the number of main (sub) scan pixels of the output image is set to M.

The greatest common divisor of N and M (GCM: gre
atest common measure). If N and M are relatively prime (GCM = 1),
Proceed to.

Set N = N / GCM and M = M / GCM, and return to.

N and M are both predetermined threshold values T
If h or less, the N and M are input to the integer ratio enlarging / reducing converted value storage means 16 and the process is terminated. If either N or M is larger than the threshold Th, proceed to.

If the threshold value Th is set to a too large value, the capacity of the conversion table in the integer ratio enlarging / reducing conversion value storage means 16 becomes large, so a small value of, for example, about 4 is used. In this example, the same threshold value was used for N and M,
The determination may be performed using different thresholds.

When N / M <0.5, N = 1, M
= 4, N and M are input to the integer ratio enlargement / reduction conversion value storage means 16 and the processing is terminated.

When N / M <0.75, N = 1,
M = 2, the N and M are input to the integer ratio enlarging / reducing converted value storage means 16, and the process is terminated.

When N / M <1.0, N = 3, M
= 4, N and M are input to the integer ratio enlargement / reduction conversion value storage means 16 and the processing is terminated.

With N = INT (N / M) and M = 1,
The N and M are input to the integer ratio enlargement / reduction conversion value storage means 16 and the processing is ended. Note that INT (N / M) is N /
Let the integer part of M be truncated.

By such a processing procedure, integer ratio expansion /
The enlargement / reduction ratio of the reduction means 11 is determined. In the above example, when the condition of the procedure is not satisfied, the magnification is set to any one of 1/4, 1/2, 3/4, 1, 2, ... In the procedure. , 1/5, 4/5, 3 /
Finer scale factors such as 2 may be used. This realization method uses 0.5, 0.75, 1.
This can be done by changing 0 and adding more processing steps to do the same.

Further, when the overall magnification is 1.0 or more, the number of pixels referred to by the projection method is limited to a maximum of 4 pixels, so that the integer ratio enlarging / reducing means 11 does not perform enlarging or reducing, The scaling may be performed only by the real number ratio scaling unit 12.

The enlargement / reduction ratio R in the real number ratio enlargement / reduction means
Is a magnification excluding N / M set by the above procedure, that is, (overall magnification) / (N / M).

Based on N and M thus obtained, the integer ratio enlargement / reduction conversion value storage unit 16 creates a conversion table as described below. For example, if the input image is 100
When the image is composed of × 100 pixels and the image is reduced to 85 × 85 pixels, N = 3 and M = 4 in the flow of FIG. That is, by performing the process of generating 3 pixels from 4 pixels, a reduction of 0.75 times is performed, and the remaining (0.85 / 0.7
5) The real number ratio enlarging / reducing unit 12 processes the multiplication.

In order to perform the process of generating 3 pixels from 4 pixels while maintaining the local density distribution, as shown in FIG. 4, each of the four input pixels (see FIG. 4A) is repeated 3 times. There is a method of obtaining 12 pixels (see FIG. 2B), and obtaining an average value of 4 pixels from the beginning of the 12 pixels to obtain 3 pixels of output (see FIG. 2C). For example, FIG. 4 (b)
The average value of the last 4 pixels of the 12 pixels shown in is (0+
1 + 1 + 1) × 255/4 = 191.25≈BF (16
It is written in hex). Such processing requires high-speed processing because it requires processing such as generating an intermediate result of 12 pixels on the way and adding / dividing pixel values. So input 4
A conversion table for collectively generating three output pixels from the pixels is created, and the conversion table is subjected to 3/4 times reduction without performing the addition / division. Specifically, as shown in FIG. 5, a lookup table of 4-bit input / 24-bit output is prepared, and output values corresponding to the input binary numbers 0000 to 1111 are calculated in advance and stored. For example, if the input 4 pixels are 1101 in binary notation, the output is FF80BF in hexadecimal notation.
Becomes By creating such a table in advance, it becomes possible to execute scaling with an integer ratio at high speed. Further, in this example, the look-up table of 4-bit input / 24-bit output is used. However, in order to further improve the efficiency, it is also possible to use the look-up table of 8-bit input / 48-bit output to perform conversion in batch with 8 pixels.

