JPS6364947B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an image predictive restoration method for restoring an image read at a low pixel density to a high pixel density.

In general, image input/output devices such as digital copying machines and facsimile machines perform coarse sampling on the input side in order to simplify the image reading device and reduce the transmission path capacity of the image signal. to perform low pixel density image reading,
In order to improve the image quality of the image reproduced on the output side, the image signal sent from the input side is processed to predictably restore the image with high pixel density. A major challenge is being able to restore an image close to the original image using binary output on the output side.

Conventionally, as one of this type of predictive image restoration methods, the sampling density of the original image in the main scanning direction is
mpel/mm, the sampling density in the sub-scanning direction is
npel/mm (n<m), and each sampling pixel is divided into two in the sub-scanning direction (m×
2n) When reproducing an image with a high pixel density of pel/mm ² , each sampled pixel information is multivalued quantized, and then binarized for each bin according to an appropriately set threshold level. Based on each pixel information binarized in this manner, the same pixel information is reproduced for each two output lines each consisting of each subpixel divided into two.

Figure 1 shows original image B with low pixel density (m=2n)
This indicates the density level state when each pixel information scanned, sampled, and read is 8-value quantized. Such an input image is converted into binary data at threshold level 3 using the conventional image predictive restoration method described above. When the sampled pixel is divided into two to increase the pixel density and reproduced, the result is as shown in FIG. 2.

As is clear from FIG. 2, in such a conventional image predictive restoration method, the line density in the main scanning direction of the reproduced image is twice the line density at the time of reading.
Even if the number of pixels is doubled, the information of each corresponding small pixel in each of the two output lines is exactly the same, so the resolution does not change at all, and there is a drawback that a high-quality restored image cannot be obtained.

The present invention has been made in consideration of the above points, and
Based on image information read at a low sampling density of (m×n) pel/mm ² , it is possible to restore an image with high pixel density of (m×2n) pel/mm ² with high resolution. The present invention provides an image predictive restoration method.

The image predictive restoration method according to the present invention uses image information obtained by decomposing a document image into pixel units, scanning and sampling, and calculates the density of each pixel of interest to be processed and its surrounding pixels. Focusing on the property that the correlation is very deep, when the ternary quantized pixel of interest is at a midtone level, a certain logic is applied according to the density state in a specific pixel area around the pixel of interest. By making this determination, the density level of each small pixel when the pixel of interest is divided into two in the sub-scanning direction is determined.

Hereinafter, one embodiment of the present invention will be described in detail with reference to the accompanying drawings.

Figure 3a shows a pixel area with a 3 x 3 configuration when a document image is read at a low sampling density of (m x n) pel/ ^mm2 . When D is ₅ , the same figure b
The density levels q ₅₁ and q 52 of each small pixel d ₅₁ and d 52 divided into two in the sub-scanning direction so as to have a high pixel density of (m×2n) pel/mm ² as shown in (m× _2n ) pel/mm ₂ are determined below. Have them do it according to the logical judgment.

Now, when each pixel information read by scanning and sampling the original image is ternary quantized (the white level is 0, the halftone level is 1, and the black level is 2), each pixel D ₁ Letting the concentration levels of ~ _D9 be _Q1 ~ _Q5 , (1) When _Q5 =0, let _q51 = _q52 =0.

(2) When Q ₅ =2, q ₅₁ =q ₅₂ =2.

(3) When Q ₅ =1, if a Q ₁ +Q ₂ +Q ₃ ≧Q ₇ +Q ₃ +Q ₉ , then q ₅₁ =1, q ₅₂ =0.

b If Q ₁ +Q ₂ +Q ₃ <Q ₇ +Q ₃ +Q ₉ , then q ₅₁ =0, q ₅₂ =1.

According to such logical judgment, by sequentially determining the density level of each small pixel when each sampling pixel is divided into two in the sub-scanning direction, a reproduced image with high pixel density and excellent resolution can be predicted. It will be possible to restore it. In addition,
FIG. 4 shows the first image obtained by the image predictive restoration method of the present invention.
This shows the reproduced image when the original image shown in Figure 2 has been processed to increase pixel density, and it is clear that compared to the image reproduced by the conventional method shown in Figure 2, the method according to the present invention can reproduce an image that is closer to the original image. becomes. In addition, in the image predictive restoration method according to the present invention, since it is only necessary to perform three-value quantization on each sampling pixel information when reading the original image, the implementation is easier than in the case of performing eight-value quantization as in the past. It gets easier.

