JPH0129112B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to a method for creating a halftone image for enlarging or reducing an image when expressing halftones using a dot matrix. A method of expressing an image having halftones using a dot matrix is disclosed in Japanese Patent Publication No. 49-42171 and the like. This method creates a dot matrix pattern corresponding to the density, as shown in Figure 1b, and
Patterns D ₃ , D ₅ , corresponding to the density for each -3,...
D ₈ , . . . are selectively recorded and a reproduced image having halftones as shown in c is reproduced. When an n×m dot matrix is used, the number of pixels of the reproduced image shown in c becomes n times as many in the column direction (vertical direction) and m times as many in the row direction (horizontal direction) as the original image shown in a. This will be referred to as a reproduced image that is n times larger in the column direction and m times larger in the row direction. FIG. 1 shows an example where n=m=4. First, a conventional enlarging method in the halftone recording method shown in FIG. 1 will be explained. Now consider expansion by k times in the column direction. If k is a multiple of 4, for example 8, then k/n = 2, so as shown in Figure 2, the pattern that corresponds twice to each pixel of the original image of a is D ₃ , D ₃ , D ₅ , D Record as ₅ , D ₈ , D ₈ , …. If the magnification k is not a multiple of 4, but for example k = 6, then k/n = 3/2, so as shown in Figure 3, pattern D ₃ corresponding to pixel 1-1 of the original image of a is 2
pattern D ₅ corresponding to pixels 1-2 is recorded once, pattern D ₈ corresponding to pixels 1-3 is recorded twice, and pattern D ₉ corresponding to pixels 1-4 is recorded once. , . . . , it is possible to obtain a reproduced image approximately enlarged six times as shown in c. However, in this case, the sampling intervals of the original image are uneven and the reproduced image becomes unnatural. On the other hand, considering the case of reduction, if the magnification k is smaller than the side length n of the dot matrix and is a divisor of n, for example k = 2, then k/n = 1/2, so as shown in Figure 4, Original image every 2 pixels 1-1, 1-3, 1
-5, 1-7, ... and corresponding patterns D ₃ , D ₈ , D ₁₀ , D ₄ ,
... is selected and recorded to obtain the reproduced image shown in c.
In this case, only 1/2 of the information in the original image a is used in the column direction, and the rest is discarded, which inevitably degrades the resolution. If the magnification k is not a divisor of n, for example k=3, then k/n=3/4, so as shown in FIG.
3 pixels from 1-3, 1-4 1-1, 1-2, 1-
3, pixels 1-4 are discarded, and the corresponding patterns D ₃ , D ₅ , and D ₈ are recorded.
In this case as well, the information is not used effectively, and the resolution of the reproduced image c deteriorates, and unnaturalness appears due to sampling at irregular intervals. That is, in either case, convert k/n to an approximate fraction and get k'/n
′, the n′ pixels of the original image will be sampled k′ times and the corresponding pattern will be recorded k′ times. Sampling is performed so that each pixel is selected as evenly as possible. This method also works in the case of reduction (k < 1) by converting k/n into the form k'/n' such that k becomes an integer.
Exactly the same applies. In the conventional method described above, if k/n is larger than 1 and is a non-integer, the reproduced image will be unnatural, and if k/n is smaller than 1, the resolution will deteriorate due to not using information effectively. It has drawbacks such as: An object of the present invention is to overcome the drawbacks of the conventional methods as described above, and to obtain an enlarged or reduced reproduced image with as little deterioration of resolution as possible and with less unnaturalness. The feature of the present invention is that the pixels of the original image are made to correspond to the corresponding pixels of the density pattern, whereas the conventional method makes the pixels of the original image correspond to the density pattern.
