JPH0778247A - Device and method for eliminating binary image noise - Google Patents

Abstract

PURPOSE:To surely eliminate noises crowded in a binary image without requiring specific hardware by applying specific processing to an image to be processed. CONSTITUTION:A noise elimination part 13 takes the AND of the image to be processed and image in which the image to be processed is shifted in X and Y axial directions or two directions at 45 deg. from X axis or Y axis, respectively according to a shift parameter, and repeats processing to set the above result as the next image to be processed for (n) times (n>=2) according to a noise elimination parameter. The noise elimination part takes the AND of obtained images to be processed in the X and Y axis or in the two directions at 45 deg. from the X or Y axis and shifted images in which the image to be processed is shifted in at least the two directions according to the shift parameter, and repeats the processing to set the result as the next image to be processed for (n) times (n>=2) according to the noise elimination parameter. The AND of all the obtained images to be processed is taken, then, the AND of the result and an original image is taken.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing system for inputting a binary image from a scanner or a facsimile and storing or retrieving the binary image, and more particularly to noise removal of the input binary image.

[0002]

2. Description of the Related Art Generally, an image input from a scanner or a facsimile contains a noise component (noise). In order to reduce the effect of this noise, in the conventional image processing system, noise removal is performed on the multi-valued input image to perform binarization, according to either indispensability or selection.
Alternatively, noise removal is performed on a binary input image, and then image accumulation / registration processing is performed.

[0003]

However, in the method of binarizing a multi-valued image by removing noise from the image (for example, Japanese Patent Laid-Open No. 4-208772), a facsimile image transmitted from a remote location is Since it is generally a binary image, it cannot be applied to this.

Further, in a noise removal method for a binary input image (for example, Japanese Patent Laid-Open No. 4-205276), a binary signal of a reference point to be discriminated and a binary signal of surrounding points of the reference point are added. However, if the addition result is greater than a certain threshold value, it is considered as a signal image, and if it is less than the threshold value, it is determined as a noise pixel, so it is a noise pixel in a situation where noise pixels are dense. , Are often regarded as signal pixels. Moreover, in the noise eliminator disclosed in Japanese Patent Laid-Open No. 4-205276, since the signal determination circuit is configured by combining the shift registers, there is a drawback that special hardware is required to remove the noise.

Further, in the noise removal method for an image input in binary (for example, Japanese Patent Laid-Open No. 64-3782), 2 is used.
It has been proposed that image processing instructions such as shift of a value image, AND and OR are combined to remove noise from a binary image. In this method, the binary image is shifted only once in the X direction, Y direction, or ± 45 ° direction, and the result is combined to obtain the removed image, or the image is removed in the X direction, Y direction. The method of repeating each of the above processes is used. However, with this method, in the case of an original image as shown in FIG. 9A, even if noise removal processing is performed, noise (area C) that is denser than a certain dot as shown in FIG. 9B cannot be removed. There was a difficulty.

It is an object of the present invention to provide a binary image noise removing device and a noise removing method which can reliably remove the noise in which a binary image is dense, without requiring special hardware.

[0007]

A binary image noise removing apparatus according to a first aspect of the present invention comprises an image data input means for inputting binary image data as an original image and a noise removing parameter for removing noise. A noise removal parameter storage unit to store, a shift parameter storage unit to store a shift parameter serving as a reference when shifting an image, a processing target image based on the image data, and the processing target image according to the shift parameter. Axial direction, Y
The logical product of the shift images shifted in two directions of 45 degrees from the axial direction, the X-axis or the Y-axis is calculated, and the result is used as the next image to be processed.
(N ≧ 2) repeating first noise removing means, and X-axis direction and Y-axis direction obtained by the first noise removing means,
A process of ORing a processing target image in each of two directions of 45 degrees from the X axis or the Y axis and a shift image obtained by shifting the processing target image in at least two directions according to the shift parameter, and the processing result and the original According to the noise removal parameter, n (n ≧
2) The second noise removing means that repeats twice, and all the logics of the processing target image obtained by the second noise removing means in the X-axis direction, the Y-axis direction, the X-axis, or two directions of 45 degrees from the Y-axis. The present invention is characterized by comprising a third noise removing means for taking a sum and taking a logical product of the result and the original image.