In this example, a table which outputs multi-valued (8 bits) for a binary input is used. However, when the magnification is an integral multiple such as N / M = 2/1 = 2, Can reduce the capacity of the table. In the case of an example, it is also possible to use a table that outputs 2 (= N / M) × k bits that are binary data with respect to k bits that are binary data. Even in this case, since the magnification is an integral multiple, the objective of preserving the local density of the image before enlargement is achieved in the image after enlargement.

Using the conversion table created by the above method, the integer ratio enlarging / reducing means 11 inputs an image from the binary image storage means 10 and performs enlarging / reducing processing for each of the main scanning / sub scanning. The result is input to the real number ratio enlarging / reducing means 12.

The real number ratio enlarging / reducing means 12 performs enlargement / reduction defined by (overall magnification) / (enlargement / reduction magnification performed by the integer ratio enlarging / reducing means 11). If this magnification is 1.0
When it becomes, the binarization means 1 without performing any processing.
The image data is sent to 3. In other cases, the scaling is performed by the following method, for example.

According to the procedure of the flowchart of FIG. 3, the enlargement / reduction ratio in the real number ratio enlarging / reducing means 12 is often set to be larger than 1.0. Therefore, it is rare that 9 or more reference pixels as shown in FIG. 2 are required, and the process can be performed by referring to at most 4 pixels as shown in FIG. Therefore, it is possible to perform scaling with a real number ratio in which case classification is simplified. FIG. 6 shows an example in the case of enlarging by 3/2 times, and pixel values P ₀ , P ₁ , P ₂ , P
The image of 2 pixels × 2 pixels of ₃ has pixel values P ′ ₀ , P ′ ₁ ,
_{P '2, P' 3,} P '4, P' 5, P '6, P' 7, 3 pixels × the P _'8
If it is enlarged to a 3-pixel image, for example, the pixel value P ′ ₄ is P ′ ₄ = w ₀ · P ₀ + w ₁ · P ₁ + w ₂ · P ₂ + w ₃ · P
Calculated by calculation of ₃ . However, w ₀ , w ₁ , w ₂ , w
₃ is a weighting coefficient according to the area occupied by each reference pixel when each converted pixel is mapped on the image before conversion.

In the example of the flowchart of FIG. 3, many reference pixels as shown in FIG. 2 are required when the overall magnification is smaller than ¼, but when the magnification is small, up to 4 pixels are required. Even if the calculation is performed by approximation using a reference, moiré does not occur so much, so there is not much problem. Further, if the flowchart of FIG. 3 is modified to change the condition-determined portion with “N / M <0.5” to a slightly smaller value such as “N / M <0.45”, Magnification of 0.47
In such a case, there are cases in which many reference pixels as shown in FIG. 2 are required. However, even in such a case, it is only a part of the whole pixel that requires 4 pixels or more, and the contribution of reference pixels of 4 pixels or more is small for that part as well. Even if the approximation calculation is limited to, the occurrence of moiré is small, and the sharpness can be maintained as compared with the case where the remaining 0.47 / 0.25 times enlargement is performed after 0.25 times. is there.

As a result of the processing performed by the method described above, the original image data is converted from a binary state to a multi-valued state. Therefore, in order to restore this multi-valued information to the original binary state, the binarizing means 13 performs binarization by using the error diffusion method which is a method of maintaining the local density. The error diffusion method is a known technique and will not be described here. In the case where the overall magnification is 2.0 times, the binary data may be directly input to the binarizing means 13 as a result of the processing in the procedure described above. This binarization process is passed. Further, although FIG. 1 is premised on a binary display device as an output device, the binarizing means 13 itself may be omitted when outputting to a device capable of multivalued display. Is.

The image data binarized by the binarization unit 13 is displayed on the binary image display unit 14 as a good enlarged / reduced image without image quality deterioration due to moire or the like.

Next, an embodiment of the second invention will be described in detail.

