JPH06301776A - Picture noise elimination device - Google Patents

Abstract

PURPOSE:To surely and easily eliminate picture noise while maintaining source pictures. CONSTITUTION:An address generation part 3 generates the addresses of picture elements in a rectangular mask whose center is a processing object picture element and transfers them to a picture memory 2 and the picture memory 2 outputs data indicated by the addresses. Window extraction parts 4 and 5 read pixel data belonging to plural windows from the picture memory 2 and transfer them to run length counting parts 6 and 7. The run length counting parts 6 and 7 investigate the run length of black picture elements in the window and transfer the control value to a majority calculation part 8. The address generation part 3 transfers the address of the processing object picture element to the picture memory 9 and the majority calculation part 8 performs output to the picture memory 9 as the output of the processing object picture element.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for smoothing an image used in a facsimile or the like, and more particularly to an image noise eliminating device for eliminating isolated pixels in a binary image.

[0002]

2. Description of the Related Art Conventionally, in an image noise eliminator used in a facsimile machine or the like, as shown in FIG. Unwanted isolated pixels as shown at points a and b occur. This isolated point makes the image difficult to read.

The technique for removing isolated points is j × k in the original image.
By taking the majority of the pixels in the local area formed by the mask of (j and k are positive integers) and changing the central pixel to the pixel value determined by the majority, the isolated pixels are removed. There is something I did. As an example, the mask is 3
In the case of × 3, when the pixel in the mask has the pattern shown in FIG. 3, the central pixel is made white by majority voting. After performing the above operation, as shown in FIG. 4, the pixel to be processed is moved to the next, and the same operation is performed for all the pixels.

[0004]

However, by using the conventional image noise removing apparatus as described above, as shown in FIG.
When the image patterns of (b), (c), (d), (e), and (f) are processed, the processing target pixel changes from a black pixel to a white pixel, which causes image deformation. As a result, when the image of FIG. 2 is processed, it becomes like the image of FIG. The pixel indicated by the diagonal line rising to the right is the removed pixel. The points a and b in FIG. 2 are removed, but the length of the horizontal line segment changes from 6 to 4 as in the area c. There are problems that the corners of the image are erased (point d), the original image is deformed (point e is removed, and point f becomes a new black pixel). If the local area is made larger, even thicker lines are erased, and if it is made smaller, the deformation of the image becomes larger.

In view of the above point, the present invention has an object to provide an image noise removing apparatus for easily removing isolated pixels.

[0006]

According to the present invention, a plurality of extracting means for extracting a plurality of windows, each of which is composed of a plurality of pixels including a pixel to be processed, and pixels included in the plurality of window areas are extracted. A plurality of run length counting means for counting run lengths, respective control values obtained from the plurality of run length counting means, and a majority decision means for taking a majority decision from the processing target pixel are provided, and a value obtained from the majority decision means is provided. , And the output value of the pixel to be processed.

[0007]

According to the present invention, with the above configuration, the run length of each pixel included in the plurality of window regions including the pixel to be processed is obtained, each control value obtained from the plurality of run length counting means, and the processing target. By taking a majority vote with the pixels, the image information around the pixels to be processed can be reliably taken in as the processing target, so that the original image in the window can be retained and the isolated pixels can be effectively removed without distortion.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the drawings. FIG. 1 is a block configuration diagram for explaining an embodiment of an image noise removing apparatus according to the present invention.

In FIG. 1, 1 is an input device, 2 is an image memory, 3 is an address generating section, 4 and 5 are window extracting sections as window extracting means, and 6 and 7 are run length counting sections as run length counting means. , 8 is a majority decision calculation unit, 9 is an image memory, and 10 is an output device.

The operation of this embodiment having the above-described structure will be described. First, the image memory 2 is connected to the input device 1.
It stores the data read from. The address generation unit 3 transfers the window address to the image memory 2 and the processing target pixel address to the image memory 9. The window extraction units 4 and 5 extract windows from the mask position referenced by the address generation unit 3.
The run length counting units 6 and 7 are the window extracting units 4 and 5.
The number of consecutive black pixels in windows A and B shown in FIG. 7 extracted in FIG. 7 is checked, and when the number is larger than half the window size, black pixels are calculated, and when the number is small, white pixels are calculated as respective control values. And output. The majority decision calculation unit 8 calculates a majority decision from the control values of the run length counting units 6 and 7 and the pixel to be processed. The image memory 9 stores data to be an output image and sends it to the output device 10, and the output device 10 outputs the data.

Below, a part of the contents of the image memory 2 is shown in FIG.
The case described in 1. will be described. In this embodiment, the size of the window A is 3 × 1, and the size of the window B is 1.
It will be described as × 3. First, the address generation unit 3 operates as shown in FIG.
A size of 3 × centered on the processing target pixel g shown in (a)
The address of the pixel in the mask of 3 is generated, and it is transferred to the image memory 2. The image memory 2 outputs the data pointed to by the address.