The basic case of this type of recording method is that of
- Disclosed in Publication No. 146582. To briefly explain this method, as shown in Figure 6, the density pattern corresponding to the density _D3 of pixel 1-1 of the original image is
Check the presence or absence of a dot in pixel 2-1 at the corresponding position of _D3 , and if there is a dot, record the dot, and if there is no dot, do not record the _dot . 2-2 and do the same. In this method, regardless of the size of the density pattern, it is possible to obtain a reproduced image in which the number of pixels in the original image is equal to the number of pixels in the reproduced image, and the number of pixels in the vertical and horizontal directions is one times as large. It is stated in the above-mentioned patent publication that this method has a superior resolution compared to the reproduced image that is 1 times as large as the conventional method. The present invention extends this to the case of enlarging and reducing, and is superior to conventional methods in terms of resolution and naturalness of reproduced images, and is easy to implement. The present invention will be explained in detail below. Although the case where a 4×4 density pattern is used will be explained as an example, the scope of application of the present invention is not limited to this case, and patterns of arbitrary sizes such as 3×3 and 3×4 can be used. It is possible. In the following, only the column direction will be described up to and including FIG. 10, but the same process can be performed in the row direction as well. Consider expansion with a magnification k=5 in the column direction. As shown in FIG. 7, the corresponding position of the density pattern is shifted one by one each time it is sampled, regardless of whether it is the same pixel or a different pixel from the original image a. That is, the corresponding pixel of pattern _D3 corresponding to the density _D3 of pixel 1-1 is initially 2-
1-1, and since there is a dot there, recording 3-1-1 is also performed on the reproduced image c. Next, sample the same pixel 1-1 of the original image again, and
Looking at the newly corresponding pixel 2-1-2 of pattern _D3 corresponding to _D3 , there is no dot there, so no dot is recorded in the reproduced image 3-1-2. Until the fourth sampling, the reference pixels of the pattern are 2-1-
3, 2-1-4, and the fifth reference pixel 2-
1-5 returns to the first pixel 2-1-1. At this time, the reference pixels of other patterns are also moving in the same way. That is, the first reference pixel of pattern D ₅ corresponding to the density D ₅ of the next pixel 1-2 of the original image of a is 2-1-1, and the following pixel is 2-2-2, 2-2-3, 2- 2-4
2-2-5 returns to 2-2-1. Therefore, if k is an integer, even if k/n is a non-integer such as 3/2, the sampling interval of the original image will be uniform, and unnaturalness unlike the conventional method will not appear. When k/n is smaller than 1, for example, when k=3, k/n=3/4 is recorded as shown in Fig. 8, and in the conventional example, the original image information is discarded or the reproduced image is Phenomena that used to look unnatural no longer occur. When the magnification k is a non-integer greater than 1, the conventional method becomes complicated. For example, when the magnification k=3/2, the conventional method involves sampling three pixels from eight pixels of the original image and recording three patterns corresponding to these three pixels, which is complicated. However, according to the present invention, as shown in FIG. 9, pixel 1-1 of the original picture a is sampled twice and pixel 1-2 is sampled once and recorded. That is, look at the corresponding pixel 2-1-1 of the pattern D ₃ corresponding to the density D ₃ of the pixel 1-1, record the dot 3-1-1, sample the pixel 1-1 again, and create the corresponding pattern. Look at pixel 2-1-2 and see dot 3-1.
-Record 2. Actually, there is no dot at pixel 2-1-2, so it is not recorded. Next, the density of pixels 1-2
Pixel 2-2-1 of pattern D ₅ corresponding to D ₅ is looked at and recorded in pixel 3-2-1. Next, the corresponding pixel 2 of pattern D ₅ corresponding to the density D ₅ of pixels 1-2
-2-1 is seen and recorded, and pixels 2-3-1 and 2-3-2 of pattern _D8 corresponding to pixel 1-3 are seen and recorded. Let's move on to this process. In the case of reduction where k is smaller than 1, information on the original image is discarded, but compared to the conventional method, the amount discarded is much smaller. For example, when k=2/3, k/n=1/6, so in the conventional method, one pixel is sampled for every six pixels of the original image, and the remaining five pixels are discarded. According to the present invention, as shown in FIG.
