JPH03282895A - Optical character reader - Google Patents

Abstract

PURPOSE:To surely perform read of a character recorded in a state that it comes into contact with a character entry frame or protrudes therefrom by detecting a character pattern from a sheet image obtained as a result of elimination executed by an image eliminating means. CONSTITUTION:To a line buffer 10, a black frame eliminating circuit 20 for eliminating a character entry frame so as to leave a character pattern in a sheet image stored in said buffer is connected. To the black frame eliminating circuit 20, a character segmenting circuit 30 for segmenting a character pattern at every one character from an image from which a frameless part is eliminated is connected. To the character segmenting part 30, a matching circuit 40 for executing a character recognition processing with regard to the segmented character pattern is connected. In such a state, the matching circuit 40 reads out a connected standard pattern from a dictionary store memory 50, executes matching to the character pattern, and the result of matching is outputted as the result of character recognition processing. In such a way, an exact character read result is obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to an optical character reading device for reading characters written on a form or the like.

(Prior Art) Generally, in an optical character reading device that reads characters written on a form, characters recorded in a frame (character entry frame) provided on the form indicating the character entry position are read. There is. In particular, if the frame indicating the position for writing characters on the form is printed with a black frame (non-dropout color), the characters written in the character writing frame may touch the frame or protrude from the frame. It is a condition for entering the text that there is no.

By the way, letters (
The recording of characters on forms in which numbers (numbers) are recorded is increasingly done manually using numbering or rubber stamps.

However, the character pitch of numbering or rubber stamps often does not match the frame pitch of the character entry frame, and because it is done manually, characters may be recorded while touching or protruding from the character entry frame. There is.

When reading such a form on which characters are not accurately recorded, the optical character reading device simply executes character recognition processing.

That is, the frame portion unnecessary for character recognition processing is deleted, and character turns are cut out for each character from the remaining image to perform character recognition.

Therefore, if a character touches or protrudes from the character entry frame, the character stroke that overlaps with the frame will also be deleted. Therefore, the character pattern used for character recognition becomes incomplete, and as a result, accurate reading becomes impossible.

(Problems to be Solved by the Invention) As described above, in the conventional optical character reading device, even if the character touches or protrudes from the character writing frame, the character recognition process takes this into consideration. There was a problem in that accurate reading could not be performed because it was not executed.

The present invention has been made in view of the above points, and it is possible to reliably read characters recorded in non-dropout colors such as black that touch or protrude from the character entry frame. The purpose of this invention is to provide an optical character reading device.

[Structure of the Invention] (Means for Solving the Problems) The present invention detects characters recorded in a character entry frame provided on a sheet of paper to be processed, and converts the characters recorded in the sheet image of the sheet into the same sheet image. This is an optical character reading device that reads a character pattern by cutting out a character pattern from a sheet and performing character recognition processing, and includes a frame image detection means for detecting a frame image of a character entry frame from a sheet image, and a frame image detection means for detecting a frame image of a character entry frame from a sheet image, and a frame image detection means for detecting a frame image of a character writing frame from a sheet image. A removal image detection means for detecting a portion to be removed that does not overlap with a character pattern from a frame image, and an image removal means for removing the portion to be removed detected by the removal image detection means from a sheet image, The present invention is configured to detect a character pattern from a sheet image obtained as a result of removal by the removal means.

(Function) According to such a configuration, even if there is a portion where the character pattern and the frame image overlap due to contact with the character writing frame or characters being recorded protruding from the character writing frame,
Only non-overlapping parts of the frame image are removed. Therefore, a character pattern is cut out from the sheet image from which unnecessary frame images have been removed, and is subjected to character recognition processing with the same quality as the character pattern of characters recorded without contact.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the configuration of the main parts of the optical character reading device according to the same embodiment. In FIG. 1, 10 is a line buffer in which one line of sheet image is stored.

That is, optical scanning is performed on the form to be processed, and a sheet image including images of characters, character entry frames (black frames), etc. is detected. In the line buffer 10,
An image of a line in which characters are recorded (a line in which a character entry frame is provided) is cut out from this sheet image and stored. A black frame removal circuit 20 is connected to the line buffer IO, which removes the character entry frame so as to leave the character pattern in the sheet image stored in the buffer. Connected to the black frame removal circuit 20 is a character cutout circuit 30 that cuts out a character pattern for each character from the image from which the frame portion has been removed. Connected to the character cutting circuit 30 is a matching circuit 40 that performs character recognition processing on the cut out character pattern. The matching circuit 40 includes a dictionary storage memory 5 connected to the matching circuit 40.
The standard pattern is read from 0 and matched with the character pattern.

The dictionary storage memory 50 stores standard patterns classified for each character to be recognized. The matching result of the matching circuit 40 is output as the result of character recognition processing.

Next, the operation of this embodiment will be explained.