The window extracting units 4 and 5 have a shape long in a certain direction, that is, a shape having directionality so that the window area can be specified by a rectangle as shown in FIG. 7, and are orthogonal to each other. Pixel data belonging to is read from the image memory 2 and transferred to the run length counting units 6 and 7. The run length counting units 6 and 7 check the number of consecutive black pixels in the windows A and B passed from the window extracting units 4 and 5. As shown in FIG. 6B, the run-length counting unit 6 determines that the number of consecutive black pixels is 1, and thus the run-length counting unit 7 sets the white pixel as the control value. A white pixel is calculated as a control value based on the result that the number of consecutive black pixels is 1, and the resulting value is transferred to the majority decision calculation unit 8.

The address generator 3 transfers the address of the pixel to be processed g to the image memory 9, and the majority decision calculator 8
Takes a majority decision from the results of the white pixel calculated in window A, the white pixel calculated in window B, and the black pixel of the processing target pixel g, and the white pixel is output as the processing target pixel g.
Output to the image memory 9.

After the above operation, the pixel to be processed is moved to the next, and the same operation is performed for all the pixels. Figure 6
The data content of the image memory 9 obtained from the present embodiment for the image memory 2 having the data shown in FIG. 8 is as shown in FIG. 8, and isolated points indicated by diagonal lines are removed while the original image is retained. There is. The example of the pixel pattern in the mask shown in FIG. 3A is an isolated pixel and should be removed, but FIGS. 3B, 3C, 3D, 3E,
The pixel pattern example shown in (f) is considered to be a part of the image and should not be deformed originally.

In the above-mentioned conventional example, as shown in FIGS.
In the pixel pattern examples shown in (d), (e), and (f), the processing target pixel that is originally a black pixel is output as a white pixel, but in this embodiment, a black pixel can be output. In the present embodiment, since these pixel patterns can be held accurately, isolated pixels can be removed without deforming the original image such as line segments.

In this embodiment, the mask size is 3 × 3, the window size is 3 × 1, 1 × 3,
Although the number of windows is two, it goes without saying that other masks, window sizes, and numbers of windows may be used.

[0017]

As described above, according to the present invention,
By taking a majority decision between the control value obtained from the pixel run length calculation means, which is included in the directional window area centered on the processing target pixel, and the processing target pixel,
Since the pixel to be processed is determined, as a result, the original image can be accurately retained, the isolated pixel can be removed, and the image quality can be improved, which has a large practical effect.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of an image noise eliminator of the present invention.

FIG. 2 is a pattern diagram showing the contents of an image memory which is an input image used in the embodiment.

FIG. 3 is a pattern diagram showing an example of a pixel pattern in a mask when a pixel to be processed is white in a conventional image noise eliminator.

FIG. 4 is a pattern diagram showing an example of moving a mask.

FIG. 5 is a pattern diagram showing a result of an image noise eliminator conventionally used for FIG.

6 is a processing target pixel and windows A and B in FIG. 6;
Pattern diagram explaining the shape of

FIG. 7 is a pattern diagram showing configurations of windows A and B of the same embodiment.

FIG. 8 is a pattern diagram showing a result of applying the present invention to FIG.

[Explanation of symbols]

 1 Input Device 2, 9 Image Memory 3 Address Generation Unit 4, 5 Window Extraction Unit 6, 7 Run Length Counting Unit 8 Majority Calculator 10 Output Device

Claims (3)

[Claims] 1. A plurality of window extraction means for extracting a window composed of a plurality of pixels including a pixel to be processed, and a plurality of run length counts for counting run lengths of pixels included in the plurality of window regions. Means, and a majority decision means for taking a majority decision from the respective control values obtained from the plurality of run length counting means and the value of the processing target pixel, and the output data of the processing subject pixel having the value obtained from the majority decision means. An image noise eliminator characterized by: 2. A first window extracting means for extracting a first window, which is composed of h × 1 pixels including a pixel to be processed, and run lengths of pixels included in the first window area. A first run length counting means for counting; a second window extracting means for extracting a second window composed of 1 × i pixels including the pixel to be processed; and a second window area in the second window area. Second run length counting means for counting the run lengths of the pixels included therein, a control value obtained from the first run length counting means, and a control value obtained from the second run length counting means and the processing target. The image according to claim 1, further comprising a majority decision means for taking a majority decision from a pixel value, and a value obtained from the majority decision means is used as output data of the pixel to be processed. Image noise eliminator. 3. The image noise eliminator according to claim 2, wherein the first and second window regions can be specified by two rectangles centering on the pixel to be processed, and the rectangles are orthogonal to each other.