Pixel 1-1, 1-2, 1-3 to 2 pixels 1-1,
Sample 1-2 and discard 1 pixel 1-3 without sampling. Thereafter, one pixel sampling is skipped every three pixels in the same way. When sampling is skipped, the reference pixels of the pattern do not move.
That is, pixel 2 of pattern D ₃ corresponds to pixel 1-1.
-Look at 1-1, pattern corresponding to pixels 1-2
Look at pixels 2-2-1 of D ₅ , skip pixels 1-3, and look at pixels 2-2 of pattern D ₉ corresponding to pixels 1-4.
-See 3-1. 2- corresponds to pixels 1-4
It becomes 2-3-1 instead of 3-2. The above is summarized in Table 1.

[Table] Although the explanation has been limited to the column direction, the present invention can be similarly applied to the row direction. For example, if we consider expansion four times in the column direction and three times in the row direction, the result will be as shown in FIG. The magnification in the row direction and column direction can be freely set independently. When moving from column to column, the row to be referenced in the pattern returns to the first row. The effects of the present invention will be illustrated by giving one example. 1st
Fig. 2c shows a reproduced image according to the conventional method when the original image in a is multiplied by 6 times in the column direction and 3 times in the row direction, and d shows the reproduced image according to the present invention. The density of each region 4, 5, 6, and 7 of the original image a consisting of 4×4 pixels is D ₀ , D ₁₅ , D ₀ , and D ₁₀ , respectively, and the density distribution of each pixel is as shown in b. do. If this is recorded using the conventional method, the horizontal direction will be 4 rows 8 of the original picture a.
- First 3 rows of 1, 2, 3, 4 8 - 1, 2, 3
Since only the area 6 is sampled, the area 6 is hardly expressed, and the area 7 is not expressed at all. In contrast, in the present invention, four rows 8-1, 8-2, 8-
3, 8-4 are all 8-1″, 8-2″, 8-3″,
8-4", and can be expressed much more faithfully to the original image. Next, a specific example of a circuit in which the present invention is implemented will be explained. FIG. The recording medium 10 is wound around the recording medium 10, and recording is performed by rotating the drum and moving the head 35 in the row direction.The rotary encoder 12 detects the rotation of the drum and generates a column clock 13 for each recording start point of each column. A pixel clock 14 is output for each pixel.A column address generator 15 outputs a pixel clock 14 for each pixel.
3, the original image memory 17 is stored according to the magnification in the row direction.
A column address 16 is generated. A row address generator 18 receives the pixel clock 14 and generates a row address 19 of the original image memory 17 in accordance with the magnification in the column direction. For example, if the magnification in the column direction is 3x,
The row address is advanced by one every three pixel clocks. Similarly, at a magnification of 1.5x, the row address is advanced by two for every three pixel clocks. Original picture memory 17
From then on, 4-bit density data 20 is output every time an address is designated and input into the pattern memory 21. The pattern memory 21 stores 16 density patterns corresponding to each density D ₁ to D ₁₅ , and one pattern is selected according to the density data 20 . The row counter 23 is a 2-bit binary counter that is reset by the column clock 13 and counts the pixel clock 14. The row count value 24 specifies one row from each pattern consisting of 4 rows as a 4-bit row pattern 25. Output. Column clock 13
The column counter value 27 of the 2-bit binary counter 26 is input to the 4/1 selector 28, which selects 1 bit from the 4-bit row pattern 25 and outputs the image signal 29. The image signal 29 is modulated onto a carrier 31 by a modulator 30 consisting of an analog switch, and the modulated signal 32 is sent to an amplifier 35.
As a result, a drive signal 34 is generated to drive the head 35. Pattern memory 21 includes random access memory RAM and read-on memory.
