JPH0425749B2 - - Google Patents

Description

[Detailed description of the invention]

[Technical Field of the Invention] The present invention relates to an image processing method for obtaining high pixel density image information from image information whose density is detected at low pixel density. [Technical background] Image processing devices that perform halftone dot processing on original images such as photographs,
In a fax machine, an output image is obtained by binarizing the image information for the original image obtained by detecting the density of a multi-value level for each predetermined pixel. In such an image processing apparatus, it is generally considered to increase the density detection density, that is, increase the number of sampling pixels, in order to increase the resolution of the output image. However, if the number of sampling pixels is increased in this way, it will take time to detect the density of the entire original image, and as a result, the overall image processing time will be extended. Therefore, it becomes necessary to convert image information whose density is detected at a low pixel density into image information at a high pixel density. [Prior Art and Problems] Conventionally, as an image processing method for converting predetermined image information into image information with a higher pixel density, there is a method disclosed, for example, in Japanese Patent Application Laid-open No. 10221/1983. As shown in FIG. 1, for example, pixel S
, and the densities of the small pixels α to δ obtained by dividing the pixel S of interest into four are newly determined in consideration of the densities of the pixels adjacent to the pixel S. The density of each small pixel is determined, for example, as follows. α={2S+(A+B)}/4 β={2S+(B+C)}/4 γ={2S+(C+D)}/4 δ={2S+(D+A)}/4 In other words, the density of the small pixel α is the sampling pixel A,
It is determined by considering the density of B, and similarly small pixels β, γ,
Each density of δ is the sampling pixel B, C, respectively.
Further, it is determined by taking into consideration the respective concentrations of C, D, and D and A. According to the image processing method described above, the density of a small pixel created by dividing a sampling pixel into four is newly calculated based on the density of the sampling pixel to which this small pixel belongs and other sampling pixels adjacent to the sampling pixel. Because it was set as
The original image information, that is, the image information obtained by detecting the density of the sampled pixels, is converted into image information with four times the density. However, the lines drawn on the original picture are very thin, and the density distribution of the thin lines is as shown in Figure 2a (2
Dimensional expression) Within the width of one pixel (for example, about 100μ)
If the density of each detected sampling pixel is, for example, A=S=C=10 B=D=0 as shown in FIG. When the image information is converted, the sampling pixel S is converted into small pixels α, β, γ, δ.
Since the density of α = β = γ = δ = {1 × 2 + (10 + 0) / 4 = 7.5, the density distribution after the conversion is as shown in Figure 2c, for the sampling pixels A, S, The density of each small pixel of C becomes uniform (7.5), resulting in a prominent distribution in that part. As a result, even if the width of the thin line drawn on the original image (the half-width in the graph of Figure 2a) is narrower than the width of the original sampling pixel, for example, half, it will be In the converted image, the width was never narrower than the original sampling pixel width (see FIG. 2d). [Object of the invention] The present invention has been made in view of the above, and
The object of the present invention is to provide an image processing method that can obtain high pixel density image information that more faithfully reproduces a thin line drawn on an original image from image information that detects the density of a thin line drawn on an original image at a low pixel density. . [Structure of the Invention] In order to achieve the above object, the present invention detects the multi-value level density for each pixel in each block divided into multiple vertical and horizontal blocks, and uses the image information on the pixel as an original pixel to obtain four times the density. When converting to density image information, 2 pixels are added vertically and horizontally to the original pixel width on the original pixel.
