JP2867487B2 - Image processing method - Google Patents

Description

Description: FIELD OF THE INVENTION The present invention relates to an image processing method, and more particularly to a method for binarizing a multi-valued image into a black and white binary image by an error diffusion method. .

(Prior Art) Conventionally, in a digital printer, a digital facsimile, and the like, there is an error diffusion method as a binarization method for expressing halftones in a pseudo manner. Floyd Steinberg: "An Adaptive Algorithm for Spatial Grayscale" SID Preprints, Vol.17 / 2,1976, PP.75-77 (Floy
d and Steinberg; `` An Adaptive Algorithm for Spatia
l Greyscale Proceedings of the SID (Society for
Information Display) Vol.17 / 2, Second Quater1976, P
P.75-77). This method calculates the density difference between the original image density and the output image density for each pixel, assigns a specific weight to peripheral pixels as an error, disperses it, and adds this error to the density of the input value to fix it. The binarization is performed using a threshold value.

(Problems to be Solved by the Invention) However, in the above-described error diffusion method, when the output of a certain pixel becomes black (white), the error is dispersed so that peripheral pixels are made white (black). There is a problem that thin lines such as characters are blurred.

In addition, when pixels having low (high) values are continuously input, errors that are dispersed due to arithmetic errors are truncated, and when the errors start at "0", the errors accumulate to a certain value. The output does not become black (white) until the sum of the input value and the error multiplied by the weight exceeds the threshold (below the threshold). For this reason, the portion where the density of the gradation portion greatly changes to a certain extent when the background has a low input value for white (here, when the input value is low), that is, the edge portion is not very good ( However, a high input value is not good for a black background).

An object of the present invention is to solve such conventional problems,
It is an object of the present invention to provide an image processing method capable of preventing a thin line of a character or the like due to error diffusion processing from being blurred and an edge portion of a gradation at a low input value or a high input value from appearing at a position different from an actual position.

(Means for Solving the Problems) According to an image processing method of the present invention, a binarization process is performed by adding a binarization error of a peripheral pixel to a density of a target pixel. Is compared with a predetermined second threshold, and a value obtained by adding a binarization error of a peripheral pixel to the density of the target pixel is compared with a predetermined second threshold. Is satisfied, a binarization error of the pixel of interest is obtained from the comparison result based on the second threshold value and a predetermined third value.

Further, in the image processing method, when the density of the pixel of interest is a value lower than the first threshold, the third value is set to a preset positive value.

Further, in the image processing method, when the density of the pixel of interest is a value higher than the first threshold value, the third value is set to a preset value of 0 or a negative value.

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

FIG. 1 is a schematic configuration diagram of an image processing apparatus showing one embodiment of the present invention.

In FIG. 1, reference numeral 11 denotes an input unit including a photoelectric conversion unit such as a CCD and a driving unit required for scanning a document, reference numeral 12 denotes an A / D conversion unit for converting analog image data into digital image data, and reference numeral 13 denotes, for example, shoeing correction. Correction unit, 14 is a binarization unit that performs binarization by the characteristic error diffusion method of the present invention,
Reference numeral 15 denotes an output unit including a printer for printing out binary data and its driving system.

 Hereinafter, the operation of FIG. 1 will be described.

First, an original is read by the input unit 11 as image data,
The density value is output for each pixel. The A / D converter 12 converts this image data into 8-bit data of 256 gradations. Further, the correction unit 13 performs correction such as shading correction, and outputs a 4 bit,
The data is output as a signal 100 as data of 16 gradations. The binarization unit 14 determines that the signal 100 is under a predetermined condition (here, “0” or less), and forcibly sets the binarization error at this time to a specific value (here, “5”). The signal is binarized by the error diffusion processing to which the above is added, and is output as a signal 300. The output unit 15 receives the signal 300, temporarily stores data for one document in the frame memory, and then outputs the data as a binary image.

FIG. 2 is a diagram showing a configuration example of the binarization unit 14 in FIG. This includes an adder 21, comparators 22 and 24, a selector 26,
Error calculation unit 28, error diffusion unit 29, thresholds 23 and 25 and fixed error
27.

