JPH0488748A - Isolated point elimination device - Google Patents

Abstract

PURPOSE:To prevent deterioration in picture quality by clearing an error data stored in an area in which a change in a picture element density is discriminated to be small and the picture element density is near a minimum value. CONSTITUTION:A peak hold detection section 10 holds a maximum value of a difference in a picture element unit in an error data calculated by a subtractor 6. An isolated point discrimination section 11 receives an output binary data being an output of a comparator 8, a maximum value of an error data stored in the detection section 10 and a prescribed threshold level T' respectively. Then the discrimination section 11 recognizes a maximum value of a difference from the detection section 10 when a black level picture element is caused from an output from a comparator 8, and corrects the output binary the prescribed threshold level T' (equal to T in this embodiment) and the maximum value of the difference is smaller than the prescribed threshold level T' as the result. Moreover, a reset signal is generated in addition to the processing above to reset a buffer memory 4. Then the error data is discriminated in excess of a half of the gradation number at first in a low density area by applying error diffusion processing to the error data of a noted picture element 25.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an isolated point removal device that can be applied to an image processing device that binarizes a monochrome halftone image, such as an image scanner or a facsimile machine.

"Prior Art" In image processing devices such as image scanners and facsimile machines that binarize monochrome halftone images, "
Dither method J (! Journal of the Institute of Image Electronics Engineers, Vol. 10, No. 5 (
A systematic dither method is known as shown in (1981)).

This is a well-known technique, but it has the disadvantage that when you try to increase the number of pseudo gradations, the resolution of text and line figures deteriorates, so text and line figures and halftone images are mixed. Regarding manuscripts, it is hard to say that this is a good solution.

As a method to eliminate this drawback, “An Adapti
ye Alogorithm for 5patial
Gray 5cale, (SID 75D engineering GES
The error diffusion method shown in T) is known.

This is the reading density and display density (white or black) of the original image.
This is a method that attempts to reduce the average error between
Use an error filter to detect binarization errors that occur around the pixel of interest!
It is designed to correct halftone images by detecting
The processes shown in 1) to (4) are performed.

fan=fmn+(1/Σakl)×Σakl
em-k n-1...(1) f an ” ≧
T gmn=1 (black) ・ −・ (2)
fan”<T gmn=o (white)---(3)s
mn=f mn--gmn ---(4)
Here, f: input multi-value data, f*n': correction value 8-n: output binary data, akl=weighting is a coefficient, e*n: binarization error, T: m value.

Further, for weighting, the coefficient α is set to 1, and the pixel of interest is marked with an asterisk.

"Problem to be Solved by the Invention" By the way, when converting multivalued data (halftone image) into binarization using the error diffusion method, when the pixel density changes in an area that is not near the maximum value or the minimum value, Good characteristics are shown in areas where the pixel density changes significantly.

However, for areas where the pixel density is close to the minimum value and where the change in pixel density is small, the number of black pixels allocated to them is extremely small, so they can be recognized as isolated points within the range that the human eye can recognize. .

Particularly in the case of facsimile machines, isolated dots (black pixels) in low-density areas not only cause deterioration of the output printed document but also lead to an unnecessary increase in the amount of code during run-length encoding.

Therefore, the present invention solves these conventional problems and proposes an isolated point removal device that can easily remove isolated points.

"Means for Solving the Problem" In order to solve the above-mentioned problem, the present invention creates a pseudo An isolated point removal device using an error diffusion method that expresses halftones includes a means for binarizing a pixel of interest, a means for calculating the amount of change in error in pixel units, and a means for processing input multilevel data and binarization. a peak hold detection unit that detects the maximum value of error data generated from the output binary data, and determines an isolated point of a pixel from the a force binary data, the maximum value of the error data, and a predetermined threshold value. and an isolated point determination section, in which the maximum error value of each pixel on the scanning line including the pixel of interest is smaller than the predetermined threshold when the pixel of interest changes from white to black. When it is determined that This method is characterized in that all error data in a range where the error data has a positive value is corrected to 0 in the opposite direction.

"Function" The error between the level (density) of monochrome input multi-value data and the level (display density) of output binary data is small and f! In the error diffusion method that expresses halftones in a 1-like manner, when determining the binarized output using equations (2) and (3), 1
Check whether the pixel is the same as the previous binary data.