A new pixel whose position can be moved up to division 1 is set, and the new pixel is formed by four subpixels, and the density of the subpixel located across the plurality of original pixels is determined by The density is determined by the arithmetic average of the densities of multiple original pixels that the small pixel spans, and the density of the small pixel located within one pixel of the original pixel is determined by the density of the original pixel and adjacent original pixels. In the area including the original pixel, calculate the density gradient in each direction by calculating the density difference with the original pixel. At a concentration of
or b: when the density gradient has an extreme value in one direction and monotonically increases or decreases in the other direction, the density is corrected by the density difference in each direction from the original pixel density in the direction having the extreme value, or c: When the concentration gradient has a convex value in one direction and a concave value in the other direction, the concentration is corrected by the concentration difference from the one direction, or d: The concentration gradient has a convex value in either direction. Sometimes, the respective densities corrected by the density difference in each direction are compared and the larger density is used; or e: When the concentration gradient has a concave value in either direction, the density in each direction is The respective densities corrected by the difference are compared, and the converted image information determined respectively is used for the smaller density. [Embodiments of the Invention] Hereinafter, embodiments of the present invention will be described based on the drawings. FIG. 3 is an explanatory diagram showing an example of the image processing method according to the present invention. In the same figure, a11 to a13, a21 to a2
3, a31 to a33 (within the frame indicated by the broken line)
are sampling pixels predetermined for the original image, and the density of the original image is detected in units of these sampling pixels. Here, on the sampling pixel surface, a quarter (l/
4) Set new pixels shifted by the ratio, and divide these new pixels into four. Then, the small pixels formed by this quartering are classified into the following four types. (1) Small pixel A The one located at the center of the original sampling pixel. (2) Small pixel B A small pixel located at the same ratio between the original sampling pixel and the sampling pixel adjacent below this sampling pixel. (3) Small pixel C A small pixel located at the same ratio between the original sampling pixel and the sampling pixel adjacent to the right side of this sampling pixel. (4) Small pixel D: A small pixel located at the same ratio between the original sampling pixel and the adjacent sampling pixels below, to the right, and to the lower right of this sampling pixel. Looking at the new pixel shifted from the original sampling pixel a22, among the four types of small pixels mentioned above, each density of B, C, and D is B = (a22 + a22) / 2 C = (a22 + a22) /2 D=(a22+a32+a33+a23)/4 Determine according to each algorithm. Furthermore, in order to emphasize the density of the small pixel A, its density is determined, for example, by an algorithm as shown in FIG. Therefore, the density determination algorithm (density determination of the small pixel A located at the center of the sampling pixel a22) shown in FIG. 4 will be explained. If the density distribution in the main scanning direction (horizontal direction in the drawing) centering on sampling pixel a22 is a22>a21, a23, and the density distribution in the sub-scanning direction (vertical direction in the drawing) is a22a>a21, a32, That is, when the density of sampling pixel a22 is outstanding compared to the density of surrounding sampling pixels, A = a22 + (a22 - a12) + (a22 - a32) = 3 (a22) - a12 - a32 and A = a22 + (a22 - a21) + (a22 - a23) = 3 (a22) - a21 - a23, whichever is larger. When the density distribution in the main scanning direction is a22<a21, a23 and the density distribution in the sub-scanning direction is a22<a12, a32, that is, when the density of sampling pixel a22 is smaller than the density of the surrounding sampling pixels, A = a22 + (a22 - a12) + (a22 - a32) = 3 (a22) - a12 - a32 and A = a22 + (a22 - a21) + (a22 - a23 = 3 (a22) - a21 - a23, whichever is smaller If the density distribution in the main scanning direction is a21≦a22≦a23 and the density distribution in the sub-scanning direction is a12≦a22≦a32, that is, in the main scanning direction, from the left side to the right side centering on sampling pixel a22. In the case of a density distribution in which the density increases sequentially from the top to the bottom with sampling pixel a22 at the center in the sub-scanning direction, A=a22 is determined.Density distribution in the main-scanning direction If a12≧a22≧a32, that is, in the main scanning direction, the density distribution decreases gradually to the left with sampling pixel a22 as the center, and in the sub-scanning direction, the density distribution decreases from the top to the bottom with sampling pixel a22 as the center. In the case of a density distribution in which the density decreases sequentially, A=a22 is determined.Hereafter, in the case where the density distribution in the main scanning direction is a22>a21, a23 a22<a22, a23 a21≦a22≦a23 a21≧a22≧a23, , a combination of cases in which the density distribution in the sub-scanning direction is a22>a12, a32 a22<a12, a32 a22≦a22≦a32 a12≧a22≧a32, and the case in which the density is not determined as described above. The density is determined based on each algorithm of A=3(a22)-a12-a32, A=3(a22)-a21-a23, or A=a22.Next, the predetermined sampling pixel plane An example of a device that determines the density of a newly set small pixel based on the algorithm described above and obtains high-density image information will be explained below. Fig. 5 shows the basic configuration of the device. This is a block diagram in which a multilevel digital video signal obtained by optically scanning an original image, photoelectrically converting the analog video signal, and further A/D converting it is stored in a video memory 10. The input video signals are sequentially inputted and expanded into a 3×3 matrix (corresponding to sampling pixels) on a video memory 10 via shift registers 11 and 12. And video memory 10
The density data (4-bit data) of each sampling pixel developed above, that is, a11 to a13, a2
1 to a23 and a31 to a33 are input to an image processing circuit 20 which performs processing according to the algorithm described above. Here, the details of the image processing circuit 20 are as shown in FIG. 6, for example. In the same figure, 2
1a is a ROM with 12 bits of input and 6 bits of output; this ROM 21a receives 4-bit density data of a12, a22, and a32 on the video memory 10, and 4-bit data of 3(a22)-a12-a32. 2-bit code data is output. This 2-bit code data indicates the density distribution state in the sub-scanning direction centered on a22, and has the relationship as shown in Table 1.

