JPH02285861A - Image processing method - Google Patents

Abstract

PURPOSE:To improve reproducibility at an edge part by deciding a threshold value to binarize a noticed picture element based on the mean value of the binary data of a peripheral picture element whose prescribed range in the neighborhood of the noticed picture element is binarized, and the mean value of multivalue data before binarization. CONSTITUTION:The multivalue data 500 inputted from a image data input part 10 is binarized at a binarization processing part 20, and the binary data 510 is outputted from a image data output part 30. At the binarization processing part 20, mean density m (i, j) and m (i, j) in the neighborhood when the noticed picture element is binarized to '1' or '0' at a threshold value arithmetic part 20b are found by using data B (i, j) after binarization and a weighing mask R (x, y), and the threshold value to binarize the noticed picture element is decided, then, the binarization by a mean density approximation method is performed at a mean density arithmetic processing part 20a.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Field of Application] The present invention relates to an image processing method, particularly for converting multivalued image data to "0",
An image binarization method that converts into binary data of "1", and an image processing method that performs pseudo-halftone processing to minimize the deterioration of resolution information while preserving the gradation information in the original. It is related to.

[Prior Art] An average density approximation method (Information Processing Society of Japan Journal Vol. 1, 1985) has been proposed as a suitable image processing method for binarizing multivalued data and performing pseudo-halftone processing. This method uses binary data that has already been binarized near the pixel of interest, and when the pixel of interest is binarized to "1" or "0", the weighting of both is determined by the neighborhood average value. This is a method of selecting the average value closer to the level of the pixel of interest and binarizing the pixel of interest.

[Problem B that the invention seeks to solve] However, although the average density approximation method is a method that can achieve both image gradation and resolution with a small amount of hardware, it is difficult to reproduce characters with low contrast, thin lines, etc. as shown below. 6 Now, let us consider the case where multilevel data as shown in FIGS. 3A and 3B is to be binarized. Here, the data assumes a value of 4 bits (0 to 15),
Assume that "O" represents white and "15" represents black. FIG. 3B is a numerical representation of FIG. 3A. As shown in FIG. 3A, the pixels are numbered i, j, pixel 201 (i-2,
The pixels before 6th pixel 201, where j-3) is the pixel of interest for binarization, have already been binarized to either "0" or "1" as shown in Fig. 3C, and the average density approximation method Now, for standardized weighting as shown in Fig. 4, a mask is placed over the binary data in Fig. 3C as shown by the dotted line, and the average value no when pixel 202 is binarized to "0" and In the case shown in Figure 6, calculate the average value m when 2i is converted to ``.
-13/23 (40,56).

ml -20723 (40,87), so this value is compared with the normalized value (10/1540.ft7) of the multivalued data ("10" in the figure) of the pixel of interest 20,
The value no that is closer to mo and mo is selected, and the pixel 202 is eventually binarized to "0".

In other words, in the conventional average density approximation method, even though the original image data is as high as 10'', it is binarized to ``0'', resulting in blurred lines and missing white areas.This is due to the density of thin line areas. This occurs because the average value m o + m r of the binarized data becomes higher than that of the original image data because it is binarized to "1" even though it is relatively low.

The present invention provides an image processing method that eliminates the above-mentioned conventional drawbacks, eliminates blurred lines, white spots, etc., and improves reproducibility at edges.

[Means for Solving the Problem] In order to solve this problem, the image processing method of the present invention includes:
In an image processing method that binarizes multivalued pixel data by comparing it with a predetermined threshold value, the average value of the binary data of the surrounding pixels that have already been binarized in a predetermined range around the pixel of interest is calculated, and the Find the average value of the multivalued data before valorization, and based on each of the average values,
A threshold value for binarizing the pixel of interest is determined.

Furthermore, in an image processing method that binarizes multivalued pixel data by comparing it with a predetermined threshold value, based on the binary data of the surrounding pixels that have already been di-sulfurized in a predetermined range around the pixel of interest, the pixel of interest is set to 0 or 0. 1 to 2
Find the average value of each value when converted into values, find the average value of the multivalued data of the surrounding pixels before binarization, and determine a threshold for binarizing the pixel of interest based on each of the average values. do.

Note that the threshold value is calculated as a weighted average of the respective average values.

Furthermore, when the original image data is composed of a plurality of images representing different colors, the coefficients of the weighted average are changed in accordance with the colors.

[Example] The image processing method of this example will be described in detail below with reference to the drawings.

FIG. 1A is a block diagram showing an example of an image processing apparatus that implements the image processing method of this embodiment.

