JPH0528307A - Character recognition device - Google Patents

Abstract

PURPOSE:To provide a character recognition device in which a wasteful recognition operation is eliminated by eliminating the character recognition of a tilted original. CONSTITUTION:In the device, a tilt detection circuit 7 detecting the tilt of an original supplied by an automatic paper feeder 1, a comparison circuit 8 preliminarily permitting an operator to set the tilt limit value of the original and comparing the set limit value with the tilt detected by the tilt detection circuit 7, and an original page counter 6 counting the page of the original supplied with the automatic paper feeder 1, are provided. The comparison circuit 8 stores the number of the original exceeding the limit value in a storage part 5, prevents the character of the original from being recognized with a recognition part 4, and controls the automatic paper feeder 1 to supply the next original.

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 for character recognition of general documents.

[0002]

2. Description of the Related Art In recent years, character recognition devices have become widespread, but their recognition rate is an important issue.

Hereinafter, a conventional automatic paper feeder (hereinafter, referred to as AD
A character recognition device provided with (F) will be described with reference to the drawings. FIG. 2 is a block diagram showing the configuration of a conventional character recognition device. In the figure, reference numeral 1 is an ADF in which several tens of originals are set, and the originals are fed one by one to the image reading section as needed. The image reading unit 2 scans a document and stores image data for one document in the image memory 3. The recognition unit 4 reads the image data from the image memory 3 to perform character recognition, and stores the text data converted into a character code in the storage unit 5 such as a hard disk.

[0004]

In such a conventional character recognition apparatus, when a large number of originals are collectively subjected to character recognition processing by using the ADF, even if the originals are greatly tilted due to an ADF feeding error. The recognition process was performed as it was. However, when the manuscript is greatly tilted, the recognition rate is poor and it cannot be used. As described above, in the conventional character recognition device, since the recognition process is performed even on a document that is greatly inclined, useless processing time is generated. In addition, since there is no means for notifying the operator of a recognition error due to a tilt failure, the operator needs to examine the recognition result in detail, which causes an additional burden on the operator.

The present invention solves the above problems, and an object of the present invention is to provide a character recognition apparatus which can automatically exclude a document having a large inclination from the recognition process and can easily know the document. And

[0006]

In order to achieve the above-mentioned object, the present invention comprises an automatic paper feeder for feeding a large amount of originals to a predetermined position in sequence, an image reading unit for reading the originals,
An image memory unit that stores the read image data,
A recognition unit that recognizes the image data of the memory unit, a document skew detection circuit that detects the skew of the document, a counting unit that counts the number of documents fed by the automatic paper feeder, and a recognition unit that recognizes the document data. In a character recognition device having a storage unit for storing the result, a document skew detection circuit and a comparison circuit for comparing the skew of the document detected by the skew detection circuit with a limit value are provided, and the recognition unit has the automatic paper sheet. When the document placed at the predetermined position by the feeding device is tilted beyond the limit value when compared by the comparison circuit, the recognition process is not performed, and the process proceeds to the recognition process of the next document, and the page is processed. The counting unit is a character recognition device that stores the page number for which the recognition process has not been performed in the storage unit.

[0007]

According to the present invention, in the above configuration, when a large amount of originals are processed at a time by using the ADF function, useless recognition processing such as character recognition processing of a greatly inclined original is eliminated by the above-mentioned means. Therefore, the processing time of the entire document set on the ADF can be shortened. Further, the operator can retrieve the page number of the original document for which the recognition processing is omitted from the storage unit and refer to it, and can quickly extract the original document for which the re-recognition processing is required.

[0008]

【Example】

(Embodiment 1) A character recognition apparatus according to an embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a character recognition device according to an embodiment of the present invention. In the figure, 1 is an ADF that sets several tens of originals at once, and 2 is an image reading unit that optically scans the originals and converts optical signals into electrical signals.
3 is an image memory for storing image data, 4 is a recognition unit for performing character recognition processing, 5 is a storage unit such as a hard disk,
6 is a document page counting section for counting the number of documents,
Reference numeral 7 is an original inclination detection circuit for detecting the inclination of the original, and 8 is a comparison circuit for comparing the detected inclination with the inclination set by the operator.