The detailed operation of the binarization unit 14 is as follows.
The signal 100 that has entered the binarization unit 14 is branched into two, one enters the adder 21 and the other enters the comparator 22. The adder 21 adds the error signal 200 from the error diffusion unit 29 and outputs the result as a signal 101. This signal 101 is split into two, one of which enters the comparator 24. The comparator 24 outputs the signal 300 as “1” if it is larger than the threshold value 25 (here, “7”), and “0” otherwise. One of the signals 101 enters the selector 26. The selector 26 outputs the signal 101 as the signal 103 when the signal 102 is “0”, and outputs the fixed error 27 (here, “5”) as the signal 103 when the signal 102 is “1”. Here, the signal 102 input to the selector 26 is
Output from 2. The comparator 22 receives the signal 100 and outputs “1” as the signal 102 when the input value is equal to or smaller than the threshold value 23 (here, “0”), and outputs “0” otherwise. One of the output signal 103 from the selector 26 and the branched signal 300 enters the error calculator 28. In the error calculation unit 28, the signal 30
When 0 is “1”, a value obtained by subtracting the maximum value (here, “15”) of the signal 100 from the signal 103 is output as the signal 104,
Otherwise, signal 103 is output to error diffusion section 29 as signal 104. The error diffusion unit 29 spreads the signal 104 given as an error to the position as shown in FIG. Further, an error of a pixel which is next input to the binarizing unit with respect to the given target pixel is output to the adder 21 as a signal 200. And this signal 200 is added to the next signal 100,
It is output as a new signal 101 and further output by the comparator 24 as the binarized signal 300 to the output unit 15 in FIG.

A supplementary description of the error calculator 28 will be given below. The signal 103 input to the error calculator 28 is g ₁₀₃ , and the output signal 104 is g ₁₀₄ . Further, in this embodiment, the digital data after the input correction is 16 gradations from "0" to "15", and the maximum value here is "15". Then, when the signal 300 is "1" to enter the error calculation unit 28 is a g ₁₀₄ = g ₁₀₃ -15, when "0" is a g ₁₀₄ = g _103. When the signal 300 is “0”, the signal 104 subtracts “0” (the minimum value of the input data) from the signal 103, but is not described here because it is “0”. Here it is calculated the g ₁₀₄ becomes a binary-coding error.

In the example of FIG. 2, the error diffused when the signal 100 is equal to or smaller than the threshold value 23 is a predetermined value. However, the signal 101 before entering the comparator for binarizing may be the predetermined value. Also, signal 1
When 00 is equal to or smaller than the threshold value 23, the signal 200 from the error diffusion unit 29 may be set to a predetermined value.

Although the threshold value 23 is "0" here, the threshold value may be further increased when the number of gradations increases.

The fixed error 27 also needs to be increased at the same time as the number of gradations increases. However, if the fixed error 27 is too high, a pseudo contour appears, and if it is too low, the effect as in the present embodiment is lost.
As a result of various experiments, a value of 20 to 40% with respect to the number of gradations may be set as the fixed error.

In this embodiment, the case where the background is white has been described.
It is also effective for a white thin line and a gradation portion at a high input value when the background is black. In this case, the threshold value 23 is set to “15”, the comparator 22 sets the signal to “1” when the threshold value is 23 or more, and sets the signal to “0” otherwise. Further, the fixed error 27 is set to “−5”.

FIG. 3 is a diagram showing a configuration example of the error diffusion unit 29 in FIG. These are multipliers 31, 32, 33, 34, adders 35, 36, 37,
The register is composed of registers 38 and 39, a line memory 40, and a divider 41, and performs diffusion of a binarization error of the target pixel and output of an error of the target pixel which is input next.

Next, a detailed operation of the error diffusion unit 29 will be described.

The signal 104 input to this error diffusion unit is branched into four, and the multipliers 31, 32, 33, and 34 respectively multiply the signal by "7", "3",
It is multiplied by “5” and “1”. The line memory 40 outputs a binarization error to the right of the target pixel position. This error is calculated by adder 35
And the sum of the numbers multiplied by the multipliers 31, 32, 33, and 34 in the addition divider 41 and the addition divider 41 (16 = 7 + 3 + 5 +
Divide by 1) and output as signal 200. This signal 200 is output as the error of the next target pixel. The adder 36 is a multiplier
After adding the output from the register 32 and the output from the register 39, the result is stored in the line memory 40 as the lower left error of the pixel of interest. The adder 37 adds the output from the multiplier 33 and the output from the register 38 and outputs the result to the register 39. Register 39
Then, this value is temporarily stored, and this value is output when the next input is made from the adder 37. The output from the multiplier 34 is temporarily stored in the register 38, and this value is output the next time there is an input.