In the figure, multivalued data 500 inputted from the image data input section 10 is binarized by the binarization processing section 20, and binary data 510 is outputted from the image data output section 30. Here, in the binarization processing 20, while the threshold value calculation unit 20b calculates a threshold value to a desired value, the average density calculation processing unit 20a performs binarization using the average density approximation method. Note that the multivalued data input to the image data input section 10 may be data read in by the image receiving section 11 or an image reading section 12 such as a scanner and processed by the image preprocessing section 13. or,
The binary data from the image data output unit 30 may be displayed from the image transmission unit 31 to the transmission 1 image display unit 32 or recorded to the image printing unit 33.

Next, the principle of the image processing method in this embodiment will be explained.

In FIG. 2A, f (t, J) indicates the normalized multi-value data (takes a value of 0 to 1) of the pixel of interest that is now to be binarized, and the pixels above the broken line are already binary data. It is assumed that after the pixel of interest is binarized, the same processing is sequentially performed on f (i+1.j). Similarly, in FIG. 1B, B(i.

j) shows the binary data after binarization, and R(
χ, y) is a standardized weighting mask for determining the average density, and in this example, a 3×2 mask shown in FIG. 4 is used.

In this method, first, similarly to the conventional average density approximation method, the average density m+(i) in the vicinity when the pixel of interest is binarized to 1' or "O".

j), no (i, j), B (t, j), RCx.

y) using the following formula.

However, B (i, j) = 1 However, B (i, j) = 0 Next, the weighted average value m (i, j) of the original multi-valued data f (i, j) is determined by the following equation.

Considering the meaning of m(i, j) here, m(t, j) is the location average value of the original image and is approximately equal to the normalized multivalued data f (i, j) of the pixel of interest, especially If the original image has a gentle gradation change, f (i, j) = m (i, j).

On the other hand, when the gradation changes in the original image are severe, that is, in thin lines and edge parts, f (i, j) ≠ m (i, j), and especially in the rising edge part, m (i, j) (f ( i, j) In the falling part, the inequality m, (i, j) > f (t, j) holds true. Therefore, in order to improve the reproducibility of thin lines and edge parts, m (t, j) The binarization threshold may be set low when <f (i, j), and set high when (i, j)>f (i, j).

Usually, in the average concentration approximation method, the threshold for binarizing f (i, j) is (mo + m,) / 2, and when f (t, j) ≧ (m 0 + m 1 ) / 2, B (t, j) = 1°f When (t, j) < (no + m,)/2, B
(i, j)=O, but in this example, in order to improve the reproducibility of the edge,
This binarization threshold is defined as threshold - (no (1, 1) + 111 (i, j)
)/2-k (f (i, j)-m (1, j))・
...■ However, k is a positive constant, and this is compared with the normalized multivalued data f(1, 4) to determine the value of B (i, j) as follows.

f(i, J)≧ (ago (i, J)+n+ (
l , J))/2-k (f (i, J) - m (
i, j)), then B(t, j)-1...■ f(1, J)<(m.(1, J)◆m+ (i, J))
/2-k(f(1,J)-m(i,j)), then B(
i, J) -0. ...■Here, the formulas ■ and ■ are equivalent to the following formulas ■ and ■, when f(i, j) on the right side is rearranged by multiplying it to the left side.

...■ ...■ Note that the calculation is simplified by setting (no +m,)/2=m'+R(o, O)/2. However, in R(0,0
)=O.

FIG. 5A shows the binarization results obtained by the conventional average density approximation method. FIG. 5B is based on the variable threshold value of this embodiment,
0.0 formula with k set to 1.0. The results shown in FIG. 5B show that, compared to FIG. 5A, the reproducibility of characters is improved, although the reproducibility of halftone portions is the same. If the value of k is increased, character reproduction will further improve, but the smoothness of halftones will deteriorate.

Figure 1B shows an example of the circuit configuration of the binarization processing unit that implements the image processing method of this embodiment. However, the standardized weighting and mask for calculating the average value are shown in Figure 2C. 3
x2.

Standardized multivalued data f input from input line 500
(1, J) is a D flip-flop (hereinafter referred to as D-F/F
) 501 for one clock, and at the same time delay RA
M502 (delays data by one line) and is input to the average density calculation circuit 503, and the average density calculation circuit 503
As for the weighting of f (i, j), the average value 1f (1, J) is calculated according to the formula 0.

The output of the D-F/F 501 is compared in the comparator 504 using the formula
It is output from 10. B(i, J) is also D −F/
B(i-1,j),
B(i-2j-1), B(i,j-1), B(i
◆1. J-1) is manually input to the threshold calculation ROM 505 along with m(i, J) from the average concentration calculation circuit 503. In the threshold calculation ROM 505, the right side of the 0.0 equation may be calculated and stored in advance for these manual efforts.