The interrelationship and operation of the above components will be described below. First, a large number of originals are set on the ADF 1 and fed one by one to the image reading unit 2. At this time, the page counting unit 6 counts the number of fed documents. An image of one original sentence is converted by the image reading unit 2 into binary image data and stored in the image memory 3. The document skew detection circuit 7 detects the skew of the document in accordance with the image reading, and sends the result to the comparison circuit 8. After the image of one document is completely stored in the image memory 3, the comparison circuit 8 determines whether or not the recognition process is performed on the document.
If the detected inclination is smaller than the inclination set in advance by the operator, the recognition processing is performed, and if the detected inclination is large, the recognition processing is not performed.

When the recognition process is not performed, the value of the original page count section 6 at that time is stored in the storage section 5 as an unrecognizable page number. Such an operation is repeated until there are no documents set in the ADF. After the character recognition of all the originals is processed, the operator refers to the unrecognizable page number stored in the storage unit 5 to determine whether or not the original is skewed due to a paper feed error of the ADF and which original is generated. Can be found quickly.

The means for detecting the inclination of the original will be described below. First, the color of the surface of the document cover that holds the document is set to a color with low light reflectance. With this setting, it is generally possible to prepare to detect a white document edge.

FIG. 3 is a block diagram showing the structure of the document skew detection circuit, and FIG. 4 is a pattern diagram showing the skew of the document. Figure 4
First, as the edge coordinate data of the original,
4 points on the edge of the document (1, 2, 3 and 4 in the figure)
To detect. Next, in order to detect the inclination of the document, the sizes of the Y coordinates of the upper left corner point and the upper right corner point are compared. as a result,
When the upper right corner point is larger, the inclination is as shown in (a) of FIG. 4, and when the upper left corner point is larger, (b) of FIG.
It means that the inclination is as shown in. In the case of the inclination as shown in (a), the inclination of the original is calculated from the coordinates of points 1 and 4, and in the case of the inclination as shown in (b), the inclination of the original is calculated from the coordinates of points 2 and 3.

The details of the document inclination detection circuit will be described below with reference to FIG. In the figure, 1 is a horizontal dot counter that counts X coordinates, and 2 is a vertical dot counter that counts Y coordinates. In this embodiment, the X-axis direction is set to the right and the Y-axis direction is set to the downward. Reference numerals 3 and 4 are an X coordinate register and a Y coordinate register for temporarily storing the coordinates of the white pixel.
Reference numerals 5, 6, 7 and 8 denote registers, respectively. The register 5 is controlled by the output of the comparator 9 and stores the coordinate value of the white pixel having the minimum horizontal coordinate value.
Is controlled by the output of the comparator 10 to store the coordinate value of the white pixel having the maximum horizontal coordinate value, and the register 7 is controlled by the output of the comparator 11 to have the minimum vertical coordinate value and the minimum horizontal coordinate value. The coordinate value of the white pixel that is the value is stored, and the register 8 is a register that is controlled by the output of the comparator 11 and that stores the coordinate value of the white pixel that has the minimum vertical coordinate value and the maximum horizontal coordinate value. is there. 9, 10
And 11 are comparators, respectively. The comparator 9 compares the X coordinate of the white pixel with the X coordinate X1 of the register 5, and stores the coordinate values of the registers 3 and 4 in the register 5 when smaller than X1, and the comparator 10 White pixel X
When the axis coordinate value is larger than the X coordinate value X2 of the register 6, the coordinate values of the registers 3 and 4 are stored in the register 6, and the comparator 11 determines that the Y coordinate value of the white pixel is smaller than the Y coordinate value Y3 of the register 7. The coordinate values of registers 3 and 4 are stored in register 7, and the coordinate values of registers 3 and 4 are controlled to be stored in register 8 when the Y coordinate value of the white pixel is equal to the Y coordinate value of register 7.

When the image data sent from the image reading unit is a white pixel, it is stored in the X coordinate register 3 and the Y coordinate register 4, and the value at this time is stored in the X coordinate 1 register and the X coordinate register.
The respective values of the coordinate 2 register and the Y coordinate 3 register are compared. If it is smaller than the X coordinate 1 register, the value at that time is stored in the coordinate 1 register. If it is larger than the X coordinate 2 register, the value at that time is stored in the coordinate 2 register. If it is smaller than the Y coordinate 3 register, the value at that time is stored in the coordinate 3 register. If it is the same as the Y coordinate 3 register, the value at that time is stored in the coordinate 4 register.
By repeating this operation, when the scanning of the document is completed, the coordinate value of the number 1 assigned to the white pixel portion around the document shown by the shaded area in FIG. 4 is stored in the coordinate 1 register, and the coordinate value of the number 2 is displayed in the coordinate system. The coordinate value of No. 3 is stored in the coordinate 3 register, and the coordinate value of No. 4 is stored in the coordinate 4 register. Next, the value of the Y coordinate 1 register and the value of the Y coordinate 2 register are compared by the comparator 13. as a result,
When the Y coordinate 2 is large, the values of the coordinate 1 register and the coordinate 4 register are selected by the selector 12 and output to the arithmetic circuit 14. When the Y coordinate 1 is large, the values of the coordinate 3 register and the coordinate 2 register are selected. Select by arithmetic circuit 1
Output to 4. The arithmetic circuit 14 performs the following arithmetic operation to obtain the inclination from the two coordinate values.

