JPS6010352B2 - Stain removal method in figure recognition device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a stain removal method in a figure recognition device, and in particular a figure recognition method in which a figure such as a character written on a form is removed and recognized after the stain is removed through a stain removal circuit. A figure recognition device that measures the size of a stain on a blank form used by a user, and sets a stain removal area in which the stain removal circuit section considers it to be a stain and removes it accordingly. relates to a stain removal method.

In a figure recognition device, in a so-called pre-processing process, a stain existing on a form is extracted while distinguishing it from the figure to be recognized, and the stain is removed by a stain removal circuit.

In this case, for example, the operator assumes the size of the stain in advance, and removes isolated points within the assumed size, considering them to be stains. However, the size of a stain that exists on a blank form used by a user is undefined, and it is desirable to set the threshold value of the stain removal area of the stain removal circuit unit according to the user's current situation.

The purpose of the present invention is to solve this problem, and the stain removal method in the figure recognition device of the present invention extracts the figure written on the form, converts it into an electrical signal, and sends it through the stain removal circuit. In a figure recognition device that performs the figure recognition after changing the black mesh points corresponding to the stains to white mesh points, the blank form before the figure is written has m mesh length and n mesh width (where m = A scanning window having a size corresponding to mxn meshes (including n), which scans with a plurality of types of scanning windows having different values m and/or n, and scans the scanned m
In a scanning window with a size corresponding to xn mesh, all 2(m-1)12(n-1) mesh positions corresponding to the outermost frame of the window are white mesh points. A stain area determination circuit section is provided to set a minimum scanning window having the minimum size for determining the size of a stain that can be distinguished from minute dots on the recognition target figure, and the stain area determination circuit section is configured to perform a process using the minimum scanning window. The present invention is characterized in that the size of the determined stain is determined, and the threshold value of the stain removal area of the stain removal circuit section is set to the size given by the minimum scanning window.

This will be explained below with reference to the drawings. FIG. 1 is an explanatory diagram illustrating how a form is scanned by a scanning window, FIG. An embodiment of the structure of a stain removal method in a recognition device, FIG. 4 is an explanatory diagram conceptually explaining an embodiment of the process of the stain area determination circuit section according to the present invention, and FIG. 5 is a diagram showing the stain removal circuit according to the present invention. FIG. 4 is an explanatory diagram illustrating an example process of the part.

In FIG. 1, 1 represents a form, 2 represents a figure recording frame area on the form, and 3, 3--, and 3-2 represent scanning windows, respectively.
Also, in Figure 2, 3-3×3, 3-4×4,...
...3-10×10 is a code that specifies scanning windows of various sizes. For example, scanning window 3-10×1 has 10 meshes vertically and 10 meshes horizontally. It represents a scanning window with a size corresponding to a ×1 solid mesh.

In the present invention, for example, the description that scanning is performed using a 10×10 mesh scanning window means the following, as is well known. That is, the graphic recording frame area 2 shown in FIG.
This means examining the figure that can be seen through the upper window that has a size of 0 mesh. In the case of the present invention, when the form 1 used by the user is in a blank state, the figure recording frame area 2 is scanned by the scanning window 3 to measure the size of the stain existing on the blank form.

In this measurement, many blank forms are scanned and the results are measured. That is, for example, by using a scanning window 3 of 10 x 10 meshes as shown in Fig. 2 (10 x 101), the figure recording frame area 2 in a blank state can be scanned by the arrows a, b, . . . shown in Fig. 1. It scans like a raster.
Then, if the stain existing on the figure recording frame area 2 which is in a blank state is within the mesh range excluding the outermost 36 meshes of the scanning window 3-10x10 (in this case, 8x8 meshes), If it appears that the scanning window 3-10.times.1 is within 10 minutes, the scan window 3-10.times.1 is replaced with a smaller scanning window 3-9.times.9. 4 excluding the outermost 32 meshes of the scanning window 3-9×9
If it appears within the 9 mesh range (7x7 mesh), scanning is performed using the next smaller scanning window 3 - 8x8. Thereafter, the scanning window 3 is successively reduced in size in the same manner, and the size of the stain on the blank form is measured.

