JPH01245376A - Character segmenting device for character reader - Google Patents

Abstract

PURPOSE:To attain printing by means of a non-dropout color by detecting a character frame field based on a formed histogram and executing character segmenting processing in each character based on the character frame field. CONSTITUTION:A character frame field consisting of a character frame partitioned into each character and a document to be read out recored in the character frame are scanned by a scanning part 20 and the character frame field and pattern data corresponding to the character are stored in an image buffer 21. A histogram forming part 22 forms a histogram consisting of projection data in line and row directions from image pattern data. A black character frame field detecting part 23 detects a character frame field based on the histogram and a character segmenting part 24 executes character segmenting processing in each character obtained from the buffer 21 based on the character frame field. Consequently, it is unnecessary to previously print out a character on a document by means of a dropout color.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to a character cutting device for a character reading device that performs character recognition processing for each character.

(Prior Art) Conventionally, in an optical character reading device, a form to be read is optically scanned, and image pattern data obtained by charge conversion processing is stored in an image buffer. Image buff? For example, image pattern data for one sheet (one form) is stored. And image buffer 1
Character-by-character character extraction processing is performed, character by character! i! 1 process is executed.

In the character cutting process, as shown in FIG. 5, a histogram consisting of projection data 10a to 10d corresponding to each image pattern data (character pattern) P in the image buffer is created. Here, each character is recorded in a character frame 11 printed in dropout color on the form.

The character cutting section of the optical character reading device scans a histogram corresponding to, for example, one line of field, detects white data corresponding to the space between character frames, and determines character cutting positions 12a to 12c.

Image pattern data for each character is cut out from the image buffer using the character cutout positions 12a to 12C as references, and transferred to the character recognition section. As a result, the character recognition section performs character recognition processing for each character.

By the way, in the above-mentioned character extraction method, a character frame is printed in advance on the form in order to set the recording position of each character on the form to be read, but the character frame is printed so that it cannot be read. must be printed in dropout color. Is this an image buffer where the text frame is used along with the text? This is because if the data is stored in , the data becomes inconvenient for character extraction processing. However, printing the character frame in dropout color is a major constraint when creating a form, complicating the form creation process and increasing the cost of the form.

Furthermore, since it is necessary to provide a predetermined space between character frames, the entire form becomes large, making it difficult to create a small form.

(Problems to be Solved by the Invention) In the conventional character cutout method, character frames provided in advance on a form need to be printed in a dropout color.

For this reason, there is a drawback that the process of creating the form becomes complicated and the cost of the form increases.

Furthermore, since it is necessary to provide a predetermined space between character frames, it is difficult to create small forms.

The object of the invention is to print with non-dropout colors,
To provide a character cutting device of a character reading device which can reliably cut out characters recorded in the character frames of a document when reading a document having character frames with unnecessary spaces between the character frames. It is in.

[Configuration of the Invention (Means and Effects for Solving the Problems) The present invention provides a system for scanning a character frame field consisting of a character frame partitioned for each character and a document to be read recorded within the character frame. And a pamphlet that stores the character frame field and image pattern data corresponding to the characters? Equipped with memory means. From the image pattern data stored in the buffer memo 9 means, a histogram consisting of projection data in each direction of rows and columns is generated by the histogram generating means. Based on the histogram generated by the histogram generation means, a character frame field is detected by the character frame field detection means. Based on the character frame field detected by the character frame field detection means, the character extraction means performs character extraction processing for each character from the buffer memory means.

With a device having such a configuration, it is possible to eliminate the need to print character frames on a form in advance in a dropout color. Further, since the character extraction process is performed based on the character frames, there is no need to provide any particular space between the character frames.

(Example) The present invention will be described in detail below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a character reading device according to the same embodiment. As shown in FIG. 1, this apparatus includes a scanning section 20, an image buffer 21, a histogram generation section 22, a black character frame field detection section 23, a character cutting section 24, and a character recognition section 25. The scanning unit 20 is a circuit that optically scans a form to be read and outputs image pattern data obtained through photoelectric conversion. Image buff? 21 is
This is a buffer memory that stores one sheet of image pattern data output from the scanning section 20.

Is the histogram generation unit 22 an image buffer? This circuit generates a histogram corresponding to a prespecified area in the image pattern data stored in 21.

The black character frame field detection unit 23 is a circuit that detects a black character frame field based on the pistogram generated by the histogram generation unit 22. Character cutting section 24
extracts character pattern data for each character from the image buffer 21 based on the detected black character frame field and outputs it to the character recognition unit 25. The character recognition EB 25 is a circuit that performs character recognition processing on character pattern data for each character.

