JPH05143776A - Character recognizing device - Google Patents

Abstract

PURPOSE:To provide the character recognizing device exactly performing the segmenting and recognition in a short time even when the character width is fluctuated due to the mixing of the special characters such as alphabets in the general characters with the same character width such as KANJI (Chinese character) and KANA (Japanese syllabary writing). CONSTITUTION:The external rectangular of the black picture element connection component is extracted from a picture inputted from an input part 1 by an external rectangular extraction part 2. The basic rectangular is prepared by a basic rectangular preparation part by unifying the external rectangular overlapped in the vertical direction in the extracted external rectangular. Based on the shape position of the basic rectangular, a character string discriminating part 6 discriminates whether the character string is the general or special character string. In the case of a general character string, the basic rectangular is unified based on the estimated reference character width, performing the recognition. In the case of a special character string, the basic rectangular is recognized separately without unification. Thus, the repetition of segmenting and recognition is reduced to the minimum, realizing the improvement of the segment accuracy and the shortening of the processing time.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition device which reads a document as image data from an image reading device such as a scanner and cuts out a character pattern for each character and recognizes it.

[0002]

2. Description of the Related Art Generally, Japanese documents have uniform character widths such as kanji and kana characters. Utilizing this fact, conventionally, the standard character width is estimated from the input character image data, and a character pattern is cut out from the estimated standard character width by a rectangle that circumscribes one character at a time. It recognized and recognized the character pattern. When the recognition result was rejected because there was no corresponding character, the character was cut out again, the position and unit were changed, recognition was performed, and the process was repeated until the recognition result was accepted.

[0003]

However, in a document in which characters other than kana-kanji such as alphanumeric characters are mixed, the character width varies, and it is necessary to cut out and repeat recognition many times, resulting in an error in cutting out, an increase in processing time, etc. Was the cause of.

The present invention has been made in consideration of the above problems, and an object of the present invention is to provide a character recognition device capable of performing high-speed character recognition of a character string in which characters other than Japanese are mixed.

[0005]

In order to achieve the above object, the input character image data is cut out in a predetermined unit,
In a character recognizing device for recognizing, a circumscribing rectangle extracting means for extracting a circumscribing rectangle for each black pixel connected component unit of character image data and a vertical rectangle of the extracted circumscribing rectangles are integrated to form a basic rectangle. The character string determining means includes a rectangle creating means and a character string determining means for determining whether the character string is a general character string, a kana character such as an alphabetic character, or a special character string other than kanji, depending on the shape and position of the basic rectangle. It is cut out and recognized according to the character string determined by.

[0006]

The character recognition device of the present invention having the above-mentioned structure extracts the circumscribing rectangle in the minimum unit in which the black pixel is connected to the character image data, and the adjacent circumscribing rectangles are integrated to form the basic rectangle, and the width of the basic rectangle. The character string determination means selects the general character string, the kana character, and the special character string other than the Chinese character from the size and position such as the position. In this way, the basic rectangles are integrated according to the properties of the selected character strings to determine the character cutting range and perform character recognition. Improvements and reductions in processing time can be realized.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is an overall block diagram of a character recognition device in an embodiment of the present invention. As constituent elements, 1 is an input unit for inputting image data from an image reading device such as a scanner, 2 is a circumscribing rectangle extracting unit as circumscribing rectangle extracting means for extracting a circumscribing rectangle of a black pixel connected component from the input image data, 3 Is a basic rectangle creating unit as a basic rectangle creating unit that creates a basic rectangle by integrating circumscribing rectangles that overlap each other in the vertical direction, 4 is a basic rectangle creating unit, and a basic rectangle integrating unit that cuts out a character pattern, and 5 is a cut out character A character recognizing unit for recognizing a pattern, 6 is a basic rectangle integrating unit 4 to a character recognizing unit 5
In the processing, the character determination unit that determines whether the character string is a general character string or a special character string based on the width and position of the basic rectangle, and 7 is an output that outputs the recognized character code obtained by the character recognition unit 5. It is a department.

With respect to the character recognition apparatus of this embodiment having the above-described structure, the operation of the constituent elements in relation to each other will be described below with reference to the overall flow chart shown in FIG.
In step s1, image data is input from the image reading device. In step s2, the circumscribed rectangle of the black pixel connected portion of the input image data is extracted by the circumscribed rectangle extraction unit 2. With the image data of the character string shown in FIG. 6a, a circumscribed rectangle as shown in FIG. 6b is extracted. In this case, the black pixel connected component has one circumscribing rectangle, and for example, the character “be” has a circumscribing rectangle at the “he” and the dakuten.