Next, focusing on the error data eIn given by equation (4), the maximum value of the amount of change in the error data is maintained, and if there is a change from white to black, the data is below a certain threshold T'. Moreover, when it is determined that the pixel density is near the minimum value and the change in pixel density is small, the output binary data is inverted to white.

This will remove isolated points.

Then, the range in which the error data has a positive value from one pixel above the pixel of interest (pixel at the point of change) in the main scanning line direction, and the range in which error data has a positive value in the direction opposite to the main scanning line direction from the pixel of interest (pixel at the point of change). All error data in a range with positive values is set to 0.

By doing so, the error data accumulated up to the point of change in the scanning line can be eliminated.

By repeating this process, it is possible to prevent isolated black pixel points from occurring until high-density data is received.

Embodiment Next, a case in which an example of the isolated point removal device according to the present invention is applied to the above-mentioned image processing device will be described in detail with reference to FIG. 1 and subsequent figures.

FIG. 1 is a block diagram showing an example of an isolated point removal device.

In the figure, reference numeral 1 denotes an input terminal for multi-gradation original image data (256 gradations here), that is, multi-value data, which can be input from an image scanner or the like. FIG. 2A shows an example of the input multilevel data.

The pixel marked * is the pixel of interest.

This input multivalued data is subjected to addition of the product-sum calculation output of the error data by the adder 5, and then is set to a predetermined threshold value T in the comparison N8.
(in this example, 128 levels). The output binary data, which is the comparison output, is supplied to the output terminal 2 after being subjected to a predetermined determination process in an isolated point determination section 11, which will be described later.

Figure 2 shows an example of output binary data. This corresponds to FIG. 2A.

Output 2 from comparison N8 with the input multi-value data that has passed through addition M5
The value data is supplied to a subtractor 116, where subtraction processing is performed and error data is calculated from this. This error data is supplied to the peak hold detection section 10, and is also supplied to the buffer memory 4 and stored at the address corresponding to the pixel.

An example is shown in FIG. 2C. In this example, error data A, B, C, and D for four pixels centered on the pixel of interest out of two lines of buffer memory are shown.

The error data that has been completely read from the buffer memory 4 is subjected to a sum-of-products operation N7, in which the weighting output from the coefficient generating section 3 is subjected to a sum-of-products process with the coefficient α.

This product-sum output is supplied to the adder 5 as correction data and added to the input multi-value data. This is to weight and correct the binarization error that occurs around the pixel of interest.

The second diagram shows an example of weighting coefficients. In this example, the weighting of one pixel before the pixel of interest is a coefficient αA,
The weighting for the pixel one line before the pixel of interest is the coefficient αB.
Correction data is formed using a total of four weighting coefficients, ie, weighting coefficients αC0αD for the pixels before and after 1.

Now, in this invention, as described above, the peak hold detection section 10 and the isolated point determination section 11 are newly provided.

In the peak hold detection unit 1o, the maximum value of the difference in pixel units in the error data completely calculated by the subtracter 6 is held.

The isolated point determination section 11 is fully supplied with the output binary data that is the output of the comparison N8, the maximum value of error data that can be stored in the peak hold detection section 10, and a predetermined threshold value T', respectively.

Then, when a black pixel occurs in the output from the comparison N8, the isolated point determination unit 11 recognizes the maximum value of the difference from the peak hold detection unit 10, and sets the data to a predetermined threshold T- (which is equal to T in this example). ), and if the maximum value of the difference is smaller than a certain threshold T', the output binary data is modified from black to white.

Furthermore, in addition to this processing, a reset signal is generated, and the buffer memory 4 is reset with this.

In the low density area shown in FIG. 3, zl is the area where the a-value processing has been completed, 22 is the area where binarization has not been processed, and 25 is the pixel (target pixel) that is currently being binarized. . Then, by performing the error diffusion process, it is assumed that the error data of the pixel of interest 25 exceeds a predetermined threshold value T' (usually half the number of gradations) for the first time in the low density region.

In this state, since error data close to the threshold T' is concentrated in the vicinity of the pixel of interest 25, that is, on the right side of the regions 23 and 24, black pixels are again placed below the regions 23 and 24. is more likely to occur. Therefore,
When the isolated point determination unit 11 determines that the area is a low concentration area, it sends a reset signal to the buffer memory υ4 to clear the error data.

