JPH05205104A - System for coupling blurred character - Google Patents

Abstract

PURPOSE:To exactly couple character areas related to blurred characters even though it is a blurred character area written in the normal typeface or the one written in the inclined typeface. CONSTITUTION:An area information check means 13 checks the area information including the size of the character area, the dimension and coordinate of the contact area, the distance between contact areas, and coupling character width for a character line picture to be processed. A blurred character discriminating means 14 judges whether or not the character in the character area is one part of the blurred character based on the area information. A character area coupling means 15 generates a coupling area by coupling the adjacent character areas where internal characters are judged to be one part of the blurred character based on the recognition result of the blurred character discriminating means 14.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a blurred character combining method, and more particularly to a blurred character combining method in an English type recognition system.

[0002]

2. Description of the Related Art Conventionally, in this kind of blurring character combination method,
The presence of italic characters that are diagonally inclined characters is not taken into consideration, and processing is performed assuming that only normal characters that are upright characters exist.

A conventional blurred character combining method will be described with reference to FIGS.

Here, in the English type recognition system, it is estimated that the character region is the character region related to the character line image (the image of the character group for one line) shown in FIG. The character pitch P is shown in Fig. 2.
It is assumed that it is determined as shown in (b) and (c).

First, based on the estimated character pitch P in FIG. 2C, the estimated character width W and the estimated inter-character space S in FIG. 2 are obtained. The relationship between the estimated character pitch P, the estimated character width W, and the estimated inter-character space S is as follows. In the equation, C ₁ and C ₂ are constants, and are values obtained by examining some of the actual character line images (or by performing a predetermined experiment). W = C ₁ × P (where 0 <C ₁ <1) S = C ₂ × (P−W) (where 1> C ₂ )

On the other hand, the inter-character space shown in FIG. 2 (d) and the combined character width shown in FIG. 2 (e) are obtained based on the character area shown in FIG. 2 (b). The inter-character space is the length from the right side of one character area to the left side of the next character area, and the combined character width is the length from the left side of one character area to the right side of the next character area.

Finally, based on the estimated character width W, the estimated inter-character space S, the inter-character space and the combined character width, adjacent character areas (combined areas) to be combined are obtained. That is, two adjacent character areas whose combined character width is smaller than the estimated character width W and whose inter-character space is smaller than the estimated inter-character space S are searched for, and the area where the searched character areas are combined is obtained as the combined area ( 2 (f)).

In the above-mentioned blurred character combination method, adjacent character areas satisfying the condition that "the combined character width is smaller than the estimated character width W and the inter-character space is smaller than the estimated character width S" are recognized as the combined area. .. Here, since the italic characters are inclined, the inter-character space between adjacent character areas related to the italic characters is generally small, and the combined character width is small. Therefore, with respect to the character area related to italic characters, the above-mentioned condition is often satisfied even though the character area is not related to blurred characters. That is, in the conventional blurred character combination method, should all the characters to be processed be ordinary characters (without considering the presence of italic characters), should the character areas be combined only under the above conditions? It was determined whether or not.

[0009]

In the above-mentioned conventional blurred character combination method, since the existence of italic characters is not taken into consideration, there is a drawback that the character areas related to italic characters are accidentally combined.

In view of the above points, an object of the present invention is to accurately and commonly connect a character area related to a blurred character of a normal character and a character area related to a blurred character of an italic character. It is to provide a blurred character combination method that enables

[0011]

The blurred character combination method of the present invention is an area including a character area size, a contact area size and coordinates, a contact area distance, and a combined character width for a character line image to be processed. Area information survey means for information,
Based on the judgment result by the blurred character judgment means for judging whether or not the character in the character area is a part of the blurred character based on the area information obtained by the area information investigation means, and the judgment result by this blurred character judgment means. And a character area combining unit that combines the adjacent character areas in which the internal character is determined to be a part of the blurred character to generate a combined area.