A noise removing method according to a second aspect of the present invention uses binary image data as an original image, a noise removing parameter for removing noise, and a shift parameter serving as a reference when shifting an image. The logical product of the processing target image based on the image data and the shift image obtained by shifting the processing target image in the X-axis direction, the Y-axis direction, and the X-axis or two directions of 45 degrees from the Y-axis according to the shift parameter is obtained, and the result is obtained. The first noise removal procedure that repeats the process of using as the next processing target image n (n ≧ 2) times according to the noise removal parameter, and the X-axis direction, the Y-axis direction, and the X-axis obtained by the first noise removal procedure. 45 degrees from axis or Y-axis 2
Processing of each processing target image in each direction and a shift image obtained by shifting the processing target image in at least two directions according to the shift parameter, and the logical product of the processing result and the original image, and the result is A second noise removal procedure in which the processing of the image to be processed in the logical sum processing is repeated n (n ≧ 2) times according to the noise removal parameter, and the X-axis direction obtained in the second noise removal procedure, Y
And a third noise removal procedure for obtaining a logical product of the result and the original image by taking the logical sum of all the images to be processed in the two directions of 45 degrees from the axial direction, the X axis or the Y axis. To do.

[0009]

In the binary image noise eliminator and the noise eliminator, the image to be processed and the image to be processed are respectively in the X-axis direction, the Y-axis direction, and the X-axis or two directions of 45 degrees from the Y-axis according to the shift parameters. The logical product of the shift image and the shifted image is obtained, and the result is used as the next image to be processed.
2) Repeatedly, and the processing target images in the X axis direction, the Y axis direction, and the two directions of 45 degrees from the X axis or the Y axis, and the processing target image are shifted in at least two directions according to the shift parameter. The process of obtaining the logical sum of the shift image and the original image and the process result of the logical product of the process result and the original image are set as the process target image of the process of obtaining the next logical sum n (n ≧ 2) times. Repeatedly, take the logical sum of all the images to be processed in the X-axis direction, the Y-axis direction, and the two directions of 45 degrees from the X-axis or the Y-axis, and take the logical product of the result and the original image. I have to.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a binary image noise removing apparatus and a noise removing method according to the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing a functional configuration of an image processing system to which a binary image noise removing apparatus and a noise removing method according to the present invention are applied. The image processing system 10 includes an image data input unit 11, a noise removal parameter input unit 12, a noise removal unit 13, and an image data output unit 14.

The image data input unit 11 reads binary image data as an image to be processed from a file (not shown) on a memory and transfers it to the noise removing unit 13.

The noise removal parameter input unit 12 receives a noise removal parameter input by the user for noise removal, and transfers this noise removal parameter (parameter data) to the noise removal unit 13.

The noise removal unit 13 stores the noise removal parameter input from the noise removal parameter input unit 12 and the shift parameter input from the shift parameter input unit (not shown) in an internal memory (not shown), and the image data input unit Noise removal processing is performed on the image data input from 11 according to the noise removal parameter and the shift parameter.

The image data output unit 14 is a noise removing unit 1.
The noise-removed image data processed in 3 is written in a file from the memory and output to the outside as necessary.

Next, the configuration of the noise removing section 13 is shown in FIG.
Will be described with reference to the detailed block diagram of FIG. The horizontal direction filter means 16 performs the AND operation of the processing target image based on the image data and the shift image obtained by horizontally shifting the processing target image according to the shift parameter, and sets the result as the next processing target image. N according to the denoising parameter
A first noise removal process that is repeated (n ≧ 2) times, a processing target image based on the processing result obtained by the first noise removal process, and a shift image obtained by shifting the processing target image in three directions according to shift parameters. After performing the process of taking the OR, the process of ANDing the process result and the original image, and the process of setting the result as the next image to be processed is repeated n (n ≧ 2) times according to the noise removal parameter. Execute the removal process.