Next, the operation of this embodiment will be explained. First, when a document to be read is scanned by the scanning unit 20, one sheet of image pattern data corresponding to the characters and black character frames recorded on the document is stored in the image buffer 21. Here, the black character frame is printed on the form in a non-dropout color, and is a field for dividing a character group into individual characters using vertical and horizontal ruled lines. image buffer 21
As shown in FIG. 3, character pattern data corresponding to characters (in this case, numbers) and image pattern data B corresponding to black character frames are stored.

Next, the histogram generation unit 22 generates the image buffer 2
In the image pattern data stored in 1, a histogram is generated for a prespecified area (the range indicated by the dotted line in FIG. 3). This area designation is given as format control information from the host computer, for example. The pistogram generation unit 22 performs a histogram generation process for the image pattern data of the designated area A, as shown in FIG.

The histogram generation unit 22 scans the area A in the horizontal direction, as shown in step S1 in FIG. 2, and generates a histogram H1 as shown in FIG. 4.

This histogram H1 consists of projection data D1 corresponding to characters in the horizontal direction and projection data D2.03 corresponding to the black character frame field F. Black character frame field detection section 2
3 is projection data 0 corresponding to the black character frame of the horizontal histogram H1 generated by the pistogram generation unit 22.
Based on 2.03, the upper and lower 1 of the black text frame field F
40.41 is detected (step S2).

Next, as shown in FIG. 4, the histogram generation unit 22 sets a predetermined left and right end range within a predetermined area A, and each t! The histogram H2 is scanned vertically with respect to the
, H3 (step 84). The black character frame field detection unit 23 uses projection data D4 . Based on D5, the left and right positions 42 and 43 of the black character frame field F are detected (step S5
).

Here, the black character frame field detection unit 23 detects the upper and lower positions 4.
When detecting 0.41 or the left/right position 42.43, it is determined whether the peaks of the projection data D2 to D5 corresponding to each black character frame satisfy predetermined width and width reference values. It turns out.

In this way, the black character frame field detection unit 23 detects the histograms H1 to H1 generated by the histogram generation unit 22.
The position IF (40 to 43 in FIG. 4) of the black character frame field is detected based on H3 (step 86). Then, the character cutting unit 24 extracts character pattern data for each character (the fourth
Pattern P) in the figure is cut out (step S7), character 2
Output to part 1 25. The character recognition unit 25 performs character recognition processing on the character pattern data for each character extracted by the character extraction unit 24.

Here, the character cutting unit 24 detects the vertical and horizontal edges of the black character frame for each character based on the detected black character frame field, and extracts character frame position data for one character consisting of these edges. create.

The cutting position is determined based on this character frame position data, and character cutting processing is performed for each character.

Incidentally, if a black character frame field is not detected in step S3, character extraction processing using the conventional method (FIG. 5) is executed (step 88).

[Effects of the Invention] As detailed above, according to the present invention, when a form having a black character frame field printed in a non-dropout color such as black is used, image pattern data corresponding to the black character frame field is Using this, it is possible to perform character-by-character extraction processing from a group of characters recorded in a field. Therefore, it is possible to eliminate the step of printing the character frame in a dropout color, thereby simplifying the process of creating a form and reducing the cost of creating it.

Also, in order to cut out characters using the black character frame,
There is no need to specifically secure the space between character frames that is required in the conventional method. As a result, the space on the form can be omitted, making it possible to reduce the size of the entire form.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a character reading device according to an embodiment of the present invention, FIG. 2 is a flowchart for explaining the operation of the embodiment, and FIGS. 3 and 4 are respectively of the embodiment. FIG. 5 is a conceptual diagram for explaining the conventional character extraction method. 20... Scanning unit, 21... Image buffer, 22
. . . histogram generation section, 23 . . . black character frame field detection section, 24 . . . character cutting section, 25 . . . character recognition section. Figure 1 Figure 03 Mouth Figure 4 Prisoner Figure 5

Claims (1)

[Scope of Claims] Image pattern data obtained by scanning a character frame field consisting of a character frame divided into character frames and a document to be read recorded within the character frame and photoelectrically converting it is output. scanning means for scanning, buffer memory means for storing image pattern data corresponding to the character frame field and characters outputted from the scanning means, and scanning means for scanning in each direction of rows and columns from the image pattern data stored in the buffer memory means. a histogram generating means for generating a histogram consisting of projection data of; a character frame field detecting means for detecting the character frame field based on the histogram generated by the histogram generating means; A character cutting device for a character reading device, comprising character cutting means for cutting out characters character by character from the buffer memory means based on a character frame field.