[0012]

In the blurred character combination method of the present invention, the area information examining means includes area information including the size of the character area, the size of the contact area, the coordinates, the distance between the contact areas, and the combined character width with respect to the character line image to be processed. The blurred character determining means determines whether or not the character in the character area is a part of the blurred character based on the area information obtained by the area information examining means, and the character area combining means determines the blurred character determining means. Based on the result of the determination by, the adjacent character areas, which are determined to have the internal character as a part of the blurred character, are combined to generate a combined area.

[0013]

The present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of the blurred character combination method of the present invention. The blurring character combination method of the present embodiment is connected to the existing character extraction method 11 and the existing character recognition method 16 in the English type recognition system. Further, the blurred character combination method of the present embodiment is configured to include a character line image storage means 12, an area information investigation means 13, a blurred character determination means 14, and a character area combination means 15.

FIGS. 3A to 3C are views for explaining the specific operation of the blurred character combination method of this embodiment.

FIGS. 4A and 4B are views for explaining the processing by the area information examining means 13.

FIG. 5 is a flow chart showing the processing of the blurred character judging means 14. This processing is performed in the character area extraction step 5
1, a character line image processing end determination step 52, a combined character width determination step 53, a contact area size determination step 54, a contact area perspective determination step 55, an aspect ratio range determination step 56, and a character height. A line height ratio size determination step 57 and a combined area recognition step 58.

FIGS. 6A to 6E are views for specifically explaining the processing of the blurred character determining means 14.

Next, the operation of the blurred character combination system of the present embodiment having the above-described structure will be described. In addition, here, as shown in FIG. 3 and FIG.
The operation in the case where the processing is performed on the character line image having the italicized characters "nder" will be described.

The character line image storage means 12 stores an arbitrary character line image in the document image to be processed in the English type recognition system. Here, it is assumed that the character line image storage means 12 stores the character line image shown in FIG.

The character segmentation method 11 determines the character area, line height and estimated character pitch for this character line image. FIG. 3B is a diagram showing a set of character areas obtained by the character cutout method 11.

The area information checking means 13 obtains area information by referring to the character line image stored in the character line image storage means 12 and the character area determined by the character cutting method 11. The area information includes the size (width and height) of the character area, the size and coordinates of the contact area in the character area, the distance between the contact areas in two adjacent character areas, and the two adjacent character areas. And the combined character width in.

The process of obtaining the area information by the area information checking means 13 will be described in detail with reference to FIGS. 4 (a) and 4 (b).

FIG. 4A is a view showing the character line image stored in the character line image storage means 12 and the character area obtained by the character cutout method 11 in an overlapping manner.

FIG. 4 (b) is an enlarged view of the area surrounded by the alternate long and short dash line in FIG. 4 (a).

In FIG. 4B, the width of the character area R ₁ is w, and the height of the character area R ₁ is h. These values are values that are basic judgment criteria when judging whether or not the character in the character area R ₁ is a part of the blurred character.

The contact area means an area where black pixels of a character in the character area are in contact with each other on the right side and the left side of the character area (a plurality of contact areas can exist on one side of one character area). From point rs _i to point re _{i in} FIG.
The area up to (line range). Note that i = 0, ...,
2N-1 (N is the number of character areas in the character line image),
In FIG. 4B, the range of 0 ≦ i ≦ 3 is shown.

The coordinates of the point rs _i are represented by (rs _i x, rs _i y) with the origin as the upper left point of the character line image. The coordinates of the point re _i are represented by (re _i x, re _i y) with the origin as the upper left point of the character line image. The relationship of these coordinates is rs _i x = re _i x, and rs _i y ≦ re _i y.

The size of the contact area means the points rs _i and re
_It means the interval rw _{i from i,} and the value obtained by the following equation. rw _i = re _i y−rs _i y (where rw _i ≧ 0)

The coordinates of the contact area are the point rs _i and the point r
The coordinates of e _i are (rs _i x, rs _i y) and (re
_i x, re _i y).