The vertical direction filter means 17 is an AN of a processing target image based on image data and a shift image obtained by vertically shifting the processing target image according to a shift parameter.
The process of taking D and setting the result as the next image to be processed is
A first noise removal process that is repeated n (n ≧ 2) times according to the noise removal parameter, a processing target image based on the processing result obtained by the first noise removal process, and the processing target image in three directions according to the shift parameter. After performing a process of ORing the shifted shift images, a process of ANDing the process result and the original image and setting the result as the next image to be processed is performed by n (n
A second noise removal process is repeated ≧ 2) times.

The lower left 45 ° direction filter means 18 ANDs the processing target image based on the image data and the shift image obtained by shifting the processing target image in the lower left 45 ° direction according to the shift parameter, and the result is calculated as follows. The process to be the image to be processed is n
A first noise removal process that is repeated (n ≧ 2) times, a processing target image based on a processing result obtained by the first noise removal process, and a shift image obtained by shifting the processing target image in two directions according to a shift parameter. After performing the process of taking the OR, the process of ANDing the process result and the original image, and the process of setting the result as the next image to be processed is repeated n (n ≧ 2) times according to the noise removal parameter. Execute the removal process.

The upper left 45 degree direction filter means 18 ANDs the processing target image based on the image data and the shifted image obtained by shifting the processing target image in the upper left 45 degree direction according to the shift parameter, and the result is calculated as follows. The process to be the image to be processed is n
A first noise removal process that is repeated (n ≧ 2) times, a processing target image based on a processing result obtained by the first noise removal process, and a shift image obtained by shifting the processing target image in two directions according to a shift parameter. After performing the process of taking the OR, the process of ANDing the process result and the original image, and the process of setting the result as the next image to be processed is repeated n (n ≧ 2) times according to the noise removal parameter. Execute the removal process.

An example of shift parameters held by each filter means is shown in FIG. The shift parameter is the first
And the parameter 2 used for the second noise removal processing. FIG. 4 shows the shift direction of the image according to each parameter. FIG. 4 shows the shift direction (arrow) and its coordinate data when viewed from the central segment indicated by the diagonal lines.
For the horizontal and vertical directions, shift to the four directions of up, down, left and right respectively.
Shift in two directions diagonally opposite each other, such as upper left and lower right. Each filter means, in addition to the noise removal parameter input from the noise removal parameter input means 12,
The shift parameter according to the direction of oneself is held in an internal memory (not shown). These parameters can be changed through a given interface.

The filter image synthesizing means 21 is arranged in the vertical direction, the horizontal direction, and the lower left 4 obtained by the filter means 16 to 19.
A third noise removal process is performed in which all ORs of the processing results in the 5 ° direction and the upper left 45 ° direction are taken and the results are ANDed with the original image.

The image thinning means 22 further thins the image obtained by the above process to create a final image. The thinning processing by the image thinning means 22 is not an essential processing requirement of the present invention, and thus can be omitted.

FIG. 5 shows the image processing system 1 shown in FIG.
It shows a concrete example for realizing 0, and shows the hardware configuration of, for example, a workstation or a personal computer.

The display 24 is a display device such as a CRT and displays input / output images and the like on the screen. The display on the display 24 is controlled by a display control unit (not shown).

The keyboard / mouse 25 is an input device for inputting data such as commands and character strings and for selecting instructions on the screen. Various input data and instructions are stored in a keyboard / not shown. It is sent to the processor unit 28 through the mouse control unit. The keyboard / mouse 25
The function of the noise removal parameter input unit 12 of FIG. 1 is realized.

The hard disk 26 is composed of a secondary storage device such as a magnetic disk, and stores input / output images in a file format. Input / output of data to / from the disk 26 is controlled by a disk controller (not shown). The hard disk 26 and a disk control unit (not shown) realize the functions of the image data input unit 11 and the image data output unit 14 of FIG.

The memory 27 is a buffer memory composed of a memory device such as a RAM, and stores binary image data read from the hard disk 26 and various control programs, and various data input from the keyboard / mouse 25. Temporarily store instructions and the like.

The processor unit 28 is a control unit composed of a CPU and its peripheral circuits, and operates the above-mentioned units according to a control program to control the entire system. The processor unit 28 realizes the function of the noise removing unit 13 of FIG.