In step s3, a character line is cut out from the extracted circumscribed rectangle. In step s4, the circumscribed rectangles that overlap in the vertical direction are integrated by the basic rectangle creation unit 3 in the cut out character line. This integrated circumscribed rectangle is called a basic rectangle. If the circumscribing rectangle shown in FIG.
A basic rectangle such as c is created. In this case, one basic rectangle is created for the character "be" in the above example. The basic rectangle codes a to z and A to E are assigned to the respective basic rectangles.
Is attached.

In step s5, the standard character width of the cut character line is estimated. The maximum value of the height of the basic rectangle included in the character line is the standard character width. This is because, in general, in Japanese documents, most characters are 1: 1 in height and width.

In step s6, the character string determination unit 6 determines whether the basic rectangle included in the character line is a general character string or an English character string. The determination of the character string will be described later in detail. In step s7, it is determined whether the basic rectangle being processed is a general character string or an English character string. If it is a general character string, the process proceeds to step s8, and if it is an English character string, the process proceeds to step s9. Step 8
Then, the general character is cut out and recognized, and step s9
Then, the English characters are cut out and recognized. The extraction and recognition of general characters and English characters will be described in detail later.

In step s10, it is determined whether or not the processing has been completed for all the basic rectangles included in the cut out character line. If the processing has been completed, the process proceeds to step s11,
If not, the process returns to step s7 to repeat the process.

In step s11, it is judged whether or not all the character lines included in the image data have been cut out,
If it's over, it's over, if it's over, step s
Return to 3 and repeat the process.

Next, the operation of the character string determination processing in step s6 will be described with reference to the flowchart shown in FIG.

In step s12, the line upper end coordinate 1up is set to the uppermost end coordinate of the basic rectangle included in the character line, and the line lower end coordinate 1low is set to the lowermost end coordinate. In step s13, the height 1h of the character line is set to 1low-1up + 1. In step s14, the upper end coordinates up of the alphabetic characters such as a, c, e are set to 1up + 1h × T1 (here, T1 =
In 0.4), the lower end coordinate low is set to 1 low.

At step s15, the width w of the English character is set to the standard character width × T2 (here, T2 = 0.6).
If the upper end coordinates and the lower end coordinates of the basic rectangle of FIG. 6 are those shown in FIG. 7, then 1up = 97 1low = 163 1h = 163-97 + 1 = 67 up = 97 + 67 × 0.4 = 124 low = 163w = 63 × 0.6 = 38 (standard character width =
63).

In step s16, the width of the basic rectangle being processed is w or less, and the distance to the next basic rectangle is w × T3.
If (here, T3 = 0.4) or less, step s
If not, the process proceeds to step s19. If the width of the basic rectangle in FIG. 6c and the distance to the next basic rectangle are as shown in FIG. 7, the width is 38 or less and the distance is 1
Basic rectangles b, j, k, l, m, n, o, p of 5 or less
q, s, w, x, y, z and A shift to step s17.

In step s17, the upper end coordinates of the basic rectangle are up-1h × T3 (here, T3 = 0.2) to u.
is included in the range of p and is low-1h × T4
If it is included in the range of (here, T4 = 0.2) to low, the process proceeds to step s18. Other than that, no processing is performed and it is undecided. In the basic rectangle of FIG. 6c, in the range of the upper end coordinates 111 to 124 of the basic rectangle shown in FIG.
The basic rectangles j, l, n, and p whose bottom coordinates are included in the range of 150 to 163 move to step s18.

In step s18, the basic rectangle whose character string has not been determined is determined as an English character string. For example, in the case of the basic rectangle j of FIG. 6c, the undetermined basic rectangle is j, and in the case of the basic rectangle l, the undetermined basic rectangles are k and l, which are determined to be English character strings, respectively.

In step s19, the basic rectangle being processed is determined as a general character string. In step s20, if the character string of the basic rectangle determined immediately before is an English character string, step s
If the character string is a general character string, the process proceeds to step s22. For example, in the case of the basic rectangle c in FIG. 6c, since b is undecided, the basic rectangle determined immediately before is a and is a general character string, and in the case of the basic rectangle r, q is undetermined, so immediately before. The determined basic rectangle is p, which is an English character string.