The inter-contact area distance d _{j, which} is the distance between the contact area on the right side of a certain character area R _j and the contact area on the left side of the next character area R _{j + 1} , is obtained by the following equation (where j =
, ..., N-1, and k = 2j-1). In addition,
When there are a plurality of contact areas on one side of one character area, the above distances are obtained for all combinations of contact areas related to adjacent character areas, and the minimum value of those distances is the final value. Is determined as the distance between the contact areas. d _j = 1/2 × [{(rs _{k + 1} x−rs _k x) ² + (r
_{s k + 1 y-rs k} y) 2} 1/2 + {(re k + 1 x-re
_k x) ² + (re _{k + 1} y-re _k y) ² } ^1/2 ]

The combined character width means the length from the left side of one character area to the right side of the next character area, and cw (the left side of the character area R ₁ to the right side of the character area R ₂ in FIG. 4B). Up to)).

The area information checking means 13 obtains and outputs area information including the size of the character area, the size and the coordinates of the contact area, the distance between the contact areas, and the combined character width as described above.

Blurred character determining means 14 determines the area information obtained by the area information examining means 13 and the character segmentation method 1.
By using the line height and the estimated character pitch obtained by 1, the determination process for recognizing the combined area is performed.

That is, the blurred character determining means 14 refers to the area information, the line height of the character line image, and the estimated character pitch to determine whether the character in each character area is a part of the blurred character. It is determined (if a character in a certain character area is a part of a blurred character, the character area belongs to the combined area), and the determination result is output.

The processing by the blurred character determining means 14 will be described in detail with reference to FIGS. 5 and 6. Blurred character determination means 14 performs the following processing.

First, two adjacent character regions are extracted from the character line image in the character line image storage means 12 (sequentially from the first character region) (step 5).
1).

It is judged whether or not the processing of the character line image for one line is completed (step 52).

If the result of this determination is that the processing of the character line image for one line has been completed, the processing ends.

If "the processing of the character line image for one line is not completed" in the judgment of step 52, it is judged whether or not the combined character width of the two character areas to be judged is smaller than the estimated character pitch. (Step 53). Such a determination is made because if the characters in the two character regions to be determined are a part of the blurred character, the combined character width for those character regions is likely to be smaller than the estimated character pitch. ..

If it is determined in step 53 that the combined character width is not smaller than the estimated character pitch, it is unlikely that the character in the character area to be determined is a part of the blurred character, so the process proceeds to step 51. To the processing for the next character area (processing for the character area on the right side of the two character areas that have been processed up to that point and the processing for the next (right) character area).

When the combined character width is smaller than the estimated character pitch in the determination in step 53, it is determined whether or not the contact areas of adjacent sides of the two character areas to be determined are small (2 to be determined). The total value of the sizes of the plurality of contact areas on the adjacent sides of one character area is the determination target value) (step 54). In such a case, in the case of a Gothic character or the like, even if the contact area is close (see the judgment of step 55 described later), it may not be a blurred character. In such a case, the judgment is generally made. This is because the contact area of adjacent sides in the target character area becomes large (see FIG. 6A).

If the contact area is not small in the judgment in step 54, it is unlikely that the character in the character area to be judged is a part of the blurred character, so the processing is returned to step 51 and the next step is executed. Proceed to the processing related to the character area.

If the contact area is small in the determination in step 54, it is determined whether or not the contact areas of the two character areas to be determined are close to each other (step 55). This kind of determination is made when the determinations in steps 53 and 54 are “Yes”.
s ”(even if the combined character width is small, etc.), there are two character regions associated with italic characters as shown in FIG. 6B (2 associated with different adjacent italic characters). Because it is not reasonable to combine (the contact area between two character areas is generally distant) (because it is necessary to prevent accidentally combining the character areas that should not be combined). ..

If it is judged in step 55 that the contact area is not near, it is unlikely that the character in the character area to be judged is a part of the blurred character, so the processing is returned to step 51 and the next step is executed. Proceed to the processing related to the character area.