Absolute value of (Y ₂ −Y ₁ ) / (X ₂ −X ₁ ) As described above, according to the character recognition device of the embodiment of the present invention,
An automatic paper feeder for sequentially feeding a large amount of documents to a predetermined position, an image reading unit for reading the documents, an image memory unit for storing the read image data, and a recognition unit for recognizing the image data in the memory unit. Character recognition provided with a document skew detection circuit for detecting the skew of the document, a counting unit for counting the number of documents fed by the automatic paper feeder, and a storage unit for storing the result recognized by the recognition unit. The apparatus is provided with a document skew detection circuit and a comparison circuit for comparing the skew of the document detected by the skew detection circuit with a limit value, and the recognition unit detects that the document placed at a predetermined position by the automatic paper feeding device. When comparing by the comparison circuit, if the inclination exceeds the limit value, the recognition process is not performed, the process proceeds to the recognition process of the next document, and the page counting unit performs the recognition process. By using a character recognition device in which a different page number is stored in the storage unit, unnecessary recognition processing can be omitted, and the processing time of the entire document set in the ADF can be shortened. . Further, the operator can quickly find a document that needs re-recognition processing due to an ADF paper feed error, and the burden on the operator is reduced.

[0017]

As is apparent from the above-described embodiments, the present invention stores an automatic paper feeder for feeding a large number of originals to a predetermined position in sequence, an image reading unit for reading the originals, and the read image data. An image memory section, a recognition section for recognizing image data in the memory section, a document skew detection circuit for detecting skew of a document, a counting section for counting the number of documents fed by the automatic paper feeder, In a character recognition device including a storage unit that stores a result recognized by the recognition unit, a document skew detection circuit and a comparison circuit that compares the skew of the document detected by the skew detection circuit with a limit value are provided. The recognizing unit does not perform the recognizing process when the document placed at the predetermined position by the automatic paper feeder is tilted beyond the limit value when compared by the comparison circuit, and the next document is detected. Recognition processing, the page counting unit uses a character recognition device in which the page number that has not been subjected to recognition processing is stored in the storage unit, whereby unnecessary recognition processing can be omitted and set in the ADF. It is possible to reduce the processing time of the entire original document. Further, the operator can quickly find a document that needs re-recognition processing due to an ADF paper feed error, and the burden on the operator is reduced.

[Brief description of drawings]

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

FIG. 2 is a block diagram showing a configuration of a conventional character recognition device.

FIG. 3 is a block diagram showing the configuration of a tilt detection circuit in the character recognition device in one embodiment of the present invention.

FIG. 4 is a pattern diagram showing an arrangement of white pixels on an inclined document edge in the character recognition device in one embodiment of the present invention.

[Explanation of symbols]

 1 ADF (Automatic Paper Feeding Device) 2 Image Reading Section 3 Image Memory Section 4 Recognition Section 5 Storage Section 6 Document Page Counter 7 Document Tilt Detection Circuit 8 Comparison Circuit

Claims (1)

Claim: What is claimed is: 1. An automatic paper feeder for feeding a large number of originals to a predetermined position in sequence, an image reading unit for reading the originals,
An image memory unit that stores the read image data,
A recognition unit that recognizes the image data of the memory unit, a document skew detection circuit that detects the skew of the document, a counting unit that counts the number of documents fed by the automatic paper feeder, and a recognition unit that recognizes the document data. In a character recognition device having a storage unit for storing the result, a document skew detection circuit and a comparison circuit for comparing the skew of the document detected by the skew detection circuit with a limit value are provided, and the recognition unit has the automatic paper sheet. When the document placed at the predetermined position by the feeding device is tilted beyond the limit value when compared by the comparison circuit, the recognition process is not performed, and the process proceeds to the recognition process of the next document, and the page is processed. The character recognition device in which the counting unit stores the page number for which the recognition processing has not been performed in the storage unit.