JPS5825767A - Picture processing device - Google Patents

Abstract

PURPOSE:To improve the gradation while keeping resolution, by estimating density values of corresponding picture elements of an original dither picture corresponding to respective noticed picture elements in an expanded/reduced dither picture and assuming these estimated density values as density values of respective noticed picture elements. CONSTITUTION:After an original dither picture S(X,Y) is stored in an original dither picture memory 1, a coordinate counter 51 in a controlling circuit 5 starts outputting successively coordinates of noticed picture elements in an expanded/ reduced dither picture. A dither position detecting circuit 53 in the controlling circuit 5 detects dither positions L1-L16 from coordinates (X,Y) of an objective picture element and gives this signal to a density value estimating circuit 2 as 4-bit information. A 9-bit pattern around the objective picture element is given from the original dither picture memory 1 to the density value estimating circuit 2, and an estimated density value is read out from a density value estimate table on a basis of these 13-bit signals and is given to a dither matrix circuit 3, and thus, expanded/reduced binary information of the noticed picture element is generated.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an image processing device for enlarging and reducing a dip image.

One of the methods of enlarging or reducing an image in which each pixel is expressed as binary information by image processing is a method called "Nearest Neighbor". This method uses the pixel position (x,
y) is made to correspond to the pixel position (x, y) of the original screen according to the magnification, and the binary information of the pixel closest to the corresponding position is determined as the binary information of the enlarged/reduced image.

A dithered image is also expressed as binary information, and the image can be enlarged or reduced by the nearist-Napore method, but the halftone information is greatly impaired. That is, in a deep image, the binary information (white, black) of each pixel is determined by a threshold matrix called a deep matrix, and the intermediate tone of the image is expressed in a pseudo manner by the magnitude of the pixel density.

Therefore, the image quality of the enlarged/reduced image varies greatly depending on which threshold value in the dip matrix the pixels of the original image that are associated with the pixels of the enlarged/reduced image correspond to, and as a result, the halftone information is significantly lost. be exposed.

On the other hand, a unit consisting of multiple pixels corresponding to the dip matrix expresses the density value with the resolution of the dither matrix size, so this density value is demodulated and used as a scaled image. Dither matrix again 2
Although it is possible to express halftones by converting into values, in this case, the original resolution of the dip image will be greatly impaired.

The present invention essentially scales the original dithered image using the Nearist-Napore method, and the purpose is to improve the gradation while maintaining the original characteristics of the dithered image such as resolution. The density value of the pixel of interest is calculated based on the pits and turns of multiple pixels in the vicinity of the original dither image, including the corresponding pixel position, and the dip position (position within the dither matrix) of the corresponding pixel position. This was achieved by estimating the density value for each pixel of interest, performing a binary judgment on the estimated density value using a dip matrix of a scaled image prepared in advance, and outputting this as binary scaled information. .

This will be explained in detail below.

The present invention targets a dithered image as an original image.

FIG. 1 is an explanatory diagram of the original dip image, and a small rectangle in the original, dip matrix, and dip image corresponds to one pixel.

In FIG. 1, the numbers attached to the pixels of the document indicate examples of the true density values of each pixel in the document, and the numbers 0 to 15 in the dip matrix DM indicate examples of threshold values, and 0 to 15 in the dip matrix DM indicate examples of threshold values. A binary value of 1 indicates a pit/cutan of the dithered image.

The dither number matrix shown in Figure 1 is
This is one type of 17-gradation display called er) type, and a deep image is obtained by binary-judging the density value of the original using the corresponding threshold value of such a dither matrix.

In the present invention, each pixel position in the enlarged/reduced dip image is associated with one pixel position in the original dip image according to the magnification.

FIG. 2 shows an example of the correspondence of pixel positions with the original screen for a reduced image with a magnification of 1/2 and an enlarged image with a magnification of 2.

In FIG. 2, in the reduced dip image, adjacent pixel positions in the reduced dip image are associated with every other pixel position in the original dip image, and in the enlarged dither image, adjacent 4 pixel positions are grouped into the original dither image. is associated with one pixel position.

In the present invention, the density value of the corresponding pixel in the original dithered image corresponding to each pixel of interest in the enlarged/reduced dithered image is estimated, and this estimated density value is determined as the density value of each pixel of interest.

To estimate the concentration value, store a concentration value estimation table in advance,
pJ? This is done by specifying the concentration estimation clothes in turns.

In this case, the dip position can be associated with the threshold array in the neighboring pit pattern. Figure 3 shows the binary information b5 of the target pixel as the pit gloss turn of the neighboring pixel.
In this example, a pit pattern of 9 pixels including b, -1)s is used, and the dip matrix shown in FIG. 1 is used.

In FIG. 3, if the dip position of the target pixel (corresponding to bs) is high, then the bit pattern b! .

b2r ba r bn r bs *...,
bg dip position is L16#L13 r Ll4 a
L 4 e Ll +...+L6, and the threshold values are 5, 15, 7, 10, 0... ....

4, if the dither position of the target pixel is L4, the bit pattern bl, b, yt13 +b4 +bS
l,... pbg dip position is Ll5 s IJ16
・Ll3・L3・L4+...+L5, the threshold value is 13.5°15.2,10. ..., 12. In this case, the corresponding pixel neighborhood t4 turn is represented by 9 pits, and the threshold array in this neighborhood pattern is at the zoomer position L1%L1. Since the dither positions L4 to Lts can be represented by 4 bits, the density value estimation table is specified and estimated using 13 bits in total.

The following table shows the number of times a density value appears in an actual image when the depth value of the target pixel whose density is to be estimated is Ll.

Note that the sample image in this case is a half-body image of a woman (an image called Girt in the standard image database 5IDBA of the Institute of Industrial Science, the University of Tokyo).

Note that this table is displayed in 16 gradations from 0 to 15 for convenience of data collection.

In this table, B is the neighboring pit pattern b.

~b9 expressed in hexadecimal, G/liO~
The actual density value is up to 15, H is the actual number of occurrences.

B=(000), that is, bl l bl r...
, bs = (00000'000'0)O, the density of the target pixel is only 0 in the actual image.
B=(010), that is, bl rb2 p...
, b9=(ooooiooooo), the density value is most often 3, followed by the density value 211.4.5.
They also appeared in this order. B = (IFF), that is, bl', bl, ..., bl = (1111111
In the case of 11), the concentration was 15 in most cases, but in some cases it was 14.13. Since the neighboring pit pattern BUBS has 9 bits, 512 consecutive numbers and turns are possible, but only 51 consecutive numbers appear in Table 1. In the concentration value estimation table, dither positions L1 to L16 and pit no! 8192 ways (1
In the case of 3-pit correspondence), the number of occurrences of density values as shown in Table 1 is detected for representative images, and the density value that appears most frequently is stored as the estimated density value. For the specified information (Ll to L16sl) 1 to be) that did not appear in the representative image, the corresponding pip) nof turn bt to bs'
One minimum threshold value of a pixel that takes l” (black) and '0” (white)
The average value of the maximum threshold values taken can be taken as the estimated concentration value, and if the pit no amount turns b1 to b9 are all "'0# or all""l# (1): a, respectively The estimated concentration value can be 0 or 15.

In the present invention, the estimated density value of the target pixel estimated in this way is regarded as the density of each pixel of interest in the enlarged/reduced pixel,
This is subjected to binary judgment using a dither matrix of the enlarged/reduced image to obtain an enlarged/reduced dip image.

Figure 4 is an explanatory diagram of a reduced dip image.The numbers attached to the original f4 image indicate the estimated density value of the target pixel, the reduced dip matrix, and the numbers in boxes indicate its threshold value. The numbers in the rive image indicate its binary information.

FIG. 5 is a block diagram showing an embodiment of the present invention,
1 is an original dither image memory, 2 is a density value estimation circuit, 3 is a dip matrix circuit, 4 is an enlarged/reduced dither image memory, and 5 is a control circuit.

In FIG. 5h, after the original dip image s(x, y) is stored in the original f4 image memory 1, the coordinate counter 5 in the control circuit 5 calculates the value of the enlarged/reduced dip image S(x, y). Start sequentially outputting the coordinates of the pixel of interest. In the output state of each target pixel coordinate (x, y), the coordinate calculation circuit 52 in the control circuit 5 calculates the coordinate (x, y) of the target pixel according to the following equation.
y), and the control circuit 5 calculates the pit nochlear turn b! in the vicinity of the target pixel from the original dip image memory 1. ~b9 is read out, and this 9-bit information is output to the density value estimation circuit 2.

In addition, in the above formula, R is the set magnification, []
is a Gaussian symbol, and 0.5 is a bias value for matching coordinates by 5 people to the nearest 400. If this is omitted, the matching will be rounded down to the nearest whole number. The dip position detection circuit 53 in the control circuit 5 detects the dip position L1'=L16 from the coordinates (x, y) of the target pixel.
is detected and provided to the density value estimation circuit 2 as 4-bit information. The concentration value estimation circuit 2 reads the estimated concentration value corresponding to these 13 bits of information from the concentration value estimation table (not shown), and provides it to the matrix circuit 3. Pixel coordinates (x, y)
The control circuit 5 gives the dip position of the scaling dip matrix corresponding to It is output as binary information s (x ry ) of the pixel of interest in the deep image.

The binary information S(x, y) of the pixel of interest is specified by the coordinate counter 51 in the control circuit 5 and written into the enlarged/reduced dither image memory.

As is clear from the above description, in the present invention, the density value is estimated for each pixel of interest, and the density value is re-enlarged/reduced in the third flash procedure correction form (method). Indication of the case 1982 Patent Application No. 1230'93 2, Name of the invention Image processing device 3, Person making the amendment Relationship to the case Patent application Office (105) 1-7 Toranomon, Minato-ku, Tokyo Number 12 Name (029) Okill1g! L-Koli1 Co., Ltd. Representative Director and President San Ketome 4, Agent Address (105) 1-791 Toranomon, Minato-ku, Tokyo
No. 2 5. Date of correction order: January 26, 1980 (shipment date) 6. Subject of correction Z Contents of correction (1) Specification No. 4w, line 6 [A rectangle corresponds to one pixel. '' should be corrected as follows.

"A rectangle corresponds to one pixel. Figure 1 shows, from the left side, the original, dithered, and IJ sock original dip images." (2) Page 5 of the specification, line 3, "Relationship It shows.”
The statement has been corrected as follows. “It shows a relationship.

FIG. 2 shows, in order from the left, an enlarged dithered image, an original dithered image, and a reduced dithered image. (3) On page 6 of the specification, line 2, "The box shown in Figure 1 is used." is replaced with "The box shown in Figure 1 is used. The box shown in Figure 3 is used." , from the left side, the bits/mother turns in the vicinity of the target pixel, the ditherers associated with them), the IJ sock positions, and the threshold values of the dither matrix corresponding to those positions.'' (4) Page 10 of the specification. In line 16, the phrase ``obtain.'' is corrected as follows. "Get. Figure 4 is the left figure 4.

Claims (1)

[Claims] means for determining a corresponding pixel position in the original deep image corresponding to the position of the pixel of interest in the enlarged/reduced deep image according to the magnification; means for detecting t4ter/ and the dip position of the corresponding pixel position, and specifying the estimated density value stored in the density value estimator by the bit pattern and the dither position to determine the pixel position of interest. concentration value estimating means for outputting an estimated concentration value of
The image forming apparatus is characterized by comprising means for performing a binary judgment on the estimated density value using a 7"4 de-matrix that is preset according to an enlargement/reduction ratio and outputting it as binary information of the pixel of interest. Image processing device.