Use ROM. The circuit configuration shown in FIG. 13 is simply a general recording circuit consisting of a drum 11, a head 35, a memory 17, etc., and a pattern memory 21, a row counter 23, a column counter 26, and a selector 28, and is easy to implement. It is. Next, another embodiment will be described. This embodiment relates to a method for improving resolution by sacrificing the ability to express halftones to some extent. In the method of expressing halftones using a dot matrix-like density pattern, it is possible to obtain a desired number of halftones by increasing the size of the matrix, but the resolution deteriorates accordingly. There are cases, such as when recording television images in a small size, where it is desirable to improve the resolution even if it means sacrificing some intermediate tones. The present invention meets this requirement by using a plurality of density pattern groups. FIG. 14 shows two circuits around the pattern memory 21 in FIG. 13, and when the magnification is small, a small pattern 21-1 is selected, and when the magnification is large, a large pattern 21-2 is selected. If the small pattern 21-1 consists of eight types of 3×3 matrices, three bits are sufficient for the image signal 20-2. To create a 3-bit image signal from the 4-bit image signal 20-1 for the large pattern 21-2, it is sufficient to simply take the upper 3 bits. The row counter 23-2 is a ternary counter that is reset by the column clock 13 and counts the pixel clock 14, and specifies the row of the pattern. Column counter 2
6-2 is a ternary counter that counts column clocks, and a 1/2 selector 36 selects one bit from the 3-bit row pattern 25-2 output from the pattern memory 21-2. -2 and the output 38-1 from the pattern memory 21-1 according to the magnification.
Select one according to the following. As explained above, according to the present invention, a high resolution image can be easily obtained by selecting a plurality of groups of density patterns having different sizes according to the magnification. Next, a third embodiment will be described. This embodiment is intended to reduce the required storage capacity of the original image memory. Conventionally, when storing image information, an original image 41 is generally stored on a drum 42, as shown in FIG.
The original image is scanned by rotating the drum and moving the photoprint 44 in the row direction. reflected light 4
3 is received by the format 44 and an analog electrical signal 4 is received.
5 and amplified by an amplifier 46, converted into a digital signal 49 by an analog-to-digital converter (AD converter) 48, and stored in the memory 50. The rotary encoder 51 outputs a column clock 52 for each rotation of the original image 41 and a pixel clock 53 for each pixel. Column address generator 54 and row address generator 55 receive column clock 52, pixel clock 53, and column address 56, respectively.
and row address 57 are generated. In the example of FIG. 15, each address of memory 50 has a storage capacity of 4 bits. By the way, in the recording method used in the present invention, in which dots are placed on pixels to be recorded or not, the amount of information required for recording is one bit per pixel. On the other hand, the original image memory needs to have enough information to specify one of the density patterns for each pixel, and when using 16 types of patterns as described above, 4 bits per pixel is required. storage capacity is required. Column direction magnification k and row direction magnification l
If the product kl of It becomes below. Generally, when the number of density patterns is γ and R is an integer not exceeding 1 + log ₂ r,
If the product kl of the column and row magnifications is less than or equal to R, then
The amount of information required for recording is less than the amount of information in the original image memory. The present invention utilizes this fact to reduce the required storage capacity of the original image memory.If kl is less than or equal to R, the image information is binarized and converted to 1 bit per pixel at the stage of inputting the image information to the original image memory. It is converted into information and stored. Figure 16 shows the configuration. rotary encoder 6
1, a column clock 52 for each rotation of the drum, and k pixel clocks 63 for each pixel of the original image according to the magnification k in the column direction. If the magnification in the row direction is l, the photoprinter 44 scans the same column l times.
k and l are integers, and if they are non-integers, the scaling factors are equivalently set to k and l. The digital image signal 49 is input to the density pattern memory 21, and one pattern corresponding to the density is selected. Row counter 23 is a 2-bit binary counter that is reset by column clock 52 and counts pixel clock 63, and row count value 24 specifies the row of the pattern. Column counter 26 is a 2-bit binary counter that counts column clock 52, and when column count value 27 is input to 4/1 selector 28, one bit is selected from output row pattern 20 of pattern memory 21. Row address generator 64 is reset by column clock 52 and counts pixel clock 63 to generate row address 65 in memory 68, which is then reset by column address generator 66.
counts the column clocks and generates column address 67 in memory 68. A 1-bit image signal 29 is stored at a designated address in the memory 68. That is, while the original image memory 17 in FIG. 13 has a storage capacity of 4 bits for an arbitrary address, this memory 68 only needs to have a storage capacity of 1 bit for an arbitrary address. For example, if the original image memory 17 in Figure 13 has 240 and 320 pixels in the column and row directions, each pixel has 4 bits, so the total capacity is 240 x 320 x 4.
= 307200, and this value does not change depending on the magnification. On the other hand, the capacity of the memory 68 in FIG. 16 is (240 x 2) x (320 x 2) = 307200, which is the same as the memory 17, if the magnification in both the column and row directions is 2, but If both the magnifications are 1, the number becomes 240×320=76800, which saves memory. If the memory capacity is 307200, it will be possible to store four screens of images each with a magnification of 1x in the column and row directions. The circuit of FIG. 16 can also be easily implemented. Next, a fourth example will be described. This embodiment is characterized by real-time storage. In the configuration of FIG. 16, when the magnification l in the row direction is an integer greater than 1, the photoprint 44 scans the same column l times. If the original image does not disappear, such as a photograph, it can be scanned many times, but if it is a temporary signal, such as an on-air television signal, it must be stored in real time, and the configuration shown in Figure 16 is used. Can not. This embodiment solves this problem by a simple device. FIG. 17 shows an image signal 72 of an ordinary television 71 in the vertical direction (column direction) and horizontal direction (row direction).
A device for doubling each image and storing it in the memory 85 is shown. If we sample 240 out of 252 horizontal scanning lines, the number of pixels in the vertical direction is 240 x 2 = 480, the number of pixels in the horizontal direction is 240 x 3/4 x 2 = 260 pixels, and the horizontal scanning line is 63.5 μS. If the internal effective screen is 50μS, the pixel clock will be 640/5013MHz. This situation is shown in FIG. In FIG. 17, the image signal 72
is converted to 4 by the analog-to-digital converter 75.
The signal is converted into a bit digital signal 76 and inputted into a pattern memory 77 to select one pattern. Column counter 78 is a 2-bit binary counter that counts pixel clock 73, and column count value 79 specifies the column of the pattern and outputs a 4-bit column pattern 80. That is, one vertical column is output. The 4/2 selector 81 outputs the output 83 of the flip-flop 82 which is inverted by the horizontal synchronizing signal 74.
In response, the four bits of the column pattern 80 are divided into two upper and lower bits and outputted as an input signal 84 to the memory. Column address generator 86 is reset by horizontal synchronization signal 74 and pixel clock 73
The row address generator 88 counts the horizontal synchronization signal 74 and generates the column address 87 of the memory 85.
generate. The memory 85 has a storage capacity of 2 bits at the same address, and stores 2-bit signals 84 at the same time. Through the above operations, image information of twice the size in the vertical and horizontal directions is stored in the memory in real time. In this circuit, the pixel clock 73 has a high frequency of approximately 13 MHz, so a high-speed analog-to-digital converter 75 is required, but there are currently many products on the market that can be used at this speed, and it can be easily implemented using these. can. Next, a fifth embodiment will be described. This embodiment is an application of the fourth embodiment to real-time recording. In place of the memory 85 in FIG. 17, FIG. 19 shows a recording element capable of simultaneously recording a plurality of pixels, such as thermal heads 91-1, 91-.
2, . . , and records on a recording medium 92. Thermal head 91-1, 91-
Since the response speed of 2, . The image signal 72 in FIG.
It is assumed that the pixel clocks 73, 1 are transmitted slowly according to the response speed of pixel clocks 73, 1, 91-2, ...
It is assumed that the row clock 74 for each row has a frequency corresponding to that frequency. Image signal 72 to signal 84
The operations up to obtaining the value are the same as those shown in FIG. The signal 84 is input to an amplifier 94 and becomes a head drive signal 95, which drives heads 91-1, 91-2, fixed to a carriage 96, which is moved in the row direction by a pixel clock 73 and a pulse motor 97.
..., row clock 74 and pulse motor 98.
2 on the recording medium 92 driven in the column direction by
Record dot by dot. That is, by scanning the original image once, a record enlarged twice in the column direction is obtained.
Expansion in the row direction is performed using the pixel clock 73 and the image signal 72.
This can be easily obtained by multiplying the frequency of . FIG. 20 shows a case where a head capable of simultaneously recording a larger number of pixels is used. In the figure, a head 91 is used to simultaneously record one line in the row direction. From the image signal 72, by the same operation as shown in FIG.
1 records one line at a time. To enlarge in the column direction, scan the same row of the original image as many times as the magnification, or
The heads 91 may be arranged for a plurality of rows so that the number of rows corresponding to the magnification in the column direction can be simultaneously recorded.
According to the present invention, by simultaneously recording a plurality of pixels, it is possible to reduce the number of times an original image is scanned, and in some cases, it is possible to record an enlarged high-resolution image in real time. Next, a sixth embodiment will be described. This embodiment is an embodiment in which memory is used effectively. As mentioned above, in general, when the number of density patterns is γ and R is an integer not exceeding 1+log ₂ γ, if the product kl of the magnification in the column and row directions is less than or equal to R, the amount of information required for recording is The amount of information required is less than that of the original image memory. Conversely, if kl is greater than or equal to R, the amount of information required for recording will be greater than that of the original image memory. The present embodiment takes advantage of this fact and changes the way common hardware is used depending on the magnification, so that the amount of memory is less than a certain limit no matter what the magnification is. FIG. 21 shows only the flow of image signals. The connection of the interlock switch 101 is changed depending on the magnification. If kl<R, the interlocking switch connects to the upper contact as shown in the figure. In this case, the digital image signal 49 passes through the switch 101-1, is input to the pattern memory 21, and is input to the selector 102.
It becomes a dot pattern signal through the switch 1.
The signal is stored in the memory 17 through 02-2, and converted into a 1-bit signal by the selector 103 and used for recording. If kl>R, connect the switch 101 to the lower contact. In this case, digital image signal 4
9 is stored in the memory 17 through the switch 101-2, inputted into the pattern memory 21 through the switch 101-1, passed through the selector 102 and the selector 103, becomes a 1-bit signal, and is used for recording. . According to the present invention, if the capacity of the memory 17 is equal to the number of pixels of the recording medium,
When the magnification is small, multiple screens can be memorized and recorded at once, and even when the magnification is large, they can be stored without increasing the memory capacity. As described above, the present invention prepares a density pattern corresponding to the halftone to be expressed, samples the pixels of the original image a number of times according to the magnification, and moves the reference position of the corresponding density pattern in accordance with this sampling. By using a new method of enlarging and reducing the image size, a scaled halftone image can be obtained in which deterioration in resolution is minimized.

[Brief explanation of drawings]

Figures 1 to 5 have magnifications of 4, 8, 6, 2, and
FIG. 3 is a diagram illustrating a conventional halftone enlarged image creation method in the case of 3x magnification, in which a shows an original image, b shows a density pattern, and c shows a reproduced image. FIG. 6 is a diagram for explaining the halftone image creation method used in the present invention, in which a shows an original image, b shows a density pattern, and c shows a reproduced image. Figure 7 ~ 1st
Figure 0 is an explanatory diagram of the halftone image creation method according to the present invention when the magnification is 5, 3, 3/2, and 2/3, respectively, where a is the original image, b is the density pattern, and c is the reproduced image. FIG. 11 shows an example in which the method of the present invention is applied in the column and row directions, where a is the original image, b is the density pattern,
c shows a reproduced image. FIG. 12 is a comparison diagram between the conventional method and the method of the present invention, where a is an original image, b is a density distribution diagram of the original image, c is an image reproduced by the conventional method, and d is an image reproduced by the method of the present invention. FIG. 13 is a circuit configuration diagram showing the first embodiment of the present invention, and FIG. 14 is a circuit diagram showing the second embodiment of the present invention.