These respective parts are connected to each other by a bus 29.

Next, the operation of the image processing system 10 for removing noise from binary image data will be described with reference to the flowchart of FIG. Note that, here, the noise removal processing when shifting in the vertical direction will be described as a typical example. Further, the noise removal parameter is set to 2, and the shift parameter uses the values shown in FIG.

First, the vertical direction filter means 17 of the noise removing unit 13 reads out the vertical direction shift parameter stored therein (step 101), and subsequently, the binary image data input from the image data input unit 11 is stored in the memory. Copy (step 102). Next, the vertical direction filter means 17
Performs AND processing of the processing target image based on the image data and the shift image obtained by vertically shifting the processing target image according to the vertical direction parameter 1 and sets the result as the next processing target image (step 103), and this process is repeated until the number of times n reaches 2 (step 104).

Next, the vertical direction filter means 17 ORs the processing target image based on the processing result obtained in step 103 and the shift image obtained by shifting the processing target image in three directions according to the vertical direction parameter 2. And the process of taking
The processing result and the original image are ANDed, and the result is processed as the next processing target image (step 105),
These steps are repeated until the number of times n of processing reaches 2 (step 106).

Next, the filtered image synthesizing means 21 compares the processing result obtained in the processing of step 107, the horizontal direction (steps 110 to 115), and the lower left (step 116).
Up to step 121), all ORs of the processing results in the upper left direction (steps 122 to 127) are taken, and the results are ANDed with the original image (step 107). Subsequently, the image thinning unit 22 is the filter image synthesizing unit 21.
The processing result obtained in (1) and the image obtained by shifting the processing result by one dot in the vertical direction and the image in the horizontal direction are ANDed to thin the image (step 108). The image data output unit 14 writes the noise-removed image data in a predetermined file (step 109).

Here, a specific example of the noise removal processing when the above-mentioned vertical shift is performed will be described with reference to FIG.

(1) First noise removal processing AND between the original image (a) and the shifted image (b) obtained by vertically shifting the original image according to the vertical parameter 1
(1st time) is taken to obtain the processing result (c). next,
The processing result (e) is obtained by ANDing (c) the image to be processed with the shift image (d) obtained by vertically shifting (c) in accordance with the vertical parameter 1.

(2) Second noise removal processing The processing result (e) and this (e) are used as the parameter 2 in the vertical direction.
According to the above, the OR (first time) is performed with each of the shift images shifted in the three directions to obtain the processing result (f). Subsequently, AND (first time) is performed between (f) and the original image (a) to obtain the processing result (g). Further, OR (second time) is taken between (g) and the shifted image obtained by shifting this (g) in three directions according to the vertical parameter 2, and the processing result (h) is obtained. Subsequently, AND (second time) is performed between (h) and the original image (a) to obtain the processing result (i).

The above-described processing is performed in the horizontal, lower left, and upper left directions, and the OR of all the processing results is taken, and finally the result is ANDed with the original image to obtain the third noise removal processing. Ends. This makes it possible to reliably remove noise such as sesame grains smaller than the value given by the noise removal parameter, that is, a group of pixels in which the number of consecutive pixels is smaller than the noise removal parameter. For example, the original image (a) as shown in FIG.
On the other hand, by setting the noise removal parameter to 2, it is possible to obtain the noise removed image (b) from which the regions A to C in which the number of consecutive pixels is 2 or less are removed. In this case, if the noise removal parameter is set to 1, the area C can be left. If the noise removal parameter is set to 3, the area D can be erased.

As described above, the noise removal can be optimized by appropriately setting the numerical value of the noise removal parameter according to the image quality of the input image and the characteristics of the input device. The denoising parameter that is most effective for denoising a general image is 4, which is a normal FAX.
Also suitable for images. This noise removal parameter can be changed according to the noise density, and can be changed as appropriate according to the image quality.