This repeats a series of processes in which all accumulated errors up to the pixel of interest 25 are eliminated.

A black pixel occurs when data with a density change is input during this -Ryo repeating process.

Buffer memory 4 reset by reset signal
As shown in Figure 3, the upper area is an area where the error data has a positive value in the main scanning line direction from the pixel one line before the target pixel 25 (the point of change from white to black). All 23 error data are set to O. Similarly, all the error data in the area 24 where the error data has a positive value in the direction opposite to the main scanning line direction from the pixel of interest 25 is set to 0.

FIG. 4 is a flowchart showing a specific example of isolated point removal processing according to the present invention.

First, one line of image data (input multivalued data) fn is input, and the maximum value P14AX held in the peak hold detection section 10 is reset (step 31.32).

Next, correction data is calculated by adding the input multivalued data and the product-sum calculation output E of the error data (fn''=fn+E) (step 33).

Then, the correction data fn- is compared with the threshold T (step 34), and when the correction data fn' is smaller, the output binary data gn is set to ○ (logical zero) and the error data en is calculated. (steps 35 and 36).

After that, this error data en and the error data e of one pixel before
The difference Δen from n-1 is compared with the maximum value PMAX of the error data, and the larger one is stored as the maximum value P)IAX (step 37).

This discrimination process continues until one line is completed (steps 38, 33 to 37).

On the other hand, if it is determined in step 34 that the corrected data fn'' is more effective, the data of the difference between the input multilevel data and the output binary data is stored as error data.
1), then check the change in output between the previous pixel and the pixel of interest, and if the output content of the previous pixel was the same, that is, it was a black pixel, the output of this pixel of interest will be the same. The binary data is set to logic 1 (black) (steps 42 and 43).

Such processing is continued until one line is completed (step 38).

On the other hand, when it is determined in step 42 that the output is different from the previous pixel, the pixel of interest is a pixel at a change point, so in this case, peak hold detection 1
Maximum value PH8X and threshold value T held in M110
' is determined (step 44).

If the maximum value PMAXO is larger, the process proceeds to step 43; if not, it is determined that the pixel of interest is an isolated point, and the output binary data at that time is set to logic 0 (=white).
(Step 45). And area 23.24
After clearing the error data, the process moves to step 38.

Such processing is performed until the last line.

"Effects of the Invention" As is clear from the above explanation, the processing according to the present invention is incorporated into an error diffusion algorithm that binarizes a monochrome multilevel image and expresses halftones in a pseudo manner. It is.

According to this, it is possible to eliminate isolated points caused by black pixels by resetting the error data accumulated during installation when the pixel density is near the minimum value and the change in pixel density is determined to be small. It becomes possible.

Therefore, deterioration in image quality caused by the presence of isolated points can be prevented, and when applied to facsimile machines, unnecessary black pixels can be eliminated, resulting in a significant reduction in the code amount of run-length encoding.

8・ 10・ 11・ ・Product-sum calculator ・Comparison chicks ・Peak hold detection section ・Isolated point determination section

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an example of an isolated point removal device according to the present invention, Sections 2 and 3 are explanatory diagrams of isolated point removal,
FIG. 4 is a flowchart illustrating an example of isolated point removal processing.

Claims (1)

[Claims] (1) We focused on an isolated point removal device that utilizes an error diffusion method that pseudo-expresses halftones by reducing the error between the reading density of a multivalued original image and the binarized display density on average. pixel binarization processing means, means for calculating the amount of change in error in pixel units, and detecting the maximum value of error data generated from input multi-value data and binarized output binary data. It is composed of a peak hold detection section, and an isolated point determination section that determines an isolated point of a pixel from the output binary data, the maximum value of the error data, and a predetermined threshold. When the pixel of interest changes from black to black, and it is determined that the maximum error value of each pixel on the scanning line including the pixel of interest is smaller than the predetermined threshold, the output binary data is inverted to white, and the pixel of interest is All the error data in the range in which the error data has a positive value from one pixel above in the main scanning line direction, and in the range in which the error data has a positive value in the direction opposite to the main scanning line direction from the pixel of interest. An isolated point removal device characterized in that the isolated point is corrected to zero.