Here, an example of the sheet image detected from the form to be processed (a part of the row with a text entry frame) is
Shown in the figure. When a sheet image as shown in FIG. 2 is stored in the line buffer lO, the black frame removal circuit 20
detects the vertical and horizontal projection histograms of the image. An example of the vertical and horizontal projection histogram detected by the black frame removal circuit 20 is shown in FIG. When the projection histogram is detected, the image is cut out excluding the horizontal frame based on the histogram value of the horizontal projection. First, based on the histogram value (black bit number) of the horizontal projection, the frame is determined depending on whether or not it exceeds a preset constant value (TI (1). In Fig. 3, (h , i, j) is determined to be a horizontal frame.Then, by cutting out the image between i and j where characters are recorded, an image excluding the horizontal frame is obtained. This is because if the format of the form (the position where characters are recorded) is known in advance, it is possible to determine from which position the image should be cut out based on the frame position.Usually, in optical character reading devices,
Various information including the format of the form used for character recognition processing is stored in advance.

Next, the black frame removal circuit 20 detects the vertical frame position based on the vertical projection histogram value. This is determined in the same manner as described above, depending on whether the histogram value of the vertical projection exceeds a preset constant value (TH2). Third
In the figure, portions (a to g) are determined to be vertical frames. When a vertical frame is detected in this way, the black frame removal circuit 20
The image of the vertical frame is removed so as not to affect the character pattern of characters that touch or overlap the character entry frame.

Here, when removing the frame image portion, it is determined whether or not to remove it based on the following conditions. That is, as shown in Fig. 4, the dots indicating the frame are CE, CE.
Let LS be the dot on the left line of ) should be determined.

In other words, as shown in FIG. 4(a), (LS-“0”)
When there is a relationship of AND (RS="0"), CE-"0"
Also, as shown in Figure 4(b) to (d), (L
When there is a relationship of S≠“0”) OR (RS≠“0#),
CE-Set as “1” (unchanged).

Based on these conditions, for each dot of each line (a-g) determined to be a vertical frame, (LS-“0”)AN
It is sequentially determined whether there is a relationship D (R3-“0”). As a result, the dots for which the above relationship has been established are removed, and the other dots remain as they are (“
1').

For example, when frame image removal processing is performed on a frame that intersects (overlaps) a character stroke as shown in Figure 5, the pattern shown in Figure 6, which shows an enlarged view within the circled dashed line in the figure, will be changed to the pattern shown in Figure 7. become that way.

That is, the dots indicated by diagonal lines in FIG. 7 that do not overlap with the character strokes are removed, leaving the dots forming the character pattern.

In this way, from each line (a-g) determined to be a vertical frame,
As a result of removing unnecessary dots, an image consisting only of character patterns as shown in FIG. 8 is obtained. The character cutting circuit 30 takes the projection of the image shown in FIG.
-, 6s6e) and cut out each character block, that is, each character pattern. The character cutting circuit 30 transfers the character pattern thus cut out to the matching circuit 40.

The matching circuit 40 performs character recognition processing by matching the character pattern from the character cutting circuit 30 with a standard pattern stored in the dictionary storage memory 50. The character recognition result in the matching circuit 40 is output as the reading result of the characters written on the form.

In this way, even if the image of the character entry frame and the character pattern are mixed and overlap, only the image of the frame can be removed. For this reason,
Since the character pattern subjected to character recognition processing has the same quality as the character pattern of characters recorded without touching or protruding from the character writing frame, accurate results of character recognition processing can be obtained.

In the above embodiment, only the vertical frame portion of the character entry frame is removed by taking into consideration the overlap with the character stroke, but the horizontal frame is also removed using the same processing procedure. It's okay. In this case, it is determined whether or not to remove the line based on the one dot above and below the line determined to be a horizontal frame.

Further, in the above embodiment, the thickness of the frame image is represented by a line of one dot, but even if the thickness of the frame image is several dots, the same method as described above can be used to determine which dots should be deleted. .

[Effects of the Invention] As described above, according to the present invention, characters are recorded on a form provided with a character entry frame printed in non-dropout color (black frame) by touching or overlapping the frame. Even in such a case, since the image of only the frame portion is removed and the character pattern is cut out, it is possible to obtain accurate character reading results.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of the main parts of an optical character reading device according to an embodiment of the present invention, FIG. 2 is a diagram showing an example of a sheet image detected from a form, and FIG.
The figure shows an example of a vertical and horizontal projection histogram.
The figure is a diagram for explaining the removal of the frame image part, Figure 5 is a diagram showing an example of an image to be processed for frame image removal, Figure 6 is a partially enlarged view of Figure 5, and Figure 7 is FIG. 6 is a diagram showing an example in which the frame image has been removed, and FIG. 8 is a diagram showing an example of an image obtained as a result of removing the frame image. 10... line buffer, 20... black frame removal circuit,
30...Character cutting circuit, 40...Matching circuit, 50...Dictionary storage memory

Claims (1)

[Scope of Claims] Characters recorded in a character writing frame provided on a sheet of paper to be processed are detected in a sheet image of the sheet, a character pattern is cut out from the sheet image, and character recognition processing is performed. An optical character reading device for reading characters, comprising: a frame image detection means for detecting a frame image of the character writing frame from the sheet image; and a frame image detected by the frame image detection means to remove a frame image that does not overlap with the character pattern. a removal image detection means for detecting a part to be removed; and an image removal means for removing the part to be removed detected by the removal image detection means from the sheet image; An optical character reading device that detects a character pattern from a sheet image.