[0039]

As described above, in the binary image noise eliminator and noise eliminator according to the present invention, the image to be removed when the noise is removed is the X-axis direction, Y-axis direction, X-axis or Since the shift is performed in two directions of 45 degrees from the Y-axis, the process of obtaining the logical sum and the logical product is performed by the number of the noise removal parameters, respectively, and the result is combined to perform the process of obtaining the logical sum and the logical product. Even in a situation where noise pixels are dense, noise such as sesame seeds can be reliably removed. Moreover, since processing is performed by combining only the simplest image processing commands such as binary image shift, AND, and OR, in a normal workstation, noise removal of a binary image is executed without special hardware. be able to. Further, by setting the noise removal parameter to an appropriate value, it is possible to perform noise removal according to the characteristics of the input image or the input device.

In particular, when compared with the method in which noise is removed from a multi-valued input image and binarization is performed, the processing can be speeded up due to the simple software configuration, and the memory consumption can be reduced. be able to.

[Brief description of drawings]

FIG. 1 is a block diagram showing a functional configuration of an image processing system (embodiment).

FIG. 2 is a block diagram showing a detailed configuration of a noise removing unit.

FIG. 3 is an explanatory diagram showing an example of shift parameters held by each filter means.

FIG. 4 is an explanatory diagram showing an image shift direction according to each parameter.

FIG. 5 is a configuration diagram showing a specific example for realizing an image processing system.

FIG. 6 is a flowchart showing a noise removal processing procedure.

FIG. 7 is an explanatory diagram showing a specific example (vertical direction) of noise removal processing.

FIG. 8 is an explanatory diagram showing a relationship (present invention) between an original image and a noise-removed image.

FIG. 9 is an explanatory diagram showing a relationship (prior art example) between an original image and a noise-removed image.

[Explanation of symbols]

10 ... Image processing system, 11 ... Image data input unit, 1
2 ... Noise removal parameter input unit, 13 ... Noise removal unit, 14 ... Image data output unit, 16 ... Horizontal filter unit, 17 ... Vertical filter unit, 18 ... Lower left filter unit, 19 ... Upper left filter unit, 21 ... filter image synthesizing means, 22 ... image thinning means

Claims (2)

[Claims] 1. Image data input means for inputting binary image data as an original image, noise removal parameter storage means for storing noise removal parameters, shift parameter storage means for storing shift parameters, and the image data Based on the shift parameter, and the processing target image based on
According to the noise removal parameter, the logical product of a shift image shifted in two directions of 45 degrees from the axial direction, the X-axis, or the Y-axis is taken, and the result is set as the next image to be processed.
(N ≧ 2) times of first noise removing means, and each of the X-axis direction, the Y-axis direction, and the two directions of 45 degrees from the X-axis or Y-axis obtained by the first noise removing means. A process of taking a logical sum of a processing target image based on the result and a shift image obtained by shifting the processing target image in at least two directions according to a shift parameter, and taking a logical product of the processing result and the original image, Process the image to be processed as n (n (n
≧ 2) second noise removing means, and all the processing results obtained by the second noise removing means in the X-axis direction, the Y-axis direction, and the two directions of 45 degrees from the X-axis or the Y-axis. A binary image noise removing device comprising: a third noise removing means for taking a logical sum and taking a logical product of the result and the original image. 2. Binary image data serving as an original image, a noise removal parameter, and a shift parameter are used, and an image to be processed based on the image data and the image to be processed are respectively set in the X-axis direction and Y according to the shift parameter. Axial direction,
According to the noise removal parameter, n is calculated according to the noise removal parameter by performing a logical product with a shift image that is shifted in two directions of 45 degrees from the X axis or the Y axis and using the result as the next image to be processed.
2) a first noise removal procedure that is repeated twice, and an image to be processed in each of the X-axis direction, the Y-axis direction, and the X-axis or two directions of 45 degrees from the Y-axis obtained in the first noise removal procedure. A process of taking a logical sum of a processed image and a shift image obtained by shifting the processed image in at least two directions according to a shift parameter, and a process of taking a logical product of the processed result and the original image and setting the result as the next processed image; A second noise removal procedure repeated n (n ≧ 2) times according to the noise removal parameter, and the X-axis direction, the Y-axis direction, the X-axis or two directions of 45 degrees from the Y-axis obtained in the second noise removal procedure. A noise removal method comprising: taking a logical sum of all the images to be processed and taking a logical product of the result and the original image.