In the embodiment of the first aspect of the invention, since the main scanning direction and the sub scanning direction are processed simultaneously, the real number ratio enlarging / reducing means 12 needs to refer to a maximum of 4 pixels.
This simplifies the processing as compared with the case of referring to 9 pixels or more, but still requires a slightly complicated processing procedure. Further, with respect to the scaling in the integer ratio enlarging / reducing means 11, as for the scaling in the sub-scanning direction, a plurality of pixels are collectively converted, and therefore an operation of collecting pixel data continuous in the sub-scanning direction is required, which makes the processing complicated. Become.

Therefore, in the second aspect of the invention, as shown in FIG. 7, main scanning direction integer ratio enlarging / reducing means 70 and sub scanning direction integer ratio enlarging / reducing means 71 are provided, and main scanning direction real number ratio enlarging is performed. The number of reference pixels at the time of enlargement / reduction in the real number ratio is set to 2 pixels by providing the / reduction unit 72 and the sub-scanning direction real number ratio enlarging / reducing unit 73 and performing the enlargement / reduction in the main scanning direction and the enlargement / reduction in the sub-scanning direction independently. The processing is further simplified, and in the case of scaling with an integer ratio, batch scaling with the conversion table is performed only in the main scanning direction. The operation of each unit will be described in detail below.

As in the previous embodiment, the magnification conversion means 15 calculates the integer ratio scaling ratio from the overall ratio. As a result, the magnification in the main scanning direction is stored in the integer ratio enlargement / reduction conversion value storage means 16 as in the previous embodiment, and the magnification in the sub-scanning direction is expanded / reduced in the sub-scanning direction.
Input to 1. Integer ratio enlargement / reduction conversion value storage means 16
The conversion table created by
It is used only for the enlargement / reduction processing in the reduction means 70. Since pixel data in the main scanning direction is normally stored with 8 pixels as 1 byte, the integer ratio scaling factor N / M is 1/8, 2 /
8, 3/8, 4/8, 5/8, 6/8, 7/8, 8/8
If the processing in the magnification conversion means 15 is changed to any one of the above, the conversion table can be configured so that the output is created in the unit of 1 byte of the input, which is effective in terms of processing speed when the processing is performed by software. Is.

The image data scaled in the main scanning direction by the integer ratio magnification is input to the sub scanning direction integer ratio scaling means 71 for scaling in the sub scanning direction. In the scaling in the sub-scanning direction, unlike the previous embodiment, scaling is performed by the method shown in FIG. 8 without using the conversion table. N =
Taking 3 and M = 4 as an example, input image data (see FIG.
(See (a)) Each main scanning line is repeatedly output three times to once generate a three-fold enlarged output (see (b) in the figure).
After that, by averaging the pixel values of the pixels at the same main scanning position for four lines (see FIG. 7C), the processing can be performed without being aware of the case when the number of reference pixels is different. You can This kind of processing method seems to take time at first glance, but copying data such as continuous data for one line can be executed at high speed with a device having a function such as BitBlt (bit block transfer), so the overall processing speed is Can be improved.

The image data which has been subjected to the integer ratio scaling in both the main scanning direction and the sub-scanning direction by the above method is input to the main scanning direction real number ratio enlarging / reducing means 72. In the main scanning direction real number ratio enlarging / reducing means 72, as shown in FIG. 9B, the enlargement / reduction of the real number ratio magnification is performed only in the main scanning direction. In the previous embodiment, as shown in FIG. 9A, the main scanning / sub-scanning was expanded / contracted at the same time. Therefore, even if the magnification is 1.0 or more, the number of reference pixels is different from 1 pixel to 4 pixels. But
When only the main scanning is scaled up or down independently, as shown in FIG. 9B, the number of reference pixels is only two, that is, one or two. Since the number of reference pixels is within 2 pixels, if the area ratio of a certain reference pixel in the conversion pixel is W, the area ratio of the other reference pixel is (1-W), and (output pixel value) = The output pixel value can be calculated by a simple formula of (left reference pixel value) × W + (right reference pixel value) × (1-W). Further, since the area ratio W is determined by the decimal part of the coordinate value on the original image of the converted pixel, if the value of the area ratio W with respect to the value of the decimal part is calculated in advance and made into a table, higher speed processing becomes possible. .