In step s21, the basic rectangle whose character string has not been determined is determined as an English character string. In the case of the basic rectangle r, the undetermined basic rectangle is q, which is determined to be an English character string. In step s22, the basic rectangle whose character string has not been determined is determined as a general character string. In the case of the basic rectangle c, the undetermined basic rectangle is b, which is determined as the general character string.

In step s23, it is determined whether or not the processing has been completed for all the basic rectangles included in the cut out character line. If all the basic rectangles have been processed, the processing ends. If not completed, the processing returns to step s16 to repeat the processing. With the above processing, the basic rectangles j, k, l, m. It is determined that n, o, p, and q are English character strings.

Next, the operation of the general character cutout and recognition processing in step s8 will be described with reference to the flowchart shown in FIG. In step s24, 0 is set as the initial value of the number n of basic rectangles to be integrated. In step s25, if the integrated width of the n + 1 basic rectangles is smaller than the estimated standard character width, the process proceeds to step s26. If not, the process proceeds to step s27. In step s26, n is set to n + 1.

In step s27, if n is 1 or more, the process proceeds to step s28, and if not, the process ends. In step s28, the n basic rectangles are integrated into a cut-out character pattern. In step s29, the cut-out character pattern is recognized. In step s30, if the recognition result is accepted, the process ends, and if not, the process proceeds to step s31. In step s31,
Set n to n-1, return to step s27 and repeat. The basic rectangle b in FIG. 6c is two basic rectangles (b to c).
Integrate and recognize. The basic rectangle z integrates two basic rectangles (z to A), the recognition result is not accepted, and recognition is performed again with one basic rectangle (z alone).

Next, in step s9, the English characters are cut out,
The operation of the recognition process will be described with reference to the flowchart shown in FIG. At step s32, 1 is set as the initial value for the number n of basic rectangles to be integrated. In step s33, the integrated width of the n basic rectangles is
If it is less than the standard character width, the process proceeds to step s34, and if not, the process ends. In step s34, n basic rectangles are integrated into a cut-out character pattern. In step s35, the cut-out character pattern is recognized. In step s36, if the recognition result is accepted, the process ends, and if not, the process proceeds to step s37. In step s37, n is set to n + 1, and step 33
Return to and repeat. The basic rectangle j in FIG. 6c is recognized by one basic rectangle (j alone).

In the present embodiment, the character string determination is a general character string and an English character string, but it goes without saying that the English character string may be treated as a special character string other than kana and kanji.

[0027]

As is apparent from the above description, the character recognition device of the present invention integrates the circumscribing rectangle extracting means for extracting the circumscribing rectangle of the black pixel connected component and the circumscribing rectangle overlapping in the vertical direction to form a basic rectangle. By providing a basic rectangle creating means and a character string determining means for distinguishing between a general character string and a special character string based on the shape and position of this basic rectangle, the cutting range of the next character is determined according to the character string. Therefore, it is an excellent character recognition device that can minimize the repetition of clipping and recognition, improve the clipping accuracy, and shorten the processing time.

[Brief description of drawings]

FIG. 1 is a block diagram showing an overall configuration of a character recognition device according to an embodiment of the present invention.

FIG. 2 is a flowchart showing the overall operation of the character recognition device of the same embodiment.

FIG. 3 is a flowchart showing an operation of a character string determination part of the embodiment.

FIG. 4 is a flowchart showing an operation of a general character cutout and recognition part of the embodiment.

FIG. 5 is a flowchart showing the operation of the segmentation and recognition part of English characters in the same embodiment.

6A to 6C are pattern diagrams showing an example of character image data for explaining the operation of the character recognition device of the embodiment and a cutout unit thereof.

FIG. 7 is an explanatory diagram showing the nature of a basic rectangle created by the character recognition device of the embodiment.

[Explanation of symbols]

 1 input unit 2 circumscribed rectangle extraction unit 3 basic rectangle creation unit 4 basic rectangle integration unit 5 character recognition unit 6 character string determination unit 7 output unit

Claims (1)

[Claims] 1. A circumscribing rectangle extracting means for extracting respective circumscribing rectangles of black pixel connected components from an input image and the circumscribing rectangles vertically overlapped with the circumscribing rectangles extracted by the circumscribing rectangle extracting means are integrated. , A basic rectangle creating means for creating a basic rectangle, and the shape and position of the basic rectangle,
A character string determination unit that determines whether the character string is a general character string or a special character string, and a character pattern that determines the range in which the basic rectangle is integrated according to the character string determined by the character string determination unit. A character recognition device comprising: a cutting-out means for cutting out a character pattern;