In this case, in addition to simply measuring the largest stain, a predetermined number of blank forms are scanned and the scanning windows 3 and 3 are scanned.
Number of stains that can be covered by ×3, scanning window 3-4
The number of stains that can be covered with ×4,...
It is desirable to create a frequency graph by examining the above, and measure the size of the stain within a preferable range that can be distinguished from the dots of the figure to be recognized and removed as a stain, as described later. FIG. 3 shows the configuration of an embodiment of the stain removal method in the figure recognition device of the present invention, in which reference numeral 4 is a data processing section, 5 is a stain area determination circuit unit and stain removal circuit section, and 6 is a scanning window setting. , 7 is a latch circuit, and 8 is an address register which is a PROM.
9 is a fixed address register for setting the output address information for the scanning window setting section 6, which is configured with The ``video signal'' that sets the address information for the ``video signal'' represents a video signal obtained by scanning a blank form or a form with figures written on it using, for example, a flying spot. Note that in order to realize the process of scanning using the m×n mesh scanning window,
By using the video signal obtained by scanning each mesh in a raster manner using the above-mentioned flying spot, etc., as is well known, the white color of each mesh within a frame corresponding to a window of m x n meshes is The black signals (m×n) can be viewed at once, and the view is sequentially shifted in accordance with the scanning of the flying spots. The illustrated video signal is used for this purpose. The process of scanning a blank form using the scanning window 3 to measure the size of a stain is carried out as follows.

That is, {1- In the processing process, the circuit section 5 operates as a stain area determination circuit section.

(2) The video signal obtained from the blank form is passed through the stain area determination circuit section 5 to the data processing section 4.

'3} At this time, the address register 8 is initially set with address information that causes the scanning window setting unit 6 to output a signal for selecting the scanning window 3-10-×10 to the stain area determination circuit unit 5. Ru.

If a spot covered by the scanning window 3-10.times.10 is detected while the video signal obtained from the '41 blank form is being input, the latch circuit 7 is set. The latch circuit 7 may be reset at the time when the video signal for one form has been processed, or when the latch circuit 7 is set even once, it may be reset at another memory (not shown). It may be stored in the unit and then immediately reset. '5} Once the latch circuit 7 is set, a signal is output to the address register 8 to select the scanning window 3-9×9 from the scanning window setting section 6 to the stain area determination circuit section 5. address information is set.

That is, the size of the stain is examined using the scanning window 3 - 9×9. '
6- Then, the same process as shown in process (2) above is performed, and when the latch circuit 7 is set at least once, the size of the stain by the scanning window 3-8×8 is then checked.

As mentioned above, the scanning window 3-10×10, 3-9×9...
When the size of the stain is determined using the following steps, and when the maximum size of the stain that should be distinguished from the dots of the recognition target figure and removed as a stain is determined, the stain removal circuit 5 is asked to remove the stain in subsequent processing. Set the size of .

That is, predetermined address information is set in the fixed address register 9 shown in the figure, and the scanning window setting section 6 sets the stain removal circuit section 5.
A signal indicating the size of the stain to be removed is output. After that, the process of removing the stain begins. That is, after 1971, a video signal obtained from a form on which a figure was written was input to the stain removal circuit section 5.

'81 Then, the stain removal circuit section 5 removes the stains within the specified size range, and the video signal from which the stains have been removed is supplied to the data processing section 4.

That is, the data processing section 4 performs figure recognition processing as in Kosanku, based on the video signal from which the stains have been removed. FIG. 4 is an explanatory diagram illustrating one embodiment of the process of the stain area determination circuit according to the present invention, and FIG. 5 is an explanatory diagram illustrating one embodiment of the process of the stain removal circuit according to the present invention. Show the diagram.

The symbol 31mxn in the figure is a scanning window consisting of mxn mesh. bll, b12, ... is mesh information, and a logic "1" is given corresponding to a white mesh point, and a logic "1" is given corresponding to a black mesh point. Those given “0”, 10-IA
and 10-1B, 10-2A and 10-2B, 10-3A and 10-3B, . . . are shift registers, 11-1,
12-1, 13-1, . . . represent AND circuits, respectively.
When it is assumed that an m×n mesh is specified as the scanning window 3 by the scanning window setting unit 6 for the stain area determination circuit unit 5,
A blank form is scanned by the scanning window 3-m×n.

And the outermost 2(m-1)12 of the scanning window 3-mxn
(n-1) mesh positions are all white mesh points and are surrounded by the outermost mesh position (m-2
)(n-2) mesh positions.
When there are two black mesh points, the scanning window 31m×
It is determined that there is a stain with a size covered by n. In other words, since the white information is logic "1" and the black information is logic "0", when the logical formula F shown in FIG. 4 satisfies F=1, it is determined that there is a stain of the size to be covered be done. Shift register 10-IA and 1 shown in FIG.
For example, mesh information for one column on the form is serially input to 0-IB, and shift registers 10-2A and 1
Similarly, mesh information for one column of the next column is input to 0-2B, and shift registers 10-3A and 10-3B
In addition, the mesh information for one column of the next column is input, and so on.