Next, the sub-scanning direction real number ratio enlarging / reducing means 7
3, the enlargement / reduction processing is performed at the real magnification in the sub-scanning direction. As shown in FIG. 9C, when only the sub-scanning direction is enlarged, the number of reference pixels is 2 or less as in the main scanning direction. Therefore, similarly to the scaling in the main scanning direction, (output pixel value) = (upper reference pixel value) × W + (lower reference pixel value) × (1-W) Can be calculated. Magnification is 1.0
In the following cases, the number of reference pixels is more than 2 pixels and the above equation is not satisfied. However, as described in the previous embodiment, a good result can be obtained by approximately using the above equation. To be

The image data scaled in this way is
Similar to the previous embodiment, it is re-binarized by the binarizing means 13 and displayed on the binary image display means 14.

In this way, by configuring the main scanning direction and the sub scanning direction to be independently scaled, the calculation in each scaling process is simplified, and as a result, high-speed processing becomes possible.
Incidentally, in FIG. 7, the processing order is “integral ratio scaling in main scanning direction” →
"Integral ratio scaling in sub-scanning direction" → "Real number ratio scaling in main scanning direction"
→ The explanation has been made in the order of "sub-scan direction real number ratio scaling", but as is clear from the explanation, "main-scan direction integer ratio scaling" → "main scanning direction real-number ratio scaling" → "sub-scanning direction integer ratio scaling" →
The processing may be performed in an order such as “scaling in the sub-scanning direction real number ratio” or in an order in which the sub-scanning direction is first.

[0047]

As described above, according to the present invention,
Since the integer ratio expansion / reduction is performed by the table and the real number ratio expansion / reduction is performed while storing the density by interpolation, it is possible to easily and quickly process the expansion / reduction conversion without image quality deterioration due to moire. You can

[Brief description of drawings]

FIG. 1 is a block diagram of an image processing apparatus showing an embodiment of the first invention.

FIG. 2 is a diagram illustrating reference pixels in a projection method.

FIG. 3 is a flowchart showing an example of an operation of a magnification changing unit.

FIG. 4 is a diagram illustrating scaling with an integer ratio.

FIG. 5 is a diagram illustrating a conversion table of an integer ratio enlargement / reduction conversion value storage unit.

FIG. 6 is a diagram for explaining scaling by a projection method.

FIG. 7 is a block diagram of an image processing apparatus showing an embodiment of the second invention.

FIG. 8 is a diagram for explaining integer ratio scaling in the sub-scanning direction.

FIG. 9 is a diagram for explaining real number ratio scaling when scaling is performed separately for main scanning and sub scanning.

[Explanation of symbols]

10 ... Binary image storage means, 11 ... Integer ratio enlarging / reducing means, 12 ... Real number ratio enlarging / reducing means, 13 ... Binarizing means,
14 ... Binary image display means, 15 ... Magnification conversion means, 16 ...
Integer ratio enlarging / reducing converted value storage means, 70 ... Main scanning direction integer ratio enlarging / reducing means 71 ... Sub scanning direction integer ratio enlarging / reducing means, 72 ... Main scanning direction real number ratio enlarging / reducing means, 73 ...
Sub scanning direction real number ratio enlargement / reduction means

Claims (2)

[Claims] 1. A magnification conversion means for converting N / M (where N and M are natural numbers) times, which is a value approximating a magnification for enlarging / reducing a binary image, and inputting M pixels of the binary image. Correspondingly, integer ratio enlargement / reduction conversion value storage means for storing a table for outputting binary or multi-valued N pixels that are enlargement / reduction images, and integer ratio enlargement / reduction conversion value storage means. Integer ratio enlargement that forms a binary or multi-valued image by multiplying a target binary image N / M times using a table
A reducing means; a real number ratio enlarging / reducing means for enlarging / reducing the binary or multi-valued image formed by the integer ratio enlarging / reducing means while enlarging / reducing the density by M × N / magnification. An image processing apparatus comprising: a binarizing unit that binarizes an image enlarged / reduced by a real number ratio enlarging / reducing unit. 2. The integer ratio enlarging / reducing means and the real number ratio enlarging / reducing means are respectively executed in a main scanning direction and a sub-scanning direction of a target image.
The image processing device described.