JPS6327752B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a character/figure separation method, and more particularly to a character/figure separation method that allows only character information to be separated and recognized even from image information containing a mixture of characters and figures. .

For example, in ordinary drawings such as design drawings of printed circuits and logic circuits, characters and figures are usually drawn in a mixture. Therefore, when recognizing such drawings, the mixture of figures and characters makes automatic recognition difficult. This is because the processing method used for graphics and the processing method used for characters are completely different. Therefore, when recognizing a drawing and, for example, drawing a handwritten drawing using a data processing device, there is a certain certain pattern of shapes, such as straight lines, rectangles, squares, or circles, and it is necessary to draw the drawing based on that pattern. However, since the identification of characters is completely different from the identification of graphics, when recognizing a drawing, it is first necessary to separate the characters and graphics.

The object of the present invention is to provide a character/figure separation method that makes it possible to separate characters and figures in accordance with such needs. In a character/figure separation method for separating characters and figures from image information containing a mixture of images and figures, an input means for inputting the image information, an information holding means for holding input information input from the input means, and the information It has a feature extraction means for extracting features from the information held in the holding means using a quadrilateral mask, and scans the input information held in the information holding means through the mask to extract features from the left side of the mask. If the part constituting the character does not exist in , and the part exists on the right side, it is determined to be the start position of the character information area, and then the mask is scanned and the part exists on the top and bottom sides of the mask. If not, and the part exists on the right side, it is determined that it is within the character information area, and the mask is further scanned to find that the part on the left side of the mask exists, and the part exists on the right side. If not, the character information area is determined to be at the edge of the character area and the character information area is identified, thereby separating the characters and graphics.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Before describing one embodiment of the present invention in detail, an outline of the present invention will be explained based on FIGS. 1 to 3.

First, as shown in Figure 1, the horizontal width is Yl,
A hollow mask M with a vertical height of Tl and a width of each side of unit length is used. Then, as shown in FIG. 2A, the image information stored in the buffer indicates whether all the areas A in this mask M are ``0'' and there are even some ``1''s in area B. Scan about. Note that here, "1" indicates that image information exists, and "0" indicates that it does not exist. Image information is scanned in the x direction in the mode 0 state shown in FIG. 2A. As shown in Figure 4, when entering the first part of the character area,
Mode 0 is established. That is, in FIG. 2A, area A is all 0, and area B is at least 1.
There will be ``1''. This mode 0
When this is true, the mask enters mode 3, which checks whether areas E and F are all 0, as shown in Fig. 2C, and this mode in which areas E and F are all 0 is activated. 3 is established, the mask prints "1" in the area G shown in FIG. 4 at that position.
Mode 2 is entered to check whether or not exists. In this way, if mode 3 and mode 2 are established, it is determined that the character area is continuous, and
In this state, scanning in the x direction is performed. mode 0
In contrast, in mode 2, the A part in mode 0 is ignored (it can be 0 or 1), and only the part corresponding to B is verified. The reason for this is that mode 0 is for finding the beginning of a character string, and mode 2 is for finding connections between individual characters that make up the string. When checking in mode 3, for example, if "1" exists in area E, it is assumed that it is not character information but graphic information such as a line or pattern, and scanning in the x direction is performed again in mode 0. If "1" does not exist in the mode 2 check, in order to judge whether or not it is the end of the character area, mode 1 shown in FIG. Check whether "1" exists. If mode 1 is established, this is determined to be the end of the character, as shown in FIG.

By the way, in this case, the size of the mask M cannot be fixed because the size of the characters often differs depending on the screen. Therefore, in order to determine the size of this mask, that is, its width Yl and height Tl, the process shown in FIG. 5 is first performed.

(a) First, as shown in FIG. 5A, the entire screen information is scanned to find its "1" and "0". Then, from this full-screen information, the changing points of the image information, that is, the points where the image information changes from "0" to "1" and from "1" to "0" are determined. For example, as shown in Figure 5B,
Now when scanning on line l, x ₁ is "0"
→ changes to “1”, x ₂ changes from “1” to “0”, x ₃ changes from “0” to “1”, and x ₄ changes from “1” to “0”. When the change points are determined in this way, the sum of the intervals between the change points _T-1 〓 ^x=1 l and the number of change points Ty are counted. Then, the average distance between the change points is determined by dividing the sum of the intervals between the change points by the number of change points Ty.