At this time, for example, each terminal on shift register 10-IA, each terminal on shift register 10-2A, etc.
As is well known, by observing the state in which the information form changes from .
This means that one river trowel figure has been scanned. Suppose now that the stain removal circuit unit 5 is instructed to remove a stain covered by a 10×10 mesh scanning window. In this case, AND circuit 1 corresponds to shift register 0-IA.
1-1 corresponds to the shift register 10-2A, and the AND circuit 11-2 corresponds to the shift register 10-10A.
In response to this, AND circuits 11-10 are respectively enabled to be turned on. Therefore, as the video signal is serially input, shift registers 10-1A, 10-2A, .
. . . From the output terminal corresponding to each mesh from above 10-10A, the output of scanning the form through the 10×10 mesh scanning window 3-10×1 river is taken out, as is well known. When it is detected that all 36 mesh positions on the outermost periphery of the scanning window 3-10×10 become white mesh points, the shift registers 10-IA, 10-2
A, . . . , 10-10A, the shift register 10-
Reset IA, 10-2A, ..., 10-10A.

That is, the entire 10x10 mesh that overlaps with the scanning window 3-10x10 at that point in time is forcibly changed to a white mesh point. Similarly, if the stain removal circuit unit 5 is instructed to remove stains covered by a 4×4 mesh scanning window, the AND circuit 12-1 is activated in response to the shift register 10-1A. ,shift·
An AND circuit 12-2 corresponds to the register 10-2A, and an AND circuit 12-2 corresponds to the shift register 10-3A.
3, the AND circuits 12-4 are enabled to be turned on corresponding to the shift register 10-4A.

Therefore, the scanning window 3 is scanned in a 4×4 manner. When it is detected that all 12 mesh positions on the outermost periphery of the scanning window 3 - 4 x 4 become white mesh points, the 4 x 4 mesh that overlaps with the scanning window 3 - 4 x 4 at that point in time is detected. Forces the entire mesh to change to white mesh points. Therefore, mx is now applied to the stain removal circuit section 5.
When a video signal obtained from a form on which a figure is written is passed through the stain removal circuit section 5 assuming that n mesh is specified, the video signal obtained from the stain removal circuit section 5 is (m The resulting video signal is obtained by erasing all dots having an area smaller than -1)×(n-1).
In other words, a video signal is obtained in which all stains having an area smaller than (m-1)×(n-1) are removed. As described above, according to the present invention, it is possible to set the darkness value of the stain removal area of the stain removal circuit section according to the size of the stain present on the blank form used by the user.

[Brief explanation of the drawing]

FIG. 1 is an explanatory diagram illustrating how a form is scanned by a scanning window, FIG. 2 is an example of an m×n mesh scanning window according to the present invention, and FIG. 3 is a diagram of a figure according to the present invention. The structure of one embodiment of the stain removal method in the recognition device, FIG. 4 is an explanatory diagram conceptually explaining the processing of one embodiment of the stain area determination circuit according to the present invention, and FIG. 5 is the configuration of the stain removal circuit according to the present invention. An explanatory diagram illustrating one embodiment of processing is shown. In the figure, 1 is a form, 3 is a scanning window, 4 is a data processing section, 5 is a stain area determination circuit/stain removal circuit, 6 is a scanning window setting section, and 7 is a latch circuit. A 1 figure 2 figures 3 figures Ge 4 figures 75 figures

Claims (1)

[Claims] 1. A figure written on a form is read and converted into an electrical signal, and a black mesh point corresponding to the stain is changed to a white mesh point via a stain removal circuit, and then the figure is recognized. In a figure recognition device that performs this, the blank form before the above-mentioned figure is written has a size corresponding to m×n meshes of m meshes in height and n meshes in width (including the size m=n). The outermost periphery of the scanning window is scanned by a plurality of types of scanning windows in which the values m and/or n are different, and the scanning window has a size corresponding to the scanned m×n mesh. 2(m-1) corresponding to the frame
Among the scanning windows in which +2(n-1) mesh positions are all white mesh points, set the minimum scanning window with the minimum size that determines the size of a stain that can be distinguished from a minute dot on the recognition target figure. a stain area determination circuit section for determining the size of the stain determined by the minimum scanning window in the stain area determination circuit section, and determining a threshold value of the stain removal area of the stain removal circuit section using the minimum scanning window. A stain removal method in a figure recognition device, characterized in that the stain is set to a given size. 2 The stain removal circuit unit has a stain removal area of m×n mesh given by the threshold value, and scans a form on which a figure is written to determine the area corresponding to the outermost frame of the stain removal area. 1) When it is detected that +2(n-1) mesh positions are all white mesh points, the above m×n
2. A stain removal method in a figure recognition device according to claim 1, wherein all mesh points within a mesh stain removal area are forcibly changed to white mesh points.