[Table] 21b also has a 12-bit input and a 6-bit output.
ROM, and ROM21b is video memory 1
4-bit density data of a21, a22, a2 and 3 above 0 are input, and 4-bit data of 3(a22)-a21-a23 and 2-bit code data are output. This 2-bit data indicates the concentration density state in the main scanning direction centered on a22, and has a relationship as shown in Table 2.

[Table] 22a is an arithmetic circuit, and this arithmetic circuit 22a
is designed to input 4-bit data a22 and a32 and output (a22+a32)/2, and 22b is also an arithmetic circuit, and this arithmetic circuit 22b inputs 4-bit data a22 and a23 and outputs (a22+a32)/2. It is designed to output a22+a23)/2. Furthermore, 23 is also an arithmetic circuit, and this arithmetic circuit 23 is designed to input 4-bit data a22, a23, a32, and a33 and output (a22+a23+a32+a33)/4. 24 is a ROM with 16 bits of input and 4 bits of output;
4 is the 12-bit data (6 bits each) from the ROM 21a and ROM 21b and the video memory 10.
The 4-bit data of a22 above is input, and the 4-bit data based on the algorithm shown in FIG. 4 is output. 25a, 25
b, 25c, and 25d are registers, respectively;
The register 25a stores data from the ROM 24, the register 25b stores data from the arithmetic circuit 22a, the register 25c stores data from the arithmetic circuit 22b, and the register 25d stores data from the arithmetic circuit 23 in synchronization with the same clock C. It's getting old. 26 is a switching circuit for two systems, and this switching circuit 26 switches the output from the register 25a or the register 25c for one system.
Regarding the other system, the output from the register 25b or the register 25d is switched in synchronization with the switching signal S, each having a cycle that is one-half of the main scanning direction clock. 27a is a switching circuit;
28a and 28b are line buffers having a capacity for one line of a newly created video signal, and the switching circuit 27 transfers data from the one system of the switching circuit 26 to the line buffer 28a.
Alternatively, the signals are distributed in synchronization with the sub-scanning direction clock 28b. Further, 27b is a switching circuit that operates in synchronization with the clock in the sub-scanning direction like the switching circuit 27a, and 28c and 28d are line buffers having the same capacity as the line buffers 28a and 28b, which are connected to the other system of the switching circuit 26. The switching circuit 27b distributes the data from the line buffer 28c or 28d to the line buffer 28c or 28d. 29 is line buffer 28a, 2
This is a write circuit that sequentially writes data output from 8b, 28c, and 28d into a video memory (an area different from the video memory 10) as new image data. Next, the operation will be explained. Video memory 10 and shift register 11,1
With density data stored in register 2, new sampling data is written to video memory 10 each time the main scanning direction clock is output, and data is transferred from ROM 24 in image processing device 20 to register 2.