(b) Then, the entire screen is scanned, and the coordinates of the change points where the distance between the change points is smaller than the above-mentioned average distance are stored as the interpolation points that include characters in the vicinity. As a result, the information on the marquee complement points as shown in FIG. 5C is obtained.

(c) Let M _L be the maximum value of the interval between the next consecutive interpolation points. For example, in the original picture in Figure 8A,
An example of a change point is the black dot in Figure B (“0” → “1”).
Partially shown as a white dot (“1” → “0”). The intervals are l ₁ , l ₂ . . . in Figure 8 C.
Denoted as ln. Among these, the interval less than the average distance of this interval is the 8th
They are indicated by l ₁ , l ₂ , l ₄ ... shown in Figure D. The maximum value ML of the interval is shown as in FIG. 8(e).

(d) Perform a similar scan in the vertical direction, and calculate the average interval between the start and end points of consecutive interpolation points.
Tlm. Then, 5/4 times this Tlm is the height Tl of the mask M, and 5/8 is the width Yl of the mask M.

Tl=5/4Tlm Yl=5/8Tlm These values were obtained through experiments by the inventors. The above-mentioned characters are identified using the mask M having the size thus determined. In the example of FIG. 8, Tlm is as shown in F of the figure, and Tl and Yl are as shown in G of the figure.

Next, the configuration of one embodiment of the present invention is shown in FIGS. 6 and 7.
This will be explained based on the diagram.

FIG. 6 shows the configuration of an embodiment of the present invention, and FIG. 7 is a flowchart for explaining the operating state of the present invention.

In the figure, 1 is an input section, 2 is an image memory, 3 is a buffer, 4 is a parameter calculation section, 5 is an address generation section, 6 is an area ROM, 7 is a control section, 8 is an AND circuit, 9 is a feature extraction section, 10 is a gate, 11 is an address table, 12 is an image clear circuit, and 13 is an output memory.

The input section 1 reads a drawing and converts it into an electrical signal, and image information input from the input section 1 is held in an image memory 2. The buffer 3 is a buffer memory for use as a work area for determining points of change from image information and for performing mask scanning.

The parameter calculation unit 4 detects changing points from the image information, calculates parameter values such as M _L , Tl, Yl, etc., and holds them.

The address generator 5 generates an address signal for setting data in the buffer 3 or reading the set data.

The area ROM 6 is for outputting data that defines the area of the mask M, and parameters for determining the mask M are transmitted from the parameter calculation unit 4.

The control unit 7 is a control unit for generating control signals necessary for data processing up to separating characters from screen information.

The feature extractor 9 identifies whether or not the information is character information based on the mode shown in FIG.

The address table 11 is a register in which the start coordinate xs where a character exists and the end coordinate xE where the character ends are entered.

The image clear circuit 12 is a clear circuit that clears only the character area of the image information set in the image memory 2.

The output memory 13 is a memory in which the image information set in the image memory 2 with the character area cleared is held for output.

Next, the operation shown in FIG. 6 will be explained.

(1) First, when a drawing is set in the input section 1, the input section 1 converts the drawing into an electrical signal and sends it to the image memory 2. The image memory 2 holds image information of the drawing. Then, the control section 7 performs control to sequentially read image information from the image memory 2 to the buffer 3. Then, the image information set in this buffer 3 is transferred to the 5th buffer.
Scanning is performed as shown in FIG. Then, based on the change points detected in this way, the parameter calculation unit 4 performs the calculations detailed in (a) to (d) above to obtain ML, Tl, and Yl, and uses these values for storage. It is held in a buffer and the calculation result is transmitted to the area ROM6. thus the area
In ROM6, among these values, Tl and Yl
Create a mask of a predetermined size.

(2) The image information is accessed using the mask obtained in this way. First, the mask M is placed at the coordinate position of the upper left corner of the image information from the address generation unit 5, and this mask M is sequentially moved in the x-axis direction. It is checked whether or not the above-mentioned mode 0 (mode 0 is indicated as C ₁ in the flowchart of FIG. 7), that is, C ₁ is established. If C ₁ does not hold, this scanning is continued in sequence until the x-coordinate at that time reaches the right end WDH. And even if this mask is scanned to the right end, if C ₁ does not hold, then
This time, the mask is advanced one step in the y-axis direction and a similar scan is performed. Such scanning is performed according to the address generated from the address generation section 5,
For the mask area, mask control is performed by sending a signal that is sent from the area ROM 6 and becomes "1" only in the area of the mask M to the AND circuit 8, and whether or not each mode is established is determined by feature extraction. This will be done in Section 9.