In the judgment of step 55, "contact area is close"
In this case, it is determined whether or not the aspect ratio of the characters in the determination target character area is within the range of the set value (“vertical length” / “horizontal length” is larger than the set value). (It is determined whether or not the aspect ratio is larger than the set value for both the characters in the two character areas) (step 56). Even if the determinations in steps 53 to 55 are "Yes", it is possible to combine two character regions related to characters such as "-" shown in FIG. It is not valid.

If it is determined in step 56 that the character aspect ratio is not within the set value range (as shown in FIG. 6C), the character in the character area to be determined is a blurred character. Since it is less likely to be a copy, the process is returned to step 51 and the process for the next character area is proceeded to.

If it is determined in step 56 that the range of the character aspect ratio is within the set value range, the ratio of the character height of the character in the character area to be determined to the line height of the character line image is determined. It is determined whether or not it is small (“character height” / “line height” is smaller than a certain set value) (step 57). The determination in steps 53 to 56 is “Y”.
This is because even with "es", it is not appropriate to combine two character areas related to characters such as "I" shown in FIG. 6D.

If it is determined in step 57 that "the ratio of the character height to the line height is not small", it is unlikely that the character in the character area to be determined is a part of the blurred character. The processing is returned to the processing for the next character area.

If it is determined in step 57 that "the ratio of the character height to the line height is small", the determination conditions of steps 53 to 57 (the character in the character area to be determined is determined to be a part of the blurred character). Since all the judgment conditions for performing the judgment are satisfied, the two character areas to be judged are recognized as a combined area (step 58). For example, “u” shown in FIG.
Two character areas related to a blurred character such as or "n" are recognized as a combined area. In the example shown in FIG. 3B, two adjacent character areas indicated by arrows are recognized as a combined area.

In the above determination process, specific numerical values that serve as criteria for determination of "small / large", "near / far", etc. are determined on the basis of the actual character line image examination and the like.

The character area combining means 15 determines, as the combined area, two adjacent character areas in which it is determined that the internal character is a part of the blurred character based on the determination result output from the blurred character determining means 14. The characters are combined (see FIG. 3C), and a set of combined character regions (a set in which the combined region is also recognized as one character region) is passed to the character recognition method 16.

The character recognition system 16 performs a process of recognizing a character related to the set of the character areas (the character line image in the character line image storage means 12 is referred to in this character recognition).

[0054]

As described above, according to the present invention, by combining the character areas related to the blurred character in consideration of the existence of the italic character, the character area related to the blurred character in the normal character and the italic character are blurred. There is an effect that it is possible to accurately combine the character areas related to the blurred character, with the character area related to the character being the common target. Note that, due to such an effect, it is possible to improve the accuracy of character recognition performed by the character recognition method.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention.

FIG. 2 is a diagram illustrating a conventional blurred character combination method.

FIG. 3 is a diagram for explaining the operation of the blurred character combination method shown in FIG.

FIG. 4 is a diagram for explaining a process for obtaining area information by an area information examining unit in FIG.

5 is a flow chart showing a process by a blurred character determining unit in FIG. 1. FIG.

FIG. 6 is a diagram for specifically explaining the processing by the blurred character determining unit shown in FIG.

[Explanation of symbols]

 11 character cutout method 12 character line image storage means 13 area information investigation means 14 blurred character determination means 15 character area combining means 16 character recognition method

Claims (1)

[Claims] 1. An area information investigation means for obtaining area information including a size of a character area, a size and coordinate of a contact area, a distance between contact areas and a combined character width for a character line image to be processed, and the area information investigation. Blurred character determination means for determining whether a character in the character area is a part of blurred character based on the area information obtained by the means, and an internal character based on the determination result by the blurred character determination means. A blurred character combining method, comprising: character region combining means for combining adjacent character regions determined to be part of a blurred character to generate a combined region.