Note that in the logical judgment described above, the pixel of interest
When the density level Q ₅ of D ₅ is halftone level 1, each of the three surrounding pixels adjacent to it in the sub-scanning direction
The density levels q ₅₁ and q ₅₂ of each small pixel d 51 and d ₅₂ are determined by comparing the density level states of Q ₁ to Q ₃ and Q ₇ to Q ₉ , but in the present invention, at least _the pixel of interest A similar determination may be made by comparing the density levels of each of the surrounding pixels Q ₂ and Q ₈ adjacent to each other in the sub-scanning direction of D ₅ .

FIG. 5 shows an example of a configuration for concretely implementing the image predictive restoration method according to the present invention described above.

In the case of this configuration, first, the input pixel signals obtained by scanning and sampling the original image at low pixel density using a scanner are stored in the memory 1, as shown in FIG. 3a. 3×3
Density information for each pixel in a specific region of the configuration is temporarily stored collectively. At the same time, a call command is issued from the address indicator 2 to the memory 1 in response to an address signal given from the outside, and the contents of the target pixel (D ₅ ) to be processed are sent to the quantizer 3 and stored here. At the same time, the contents of the surrounding pixels D ₁ to D ₃ and D ₇ to D ₉ are sent to the arithmetic unit 4, where the sum of the density levels of the surrounding pixels is calculated as appropriate. . Next, the computing unit 4
The calculation result and the output of the quantizer 3 are respectively sent to the logic decision unit 5, where the density level of each subpixel d 51 and d 52 when the pixel of interest D ₅ is divided into two is determined, and the density level of each subpixel d ₅₁ and d ₅₂ is determined. The density level information of the determined small pixels is sequentially sent to the output buffer circuit 6 and temporarily accumulated there, and then, at a predetermined timing, the density level information of each small pixel is sent to an external plotter as an output pixel signal. It is now being sent to

As described above, in the image predictive restoration method according to the present invention, when a document image is read at a low sampling density of (m×n) pel/mm ² , each pixel is divided into two and (m×2n) pel/mm When converting to a high pixel density of ² and performing image reproduction, the pixel of interest is ternary quantized, and especially when the quantization level is a halftone, the density level of the surrounding pixels adjacent to the pixel of interest in the dividing direction is This system is designed to determine the binary density level of each small pixel divided into two according to the state, and has the excellent advantage of being able to reproduce high-quality images with good resolution.

Note that the image predictive restoration method according to the present invention is not limited to the case where pixels sampled by a vertically long aperture are divided into two in the sub-scanning direction to achieve high pixel density processing; Needless to say, the same method can be applied to the case where the pixel density is increased by dividing the pixel into two parts.

[Brief explanation of drawings]

Figure 1 shows an example of image information read by performing 8-value quantization on a document image at low pixel density, and Figure 2 shows an example of image information read out using the conventional method based on that image information. Figure 3a is a diagram showing a sampling pixel area with a 3x3 configuration, Figure 3b is a diagram showing a high pixel density state of the same area, and Figure 4 is a diagram showing the image predictive restoration of the present invention. FIG. 5 is a block diagram showing an example of a configuration for concretely implementing the image predictive restoration method according to the present invention. 1...Memory device, 2...Address indicator, 3...
Quantizer, 4... Arithmetic unit, 5... Logical judge, 6
...Output buffer circuit.

Claims (1)

[Claims] 1. When performing image reproduction in which each pixel obtained by scanning the original image at a low sampling density is divided into two and converted to a high pixel density,
When a pixel of interest is ternary quantized and the quantization level is a halftone, each small pixel divided into two is determined by a logical judgment of the sum of the density levels of three surrounding pixels in the dividing direction of the pixel of interest. An image predictive restoration method that determines the white level and black level of the image.