(3) If the above C ₁ is established at a certain position, the upper left corner position of the mask at that time is determined as the character area starting point xs, and this is stored in the register of the feature extraction unit 9. do. Then, the feature extraction unit 9 determines whether mode 3 (C ₃ ) is established by advancing the coordinate in the x-axis direction by one. Of course, at this time, a mode 2 check is also performed to check that "1" exists in area G shown in FIG. However, if mode 3 is not established, it is determined that the area is not a character area, and the character area starting point xs previously held in the feature extraction unit 9 is deleted. Then, check whether _C1 holds true again.

(4) If C ₃ is established in the above (3) and then the check of mode 2 is established, then the check of mode 3 (C ₃ ) is not established at that point. In other words, area D shown in Figure 2B
indicates that there were not all 0's but 1's. And from the x coordinate at this time, the above
If x−xs, which is obtained by subtracting xs, is smaller than the above ML, the x coordinate is incremented by 1, the mask progresses, and C ₃
It is checked whether or not it holds true. As the mask progressed, _C3 was established, but
Next, when checking in mode 2, if "1" does not exist in region G, this means that _C2 is established. Then, as shown in Figure 4, the value of the coordinate x at this time is determined as the character area end point xE, and this is set as
Fill in 1. In this way, the starting point of the character area
When xs and character area end point xE are obtained,
Open gate 10 and set these character area starting points xs
and character area end point xE is address table 11
sent to. As a result, the image clearing circuit 12 operates, and from the image information held in the image memory 2, the height Yl of the mask M (or the height inside the mask M) is determined between the character area start point xs and the character area end point xE. Clear the area of Yl′).

(5) After this operation is performed on all the image information and all the character areas are cleared, the data held in the image memory 2 is transmitted to the output memory 13. Of course, the character information of the image information transmitted to the output memory 13 at this time has been cleared. In this way, the output memory 13
Image information with separated characters can be obtained from .

As explained above, according to the present invention, only the character area can be separated using a mask, so it becomes possible to distinguish character information from image information, and it becomes very easy to extract only graphic information. .

In the above explanation, an example has been explained in which the character area is cleared by the image clearing circuit, but of course, instead of clearing the area between the character area start point and the character area end point as described above, it is also possible to extract only this area.
It is also possible to perform operations such as identifying the characters written in this character area using the retrieved information.

[Brief explanation of the drawing]

Figure 1 shows the mask used in the present invention, and Figure 2 shows the mask used in the present invention.
The figure is an explanatory diagram of the identification mode using the mask, Fig. 3 is an explanatory diagram of the scanning state of the mask, Fig. 4 is an explanatory diagram of the presence of characters, the identification mode, the character area start point and the character area end point, and Fig. 5 is an explanatory diagram of the character area start point and character area end point. An explanatory diagram of the creation of parameters necessary for determining the size of a mask, etc., Fig. 6 is a configuration diagram of an embodiment of the present invention, Fig. 7 is a flowchart explaining its operating state, and Fig. 8 is an illustration of the method of the present invention. It is an operation explanatory diagram. In the figure, 1 is an input section, 2 is an image memory, 3 is a buffer, 4 is a parameter calculation section, 5 is an address generation section, 6 is an area ROM, 7 is a control section, 8 is an AND circuit, 9 is a feature extraction section, 10 is a gate, 11 is an address table, 12 is an image clear circuit, and 13 is an output memory.

Claims (1)

[Scope of Claims] 1. A character/figure separation method for separating characters and figures from image information containing a mixture of characters and figures, comprising: an input means for inputting the image information; and holding input information input from the input means. and a feature extracting means for extracting features from the information held in the information holding means using a quadrilateral mask. When scanning, if the part constituting the character does not exist on the left side of the mask and the part exists on the right side, it is determined that the character information area is the starting position, and then the mask is scanned to find the part of the mask. If the portion does not exist on the upper and lower sides and the portion exists on the right side, it is determined that the character information area is within the character information area, and further scans the mask to find that the portion on the left side of the mask exists; If the part does not exist on the right side, it is determined that it is the end of the character area, and the character information area is identified, thereby separating characters and graphics.