The above is expressed using mathematical expressions. Let g [i, j] be the value of signal 104, i be the abscissa, and j be the ordinate. The output of the line memory 40 at the position on the right of the target pixel is M [i + 1]. The sum of the error diffusion coefficients is “16”. Assuming that the signal 200 is G, the signal G and the value M [i-1] stored again in the line memory 40 are G = (M [i + 1] + g [i, j] × 7) / 16 M [ i−1] = M [i] + g [i, j] × 3.

Also, the values M [i] and M [i stored in the registers 39 and 38
+1] is M [i] = M [i + 1] + g [i, j] × 5 M [i + 1] = g [j, j] × 1

In the example of FIG. 3, the error diffusion by the error diffusion method is shown.
5 (a) and 5 (b), multiplied by the coefficients shown in FIGS. 5 (a) and 5 (b), added up with these errors and then divided by the sum of the coefficients. It may be added to the target pixel.

FIG. 4 is a diagram showing coefficients when multiplying the error diffusion position with respect to the target pixel by a position. In the error diffusion method, a value obtained by dividing the coefficient by the sum of all coefficients is used as a weight, but the sum of all coefficients is added immediately before being added to the input signal 100 as an error in order to reduce truncation in operation. Divided by

The example shown in FIG. 4A is used as the pixel position to be diffused used in this embodiment, but the example shown in FIG. 4B may be used.

6 and 7 are views for explaining the effect of the present embodiment. Figure 6 shows the iieej facsimile test chart N
FIG. 7 is an enlarged view of a character portion of 10.5 points of an image read at 8 dots / mm at o1, and FIG. 7 is an enlarged view of a gradation portion of the iieej facsimile test chart No1. In both figures, (a) is output as a binary image by processing using the conventional error diffusion method, and (b) is processed and output according to the present embodiment.

(Effects of the Invention) As described above, according to the present invention, by performing error diffusion processing provided with a method of setting an error to a predetermined value when an input value is a predetermined value, FIG. A thin line of a character or the like as shown in FIG. 5A appears as shown in FIG. 5B. Similarly, the phenomenon in which the edge portion appears at a low input value with a shift as shown in FIG. 6 (a) can be solved as shown in FIG. 6 (b). Further, it is possible to solve a phenomenon in which an edge portion at a high input value is shifted from a black background.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram of an image processing apparatus showing one embodiment of the present invention, FIG. 2 is a diagram showing a configuration example of a binarization unit in FIG. 1, and FIG. 3 is an error diffusion unit in FIG. FIG. 4 is a diagram showing a coefficient when multiplying the error diffusion position with respect to the target pixel by a position, FIG. 5 is a diagram showing an error extracting position and a coefficient, FIG. 6, FIG. FIG. 4 is a diagram for explaining an effect of the embodiment of the present invention. 11: input section, 12: A / D conversion section, 13: correction section, 14: binarization section, 15: output section, 21: adder, 22, 24: comparator, 23, 25: threshold, 26: Selector, 27: fixed error, 28: error calculation unit, 29: error diffusion unit, 31, 32, 33, 34: multiplier, 35, 36, 37: adder, 38, 39: register, 40: line memory.

──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Kazuhiro Ishikawa 1-7-12 Toranomon, Minato-ku, Tokyo Oki Electric Industry Co., Ltd. (56) References JP-A-1-175367 (JP, A) Hei 1-332765 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) H04N 1/40-1/409 H04N 1/46 H04N 1/60

Claims (3)

(57) [Claims] 1. An image processing method for performing a binarization process by adding a binarization error of a peripheral pixel to a density of a target pixel, comprising: comparing the density of the target pixel with a first predetermined threshold value; Comparing the value obtained by adding the binarization error of the peripheral pixel to the density of the target pixel with a predetermined second threshold value, and when the comparison result with the first threshold value satisfies a predetermined condition, the second threshold value An image processing method characterized in that a binarization error of a pixel of interest is obtained based on a comparison result of the above and a predetermined third value. 2. The image processing apparatus according to claim 1, wherein when the density of the target pixel is lower than the first threshold value, the third value is a predetermined positive value. Method. 3. The apparatus according to claim 1, wherein when the density of the target pixel is higher than the first threshold, the third value is set to 0 or a negative value set in advance. Image processing method.