5a, data is stored from the arithmetic circuit 22a to the register 25b, from the arithmetic circuit 22b to the register 25c, and from the arithmetic circuit 23 to the register 25d. At this time, the data stored in the register 25a corresponds to the density of the small pixel A in FIG. Store the corresponding density data. Then, by the operation of the switching circuit 26 in synchronization with the switching signal S, the data of the register 25a or the register 25c is output alternately from one system of the switching circuit 26, and these data are sequentially sent to the line buffer 28a via the switching circuit 27a. It will be stored. That is, the data stored in the line buffer 28a via the switching circuit 27a becomes the density data corresponding to the small pixel A or the small pixel C in FIG. 3, as shown in FIG. In addition, the switching circuit 26
register 25b or register 2 from the other system of
5d data are output alternately, and these data are sequentially transferred to the line buffer 2 via the switching circuit 27b.
It will be stored in 8c. That is, the switching circuit 27b
The data stored in the line buffer 28c via the line buffer 28c becomes density data corresponding to the small pixel B or the small pixel D (see FIG. 7). When scanning for one line is completed as described above, the line buffer 28a
The density data corresponding to one line of small pixels A and C is stored in the line buffer 28c, and the density data corresponding to one line of small pixels A and C is stored in the line buffer 28c.
Density data corresponding to small pixels B and D for a line is stored. Then, the switching circuits 27a and 27b are switched in synchronization with the clock in the sub-scanning direction, which is output at the start of scanning of the next line, and the image data corresponding to the small pixels A and C similar to the above is transmitted via the switching circuit 27a. line battle 28b
The density data corresponding to the small pixels B and D is stored in the line buffer 28d via the switching circuit 27b. At this time, as shown in FIG. 8, when the density data is stored in the line buffers 28b and 28d, the density data stored in the line buffers 28a and 28C during the previous main scan are sequentially read out, and these densities are Data is written into the video memory via the write circuit 29 in a predetermined arrangement. When the above operation is repeated until the scanning of the original image is completed, the density data of each small pixel A, B, C, and D is developed on the video memory, and image information with four times the density is generated. can get. If you perform image processing as described above,
The line drawn on the original image is very thin, and the density distribution of the thin line falls within the width of one pixel as shown in Figure 9a (corresponding to Figure 2a), and each density of the detected sampling pixel is However, as shown in FIG. 9b, for example, when a11=a21−a31=0 a12=a22=a32=10 a13=a23=a33=0, each sampling pixel a12,
The small pixels A, B, C, and D that make up the newly set pixel by shifting a22 and a32 are all A=3=(a22)−a21−a23=30 B=(a22+a32)/2=10 C= (a22+a23)/2=5 D=(a22+a32+a33+a23)/4=5 Therefore, the density distribution after the conversion is as shown in Figure 9c, from the original sampling pixel a12,
Density 15 protruding in the horizontal center of a22 and a32
(In this embodiment, the converted density is represented by 4 bits, so the density higher than 15 is limited to 15) and small pixels of density 10 are arranged alternately. As a result, in an image that has been binarized (by appropriately determining the threshold level) based on this converted image information, the thin line can be reproduced with half the width (see Figure 9 d). ). In addition, in this embodiment, the density of the small pixel located at the center of the original sampling pixel is determined based on the density of the sampling pixel and each sampling pixel adjacent to this in the upper, lower, left and right directions. For example, A=9(a22)-(a11+a12+a13+a21+a23+a31+a32+a33) may be determined by further considering the density of sampling pixels adjacent in the diagonal direction. Furthermore, in this embodiment, the density enhancement of the small pixel A located at the center of the original sampling pixel is determined by considering the density distribution in the vertical and horizontal directions of the sampling pixel. It may be determined by considering the density distribution, and furthermore, the small pixel A may not be emphasized and A=a22 may always be set. Furthermore, in this embodiment, a new pixel is set on a predetermined sampling pixel plane, with each sampling pixel shifted by a quarter (l/4) in the vertical and horizontal directions. The invention is not limited to this, but for example, as shown in FIG.