Note that on the circuit side in Figure 1B, the manual data was explained as standardized multi-value data f(i, j), but when realizing the circuit, the human data uses unstandardized multi-value data. However, it is preferable that the configuration of the threshold calculation ROM 505 corresponds to each of the above equations because it is simpler in terms of design.

In addition, in the case of so-called serial scanning in which binary data is reproduced while reading a document in a strip of a predetermined width, multi-value data f (i, J), binary data B ( 1, J), and each average value is calculated using the following formulas ■′, ■′
, ■′.

However, B (i, j) = 1 However, B (t, j)=.

Furthermore, in this embodiment, the weighting of the pixel of interest is set to the maximum value, but if a method of diffusing the binarization error is combined,
Even when weighting the pixel of interest to 7, the configuration is simpler compared to the conventional error diffusion method, and compared to the conventional average density approximation method, it is possible to significantly reduce the decrease in resolution due to blurring of characters and thin lines. Furthermore, the calculation of the average concentration becomes easy, and the configuration can be simplified.

In the above embodiment, the case where monochromatic data is binarized is shown, but next, the image output device converts yellow (hereinafter referred to as Y), magenta (hereinafter referred to as M), cyan (hereinafter referred to as C), and black (hereinafter referred to as K). ), which can be printed in four colors. In this case, characters and thin lines are often printed in K, and conversely, in Y, M, and C.
Generally, halftone portions are printed frequently.

As described in the above embodiments, increasing the value of k in formulas (1) and (2) reduces blurring of thin lines, but conversely, reproducibility of intermediate tones deteriorates. So, Y.

When M and C are binarized, a threshold corresponding to a small value of k is used to binarize, and when K is binarized, a threshold corresponding to a large value of k is used. By performing binarization, it becomes possible to achieve both a high level of midtone reproducibility and resolution of characters and thin lines as a full-color image.

[Effects of the Invention] As described above, according to the present invention, it is possible to significantly reduce the decrease in resolution due to blurring of characters and thin lines while maintaining the reproducibility of halftones compared to the conventional average density approximation method.

[Brief explanation of drawings]

FIG. 1A is a block diagram showing an example of an image processing device that implements the image processing method of this embodiment, and FIG. 1B is a block diagram showing an example of the configuration of a binarization processing section that implements the image processing method of this embodiment. Figures 2A to 2C are diagrams for explaining the principle of the image processing method of this embodiment. Figures 3A to 3C are diagrams for explaining the shortcomings of the conventional average density approximation method. Fig. 4 shows a mask for calculating the neighborhood average value, Fig. 5A shows an example of the result of binarization using the conventional average density approximation method, and Fig. 5B shows the image processed by the image processing method of this embodiment. FIG. 6A and FIG. 6B are diagrams showing an example of the results, and FIG. 6A is a diagram showing multivalued data or binary data of another embodiment. In the figure, 10...image data input section, 20...binarization processing section, 20a...average density calculation processing section, 20b...
・Threshold calculation unit, 30... Image data output unit, 510°507.508,509...D flip-flop, 5
02.506...Delay memory, 503...Average concentration calculation circuit, 504...Comparator, 505...Threshold value RO
It is M. Figure 2A Figure 2B R (1, 0) = 0 Figure 2C i = I 2 3 4 5 ---- Figure 3A Figure 3B Figure 3C Figure 4 Figure 6A Figure 6B Procedural amendment ( Method) 1. Indication of the case Patent Application No. 1-5880 2. Name of the invention Image processing method 3. Person making the amendment Relationship with the case Patent Hi Canon Co., Ltd. Applicant 4, Agent Toranomon, Minato-ku, Tokyo 105 2-5- July 25, 1999 (shipped) gsA diagram 5B

Claims (4)

[Claims] (1) In an image processing method that binarizes multivalued pixel data by comparing it with a predetermined threshold value, the average value of the binary data of the surrounding pixels that have already been binarized in a predetermined range around the pixel of interest is determined, An image processing method, comprising: obtaining an average value of multivalued data before binarizing pixels, and determining a threshold value for binarizing a pixel of interest based on each of the average values. (2) In an image processing method that binarizes multivalued pixel data by comparing it with a predetermined threshold, the pixel of interest is determined based on the binary data of the surrounding pixels that have already been binarized in a predetermined range around the pixel of interest. 0 or 1 to 2
Find the average value of each value when converted into values, find the average value of the multivalued data of the surrounding pixels before binarization, and determine the threshold value for binarizing the pixel of interest based on each of the average values. An image processing method characterized by: (3) The image processing method according to claim 1 or 2, wherein the threshold value is calculated as a weighted average of each of the average values. (4) The image processing method according to claim 1 or 2, characterized in that when the original image data is composed of a plurality of images representing different colors, the coefficient of the weighted average is changed in accordance with the color. .