FIG. 15 is a block diagram for explaining the conventional image storage method.
6 is a circuit configuration diagram showing a third embodiment of the present invention,
FIG. 17 is a circuit configuration diagram showing a fourth embodiment of the present invention, FIG. 18 is an explanatory diagram showing how television images are stored in memory in the present invention, and FIGS. 19 and 20 are FIG. 21 is a circuit diagram showing the fifth embodiment of the present invention, and FIG. 21 is a circuit diagram showing the sixth embodiment of the present invention. 1-1, 1-2,... ...pixel of original picture, 2-1-
1, 2-1-2,... Pixel of density pattern, 3
-1-1, 3-1-2,... Pixels of the reproduced image,
4, 5, 6, 7...Original picture area, 8-1, 8-2,
... Column of original picture, 9-1, 9-2, ... Row of original picture, 10... Recording medium, 11... Drum, 12... Rotary encoder, 13... Column clock, 14... Pixel clock, 15... Column address generator , 16
... Column address, 17... Original image memory, 18... Row address generator, 19... Row address, 20... Image data, 21... Pattern memory, 23... Row counter, 25... Row pattern, 26... Column counter, 2
8...4/1 selector, 30...modulator, 31...carrier, 33...amplifier, 35...head, 36...2/
1 selector, 37... selection command signal, 41... original picture,
42... Drum, 43... Reflected light, 44... Photomal, 45... Photomal output signal, 46... Amplifier (current-voltage converter), 48, 75... Analog-digital converter, 50, 68, 85... Original image memory, 51, 61... rotary encoder, 52... column clock, 53, 63, 73... pixel clock, 5
4, 66, 86...column address generator, 5
5, 64, 88... row address generator, 71
... Braun tube, 72... Television image signal, 74... Horizontal synchronizing signal, 77... Pattern memory, 78... Column counter, 81... 4/2 selector, 82... Flip-flop, 91... Recording head, 92... Recording medium,
94...Amplifier, 96...Carriage, 97, 98...
Pulse motor, 99... row buffer, 101... interlocking switch, 102, 103... selector.

Claims (1)

[Claims] 1. A plurality of density patterns in which dots are arranged in a matrix according to the number of density steps to be recorded are provided, and the sampled pixels are sampled a number of times according to the magnification for enlarging or reducing the original image. An intermediate method characterized in that a density pattern corresponding to the density of is selected, a reference pixel of the density pattern is moved and referenced at each sampling, and a dot is stored or recorded depending on the presence or absence of a dot in the referenced pixel. How to create toned images. 2. The method for creating a halftone image according to claim 1, characterized in that the size of the density pattern is made different depending on the enlargement or reduction magnification. 3. The method for creating a halftone image according to claim 1, wherein the presence or absence of a dot in the referenced pixel is binarized and stored. 4. The method for creating a halftone image according to claim 1, wherein a plurality of pixels in the density pattern are referenced simultaneously and stored or recorded in real time. 5 After converting the density to be recorded into a multi-value digital signal, either the dots of the pixels directly referenced based on the multi-value digital signal are stored and then sequentially recorded, or the multi-value digital signal is first converted into a multi-value digital signal. The method of claim 1 is characterized in that it selects, depending on the magnification, whether to store and then sequentially output the dots of the referenced pixel based on the stored multivalued digital signal. The method for creating a halftone image according to item 1.