A new pixel shifted at the ratio of 3) may be set, and the shift ratio may be any value as long as it is less than one-half (l/2). Here, when shifting by one-third as shown in FIG. 10, the density of small pixels A, B, C, and D obtained by dividing the newly set pixel into four will be C and D are determined by performing weight calculations that take into account the ratio of the original sampling pixels to which each small pixel belongs, and for small pixel A, the bias within the sampling pixel to which this small pixel A belongs is taken into consideration. It is better to set it as follows. [Effects of the Invention] As explained above, according to the present invention,
When detecting the multilevel density for each pixel in each block divided into multiple vertical and horizontal blocks, and converting the image information on the pixel into image information with four times the density as the original pixel, In contrast, a new pixel whose position can be moved by half in both the vertical and horizontal directions is set, and the new pixel is formed by four subpixels, and the corrected pixel density is calculated using the following steps:
Regarding the density of a small pixel located across multiple original pixels, the density calculated by the arithmetic average of the densities of the multiple original pixels that the relevant small pixel straddles is determined, and the density of one pixel of the original pixel is determined. Regarding the density of a small pixel located within the area, the density difference with the original pixel is calculated within the area including the original pixel and the adjacent original pixel, the density gradient in each direction is calculated, and the small pixel is Since a correction value calculation formula has been prepared according to the density gradient conditions of the original pixels to be sandwiched, the density of surrounding pixels and its density gradient can be determined from image information that detects density at low pixel density for thin lines drawn on the original image. By correcting the direction-dependent density difference, it is now possible to obtain high pixel density image information that more faithfully reproduces the thin line.Furthermore, even when the pixels are enlarged, images with high resolution can be expressed, making it easier to process. It becomes possible to realize an image processing device that can more faithfully reproduce the original image without reducing speed.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram showing the state of pixel division in a conventional image processing method, FIG. 2 is an explanatory diagram when thin lines drawn on an original image are processed by the conventional method, and FIG. 3 is an image according to the present invention. An explanatory diagram showing an example of the state of pixel division in the processing method, FIG. 4 is an explanatory diagram showing an algorithm for determining the density of the small pixel A in FIG. 3, and FIG. 5 is an explanatory diagram showing an algorithm for determining the density of the small pixel A in FIG. 6 is a block diagram showing the details of the image processing circuit in FIG. 5. FIG. 7 is a timing chart showing the output states of the switching circuits 27a and 27b in FIG. FIG. 8 is a timing chart showing the operating states of the line buffers 28a, 28b, 28c, and 28d in FIG. 6, and FIG. 9 shows an example of the state when thin lines drawn on an original image are processed by the image processing method according to the present invention. FIG. 10 is an explanatory diagram showing pixel divisions in another embodiment of the image processing method according to the present invention. 21a, 21b, 24...ROM, 22a, 2
2b, 23... Arithmetic circuit, 25a, 25b, 25
c, 25d...Register, 26a, 27a, 27
b...Switching circuit, 28a, 28b, 28c, 2
8d...Line buffer, 29...Writing circuit.

Claims (1)

[Scope of Claims] 1. When detecting the multi-level density of each pixel in each block divided into a plurality of vertical and horizontal blocks, and converting the image information on the pixel into image information with four times the density as the original pixel, A new pixel is set on the original pixel whose position can be moved up to 1/2 of the width of the original pixel in both the vertical and horizontal directions, and the new pixel is formed by four subpixels, and the pixel is divided into four. The density of the small pixel located astride is determined by the density calculated by the arithmetic average of the densities of the plurality of original pixels that the small pixel straddles, and the density of the small pixel located within one pixel of the original pixel is determined. Regarding the density, calculate the density difference with the original pixel in the area including the original pixel and the adjacent original pixel, and calculate the density gradient in each direction, a: density gradient in either direction. when the concentration gradient is monotonically increasing or decreasing, or b: when the concentration gradient has an extreme value in one direction and monotonically increasing or decreasing in the other direction,
or c: When the density gradient has a convex value in one direction and a concave value in the other direction, At the density corrected by the density difference, or d: When the density gradient has a convex value in any direction, the density corrected by the density difference in each direction is compared and the density is corrected by the larger density, or e: When the density gradient has a concave value in either direction, the respective densities corrected by the density difference in each direction are compared, and the determined converted image information is used for the smaller density. An image processing method characterized by: