JP2002150218A - Name card recognition device and name card recognition method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a name card recognition device and a name card recognition method capable of detecting the direction even when pattern characters prepared in advance do not exist on a name card read image and shortening the processing time. SOLUTION: The number of picture elements is counted by a picture element counting section 17e, rows existing on and below a read name card image are detected, the row width is detected by a width detecting section 17f, and the upper and lower sides of the characters described on the name card are determined by a direction determining section 17g. Or, the space from both end sections of the name row to the portion where characters exist is detected, and the upper and lower sides of the characters described on the name card are determined by the size of the space. Or, the end section of the name row is compared with the end section of its proximity row, and the direction having less difference between the positions of both ends is determined as the upper direction of the characters described on the name card.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a business card recognizing device and a business card recognizing method used for an information management device for reading and storing character information written on a business card or the like, and more particularly to detecting a writing direction of a business card to be read. And a business card recognition method.

[0002]

2. Description of the Related Art Heretofore, there have been the following techniques for accurately recognizing characters used for determining a printing direction.

For example, Japanese Patent Application Laid-Open No. Hei 7-21318 discloses that character unit data is cut out from image data read from character information, and a plurality of data having an aspect ratio close to 1 are selected from these character unit data. Character recognition processing is performed on these selected character unit data in different directions such as a vertical direction and a horizontal direction obtained by rotating the character unit data by 90 degrees. Japanese Patent Application Laid-Open No. H10-157572 discloses a technique capable of determining whether the character information string is vertical writing or horizontal writing.

[0004] Japanese Patent Application Laid-Open No. 5-81464 discloses that
An optical character reading device that optically reads characters printed on business cards, a limited number of printed characters, including symbols and characters related to postal and telephone numbers, which are always printed on ordinary business cards, in four directions: up, down, left, right There is disclosed a technology that includes a recognition dictionary for determining the printing direction of a business card and determines the printing direction of a business card by comparing characters cut out in character units with the recognition dictionary.

Further, Japanese Patent Application Laid-Open No. Sho 62-166479 discloses that a longitudinal direction of an input image of a business card is detected by taking the entire projection in the horizontal and vertical directions, and the maximum width of the detected image in the longitudinal direction is detected. Focusing on the narrow projection of each character pattern, cut out each character pattern and compare each cut out character pattern with a pattern corresponding to a prepared rotated character to determine the distinction between vertical writing and horizontal writing of a business card and determine the rotation direction. Discloses a technique for performing recognition processing of an entire input image.

[0006]

However, in each of the above-mentioned prior arts, the direction of a character is detected by pattern matching with pattern characters prepared for four directions, up, down, left, and right. Has to perform four times of processing, and there is a problem that processing time is long.

Further, according to the above-mentioned prior art, a pattern character prepared in advance for the direction determination is not always present in the read business card image, and if the pattern character does not exist in the business card image, the direction is determined. There is a problem that cannot be detected.

SUMMARY OF THE INVENTION The present invention has been made to solve such a problem of the prior art, and can reduce the processing time, so that a pattern character prepared in advance does not exist in a business card read image. An object of the present invention is to provide a business card recognition device and a business card recognition method capable of detecting the direction of a business card image.

[0009]

A business card recognizing device according to the present invention comprises a business card reading means for reading a business card as an image, and an extraction for extracting an image consisting of rows or columns included in the business card image read by the business card reading means. Means, determining means for determining the arrangement of the image consisting of the extracted rows or columns using a preset condition, and direction determining means for determining the direction of the business card image based on the determination result of the determining means, And character recognition means for performing character recognition with the direction determined by the direction determination means as the head.

[0010] According to the above configuration, it is possible to determine the arrangement of the images using preset conditions, and to determine the direction of the business card image (the direction of the characters) based on the result.
Therefore, it is not necessary to perform pattern matching processing in four directions of up, down, left, and right as in the related art to detect the direction, and even if there is no pattern character prepared in advance in the read business card image. It is possible to determine the direction of the business card image.

[0011] The business card recognition device of the present invention is characterized in that the condition for judging the arrangement of the image is a size of a row or a column in a width direction.

According to the above configuration, the number of pixels is calculated and the upper and lower rows (or the right and left columns) are present in the image.
Is detected, and the upper and lower sides (or left and right) of a character written on a business card can be determined based on the width of this row (or column).

The business card recognition device of the present invention comprises a business card reading means for reading a business card as an image, an extraction means for extracting an image consisting of rows or columns contained in the business card image read by the business card reading means, Name specifying means for specifying the row or column to be performed, space extracting means for extracting the spaces at both ends in the business card image of the image of the row or column corresponding to the name specified by the name specifying means, Comparing means for comparing the size of the space at both ends extracted,
A character recognition unit that performs character recognition with the direction of the space determined to be large by the comparison unit as a head.

According to the above configuration, the space from both ends in the name row (or name column) to the portion where the character exists is detected, and the size of this space is used to determine the upper and lower (or left and right) of the character written on the business card. It is possible to make a determination.

The business card recognizing device of the present invention comprises: a business card reading means for reading a business card as an image; an extraction means for extracting an image consisting of rows or columns included in the business card image read by the business card reading means; Name specifying means for specifying the row or column to be performed, end detecting means for detecting both ends of the row or column extracted by the extracting means, and a row or column corresponding to the name specified by the name specifying means. An edge comparing means for comparing the ends of other rows or columns, and a character recognizing means for specifying a character direction based on a comparison result of the end comparing means and performing character recognition. I do.

According to the above configuration, the end of the name row (or name column) is compared with the end of the adjacent row (or adjacent column) of the name. It is possible to determine that the written character is upward (or leftward).

According to the business card recognition method of the present invention, a business card reading step for reading a business card as an image, an extraction step for extracting an image consisting of rows or columns included in the business card image read in the business card reading step, A judging step of judging the arrangement of the image composed of rows or columns using a preset condition; a direction judging step of judging the direction of the business card image based on the judgment result of the judging step; A character recognition step of performing character recognition starting from the determined direction.

According to the above method, it is possible to determine the arrangement of an image using preset conditions, and to determine the direction of the image (the direction of characters) based on the result. Therefore, it is not necessary to perform the pattern matching process in the up, down, left, and right directions four times as in the related art, and it is possible to determine the direction of the image even if there is no pattern character in the read business card image.

The business card recognition method according to the present invention is characterized in that the condition for judging the arrangement of the image is a size of a row or a column in a width direction.

According to the above method, the number of pixels is calculated, and the upper and lower rows (or the right and left columns) in the image are calculated.
Is detected, and the upper and lower sides (or left and right) of a character written on a business card can be determined based on the width of this row (or column).

A business card recognition method according to the present invention includes a business card reading step for reading a business card as an image, an extraction step for extracting an image consisting of rows or columns included in the business card image read in the business card reading step, A name specifying step of specifying a row or a column to be performed, a space extracting step of extracting spaces at both ends in a business card image of an image of a row or a column corresponding to the name specified in the name specifying step, It is characterized by including a comparing step of comparing the sizes of the extracted spaces at both ends, and a character recognizing step of performing character recognition starting from the direction of the space determined to be large in the comparing step.

According to the above method, the space from both ends of the name line (or name column) to the portion where the character exists is detected, and the size of this space is used to determine the upper and lower (or left and right) of the character written on the business card. It is possible to make a determination.

The business card recognition method according to the present invention includes a business card reading step for reading a business card as an image, an extraction step for extracting an image consisting of rows or columns included in the business card image read in the business card reading step, A name specifying step for specifying the row or column to be performed, an end detecting step for detecting both ends of the row or column extracted in the extracting step, and a row or column corresponding to the name specified in the name specifying step. An edge comparison step of comparing the edge of another row or column, and a character recognition step of identifying a character direction based on a comparison result of the edge comparison step and performing character recognition. .

According to the above method, the end of the name row (or the name column) is compared with the end of the adjacent row (or the adjacent column) of the name, and the end of the end determined to have a small positional difference is determined. It is possible to determine that the direction is the upward direction (or the leftward direction) of the character written on the business card.

[0025]

Embodiments of the present invention will be described below with reference to the drawings. Note that the present invention is not limited to this.

FIG. 1 is a diagram showing a photographing state in the business card recognition device according to the present invention.

In FIG. 1, reference numeral 1 denotes a main body cabinet, which has a display unit 2 and an infrared communication unit (not shown). A power supply unit for supplying power to necessary parts such as a display unit 2, an infrared communication unit, a control circuit for controlling an interface, and the like is built in the main body cabinet unit 1.

Reference numeral 3 denotes a camera unit, which includes a CCD camera lens 3a housed in a camera cabinet. The CCD camera lens 3a is configured to be rotatable. In the state shown in the figure, the front side can be photographed, and when rotated, it is possible to photograph within an angle of about 180 degrees, such as the facing side or above. The camera section 3 is detachable from the main body cabinet section 1, and can transmit and receive photographed image data to and from the main body cabinet section 1 by being attached to a connector (not shown) of the main body cabinet section 1. .

Reference numeral 4 denotes a scroll button. By operating the scroll button 4 with a finger, information displayed on the display unit 2 can be moved up, down, left, right, or diagonally. Reference numeral 5 denotes a decision button. By pressing the decision button 5 while displaying an image being photographed, a desired image to be read can be stored in the storage area of the main body cabinet 1. Reference numeral 6 denotes a return button, which can cancel various operations by the user and switch information displayed on the display unit 2 to previous information. Reference numeral 7 denotes a power switch for turning on the power of the main body cabinet unit 1;
Can be turned off. The power for driving the camera unit 3 is supplied from the main body cabinet unit 1. When the power is turned off, the power switch 7 of the main body cabinet unit 1 is pressed, so that the camera unit 3 and the main body cabinet unit 1 are pressed. Power is supplied. Since the scroll button 4 and the return button 6 are not directly related to the present invention, a detailed description will be omitted.

In the business card recognition device of the present embodiment,
An image captured by the CCD camera lens 3a of the camera unit 3 can be read and stored, and the read image can be recognized and the recognition result can be stored.

As shown in FIG. 1, the display unit 2 has a horizontally long shape,
When the camera unit 3 is mounted on the upper part of the apparatus, the CCD camera lens 3a of the camera unit 3 is turned in the direction of the image to be photographed, and a reading instruction of the photographed image is issued. The image is stored in the shooting direction.

For example, when a horizontal and horizontal writing business card such as a business card 8 is photographed and read, as shown in FIG. When you shoot with the camera lens 3a facing,
Since the business card 8 is read and stored in the same direction as that shown in the figure, character recognition can be performed without changing the direction of the stored image.

However, when photographing a vertical and vertically written business card such as the business card 90, the photographing is not performed with the apparatus kept in the horizontal position as in the case of the business card 8.
Photographing is performed with the main body cabinet unit 1 and the camera unit 3 oriented vertically. When the main body cabinet unit 1 and the camera unit 3 are photographed vertically, the mode is set to the vertical mode by a mode setting unit (not shown).

By the way, depending on the user, the apparatus may be used as shown in FIG.
There is a case where the image is rotated rightward as shown in FIG. 1A to make it portrait, and a case where it is rotated leftward and is made vertically as shown in FIG. 1B. The image instructed to be read when the device is rotated to the portrait orientation as shown in FIG.
1 is stored. Further, the image instructed to be read when the apparatus is rotated to the portrait orientation as shown in FIG. 1B is stored in the state of the business card 92.

In order to perform character recognition, first,
It is necessary to know the direction of the characters in the read image.
Therefore, in the present invention, the direction of the character is detected based on the stored image, and the character is recognized. It should be noted that the image may be captured using other reading means such as a scanner. Also, in the following embodiment, a case is described in which a horizontal (horizontal) imaging device is used to read a vertically oriented and vertically written business card to perform character recognition. The present invention is also applicable to a case where a horizontally oriented and horizontally written business card is read using an imaging device. Whether the business card is written vertically and vertically, or written horizontally and horizontally can be manually set.

Hereinafter, means and a method for determining the direction of a stored image will be described in Embodiments 1 to 3. In the following embodiment, the direction is determined for an image composed of rows, but the same processing can be performed for an image composed of columns.

(Embodiment 1) First, claim 1 of the present invention,
Embodiments according to claims 2, 5, and 6 will be described with reference to FIGS.

FIGS. 2A and 2B are diagrams for explaining the direction determination of the business card image according to the first embodiment.

In the present embodiment, the number of pixels of the read image is calculated to detect upper and lower lines (head and tail) in the image, and the width of the line is used to determine whether a character written on a business card is up or down. I do.

Normally, a business card has a company name at the beginning, a name at the center, and an address and a telephone number at the back. Looking at each line, the company name and the phone number are different in character size, and the line width is larger in the company name. From this, it can be seen that in the case of a vertically written and vertically oriented business card, the one with the larger line width is the leading direction of the business card written characters.

As shown in FIG. 1A, the main body cabinet unit 1 and the camera unit 3 are photographed by rotating the camera unit 90 to the right by 90 degrees, and the read image is stored in the state shown in FIG. 2A. . Here, when examining the top and bottom line widths in the image, the top line width 91a is larger than the bottom line width 91b. The image is re-stored by rotating it 90 degrees in the direction, and character recognition is performed based on the stored image information.

On the other hand, as shown in FIG. 1B, the main body cabinet 1 and the camera 3 are moved 90 degrees to the left.
The image taken by rotating the image by a degree and read is stored in a state shown in FIG. Here, the line widths at the beginning and end of the image are examined, and compared to the line width 92a at the beginning, the line width at the end 92
Since b is large, it is determined that the lower part of the image information is the beginning of the business card, the image information is rotated 90 degrees to the left and stored again, and character recognition is performed based on the stored image information.

FIG. 3 is a block diagram showing a detailed operation according to the first embodiment.

In this figure, reference numeral 16 denotes a liquid crystal display circuit which stores dot positions for turning on the liquid crystal as a bit map, and sends signals to the common circuit 14 and the segment circuit 15 as necessary.

Reference numeral 17 denotes a central control unit for controlling input information or output information by various commands. The input control unit 17b of the central control unit 17 performs an operation corresponding to each button by detecting the pressing of the scroll button 4, the determination button 5, and the return button 6. The input control unit 17b sends a signal to the power supply unit 20 by detecting the pressing of the power switch 7 to turn on the power,
F.

Reference numeral 18 denotes a ROM, which is a font information area 18a storing fonts of characters to be displayed on the display unit 2,
And a program area 18b in which a program showing the operation of the central control unit 17 is stored.

Reference numeral 19 denotes a RAM, which is a data storage section 19a for storing various data such as a sentence or figure input by a user through input means (not shown) or an image taken by the camera section 3, and counting pixels of the stored image. The number of pixels table 19b for storing the obtained value together with the number of lines (row direction and column direction), the head width memory 19c for storing the width of the first row of image information from the number of counted pixels, and the end of image information from the number of counted pixels. It has a tail width memory 19d for storing the line width.

Reference numeral 21 denotes an RTC for measuring time using a clock signal from an oscillator (not shown), and outputs the current date and time.

In the camera section 3, a CCD camera lens 3a
The light projected from the device is received by the CCD element 11, and the received light information is amplified by the amplifier 12 and photoelectrically converted. The photoelectrically converted analog signal is output to the A / D converter 13 and is subjected to analog / digital conversion.

The image signal converted into a digital signal as an electric signal by the A / D converter 13 is sent to the central control unit 17.
Is stored in the captured image buffer 17a, is output to the liquid crystal display circuit section 16, and is displayed on the display section 2.

The image signal from the photographed image buffer 17a is also output to the gate 17c.
Is pressed, a signal is sent from the input control unit 17b to the gate 17c, and the gate 17c is opened to release the captured image buffer 1
The image information held in 7a is output to the read image buffer 17d. When the image information is output to the read image buffer 17d, the pixels of the image information held in the read image buffer 17d are counted by the pixel counting unit 17e.
This is a preparation operation for detecting each line width included in the image information. The number of black dots for each line is counted to obtain a total, and the total number of dots corresponding to the number of lines is calculated as RA.
It is stored in the pixel number table 19b of M19.

When the pixel count is completed for all the lines for the image information held in the read image buffer 17d, the pixel count unit 17e switches to the width detector 17f.
Is sent to The width detection unit 17f compares the number of pixels (the number of black dots) between adjacent lines using the variable A and the variable B, has a certain number of pixels or more, and has a fixed ratio of both pixels. If the ratio is one, the line width is counted by the width counter as one line. Then, the line width of one line from the head of the read image buffer 17d is obtained,
It is stored in the head width memory 19c of the RAM 19. Next, a line width of one line is obtained from the end of the read image buffer 17d and stored in the end width memory 19d of the RAM 19.

When the operation of the width detecting section 17f ends, a signal is sent from the width detecting section 17f to the direction determining section 17g. In the direction determination unit 17g, the head width memory 19c of the RAM 19 is used.
Is compared with the tail width memory 19d. If the value of the head width memory 19c is larger than the value of the tail width memory 19d, the image information stored in the read image buffer 17d is rotated rightward by 90 degrees. It is stored again in the read image buffer 17d. In this embodiment, the direction is assigned by rotating the image. However, a method of preparing information of the character recognition unit for each direction and selecting the information in accordance with the character detection direction is also possible.

On the other hand, when the value of the tail width memory 19d is larger than the value of the head width memory 19c, the image information stored in the read image buffer 17d is rotated 90 degrees to the left and the read image buffer 17d is rotated. To be stored again.

When the operation of the direction determining unit 17g is completed, a signal is sent from the direction determining unit 17g to the character recognizing unit 17h. The character recognition unit 17h performs a character recognition operation based on the image information in the read image buffer 17d,
The data is stored in the data storage unit 19a of the AM 19. In the character recognition unit, a character pattern prepared in advance and the read image buffer 1 are read.
Compare with the image information of 7d.

FIG. 4 is a flowchart showing an image reading process according to the first embodiment.

First, in STEP 1, the camera unit 3 is driven to start the image shooting mode. In STEP2, an image is photographed, and in STEP3, the photographed image is stored in the photographed image buffer 17a of the central control unit 17. STE
At P4, the image information held in the captured image buffer 17a is displayed on the display unit 2.

In STEP 5, it is determined whether or not an end instruction (not shown) has given an instruction to end the image shooting mode. When an instruction to end the image shooting mode is given,
Proceeding to STEP 13, end the image shooting mode, and end the processing.

If the end instruction of the image capturing mode has not been issued, it is determined in STEP 6 whether or not the decision button 5 has been pressed. If the determination button 5 has not been pressed, the process returns to STEP 2, and if the determination button 5 has been pressed, the image capturing mode ends in STEP 7,
In STEP 8, the contents of the captured image buffer 17a are stored in the read image buffer 17d.

In the next STEP 9, the read image buffer 1
The number of black dots existing in the pixels in the row direction is counted based on the image information stored in 7d, and the counted value is stored in the pixel number table 19b of the RAM 19 together with the number of lines in the row direction.

In STEP 10, the direction of the image information is determined based on the contents of the pixel number table 19b of the RAM 19 and the contents of the read image buffer 17d, and the image is rotated to a correct position for character recognition based on the determined direction. Is stored again in the read image buffer 17d. The direction determination process in STEP 10 will be described in detail below with reference to FIG.

In STEP 11, the read image buffer 17d
In step 12, character information as a result of the recognition is stored in the data storage unit 19 of the RAM 19.
and the process ends.

Next, the direction determining process of the first embodiment will be described with reference to FIG. FIG. 5 is a flowchart illustrating the direction determination process according to the first embodiment.

First, in STEP 101-1 the RAM
The 19 pixel number table 19b is called. Next, STE
In P101-2, the variable A holds the number of counted pixels of the first line of the pixel number table 19b.
In -3, the count pixel number of the next one line is held in the variable B.

In STEP 101-4, variable A and variable B
Are compared to determine whether the value of the variable A is less than 10% of the variable B. And the value of variable A is 10
If it is not less than%, the value of the variable B is held in the variable A in STEP 101-5, and the process returns to STEP 101-3. Here, the processing from STEP 101-3 to STEP 101-5 is repeated until a line at which the line where the character exists is detected.

On the other hand, if the value of the variable A is less than 10% of the variable B, it is determined that the line at which the line in which the character exists starts is detected, and the width detecting unit 17f is determined in STEP101-6.
Is set to "1" in the width counter. And STEP
In 101-7, the value of the variable B is held in the variable A,
In P101-8, the count pixel number of the next one line is held in the variable B.

In the next STEP 101-9, the value of the variable A is compared with the value of the variable B, and it is determined whether or not the value of the variable B is less than 10% of the variable A. If the value of the variable B is 10% or more of the variable A, "1" is added to the width counter in STEP 101-10, and the value of the variable B is stored in the variable A in STEP 101-11, and STEP 101-8 is executed.
Return to Here, from step 101-8 to step 101-11, until detection of one line in which a character exists is completed.
The process up to is repeated.

On the other hand, if the value of the variable B is less than 10% of the variable A, it is determined that the detection of one line in which the character exists has been completed, and the value of the width counter is stored in the head width memory 19c of the RAM 19 in STEP 101-12. Is stored.

In the next STEP 101-13, the variable A holds the count pixel number of the last line of the pixel number table 19b, and in STEP 101-14, the count pixel number of the previous line is held in the variable B.

In STEP 101-15, the values of the variables A and B are compared to determine whether the value of the variable A is less than 10% of the variable B. And the value of variable A is 1 of variable B
If it is 0% or more, the value of the variable B is stored in the variable A in STEP 101-16, and the process returns to STEP 101-14. Here, until the line where the line in which the character is present starts is detected, STEP101-14 to STEP101-
The processing up to 16 is repeated.

On the other hand, if the value of the variable A is less than 10% of the variable B, it is determined that the line where the line where the character exists is detected, and "1" is set in the width counter of the width detection unit 17f in STEP101-17. I do.

In the next step 101-18, the value of the variable B is stored in the variable A, and in step 101-19, the count pixel number of the previous one line is stored in the variable B.

In STEP 101-20, the values of the variables A and B are compared to determine whether the value of the variable B is less than 10% of the variable A. Then, the value of the variable B is 1 of the variable A.
If it is 0% or more, "1" is added to the width counter in STEP101-21, and the value of variable B is stored in variable A in STEP101-22, and STEP101-19.
Return to Here, until the detection of one line in which a character is present is completed, STEP 101-19 to STEP 101-2 are performed.
The processing up to 2 is repeated.

On the other hand, if the value of the variable B is less than 10% of the variable A, it is determined that the detection of one line in which the character exists has been completed, and the value of the width counter is stored in the tail width memory 19d of the RAM 19 in STEP101-23. Is stored.

In STEP 101-24, the value of the head width memory 19c of the RAM 19 is compared with the value of the tail width memory 19d, and the value of the tail width memory 19d is compared with the value of the head width memory 19d.
It is determined whether the value is larger than the value of c. If the value of the head width memory 19c is larger than the value of the tail width memory 19d, the image information stored in the read image buffer 17D is rotated 90 degrees clockwise in STEP 101-25 and stored again. , And the process ends.

On the other hand, if the value of the tail width memory 19d is larger than the value of the head width memory 19c, STEP1
In step 01-26, the image information stored in the read image buffer 17d is rotated 90 degrees to the left and stored again, and the process ends.

(Embodiment 2) Next, an embodiment according to claims 3 and 7 of the present invention will be described with reference to FIG. 1 and FIGS.

FIGS. 6A and 6B are diagrams for explaining the direction determination of the business card image according to the second embodiment.

In this embodiment, the number of pixels of the read image is calculated, the line corresponding to the name is specified by detecting the line having the largest width, and then both ends of the name line and the character Is detected, and the vertical direction of the character written on the business card is determined based on the size of the both end spaces.

Normally, in a business card, the characters in the name portion are described in the largest size, and the space between the end and the upper part of the name is larger than that of the lower part of the name. From this,
In the case of a vertically written and vertically oriented business card, it can be seen that the larger one of the spaces from the character existing portion to the end of the name line is the upward direction of the business card written character.

As shown in FIG. 1A, the main body cabinet unit 1 and the camera unit 3 are photographed by rotating the camera unit 90 to the right by 90 degrees, and the read image is stored in the state shown in FIG. 6A. . Here, the name line is detected by obtaining the line having the largest width in the image. In this name line, the space between the left portion and the end portion of the name line is smaller than the space 91c between the right portion and the end portion of the name line. Since 91d is larger, it is assumed that the left part of the image is the upward direction of the business card, and the image information is
The image is rotated by 0 degrees and stored again, and character recognition is performed based on the stored image information.

On the other hand, as shown in FIG. 1B, the main body cabinet 1 and the camera 3 are moved 90 degrees to the left.
The image that has been taken by rotating the camera and read is stored in the state shown in FIG. Here, the name line is detected by obtaining the line having the largest width in the image, and the space between the right portion and the end portion of the name line is smaller than the space 92d between the left portion and the end portion of the name line. Since 92c is larger, it is assumed that the right part in the image is the upper direction of the business card, and the image information is rotated 90 degrees to the left and stored again, and character recognition is performed based on the stored image information.

FIG. 7 is a block diagram showing a detailed operation according to the second embodiment. In this figure, parts that perform the same operations as in FIG. 3 described above are given the same reference numerals, and description thereof is omitted.

As in the first embodiment, the photographed image buffer 1
The image signal from 7a is output to the gate 17c,
When the enter button 5 is pressed during image capturing, the input control unit 1
A signal is sent from the gate 7b to the gate 17c, the gate 17c is opened, and the image information held in the captured image buffer 17a is output to the read image buffer 17d. When the image information is output to the read image buffer 17d, the pixels of the image information held in the read image buffer 17d are counted by the pixel counting unit 17e. This is a preparation operation for detecting each line width included in the image information. The number of black dots for each line is counted to obtain a total, and the total number of dots corresponding to the number of lines is stored in the pixel number table of the RAM 19. 1
9b.

In the image information held in the read image buffer 17d, when the pixel counting for all the lines is completed, the pixel counting unit 17e switches to the row extracting unit 17i.
Is sent to

The row extracting unit 17i compares the number of pixels between adjacent lines using the variables A and B, and if the number of pixels is equal to or greater than a certain value and the ratio of the numbers of both pixels is a certain ratio. The row width is counted as one row by a width counter. Then, the line width of each line from the head of the read image buffer 17d is obtained, and RA is calculated together with the corresponding line number.
It is stored in the row table 19e of M19. Similarly, a row extraction operation is performed for all the lines stored in the read image buffer 17d, and the row table 19e of the RAM 19 is completed.

When the operation of the line extracting unit 17i ends, a signal is sent from the line extracting unit 17i to the name area extracting unit 17j. The name area extracting unit 17j detects the widest line from the value in the line table 19e of the RAM 19 and reads the number of the widest line. Next, the line information corresponding to the read line number is read from the read image buffer 17d, and the read line information is sent to the name both ends space extracting unit 17k.

The name both ends space extracting section 17k clears a space counter (not shown), reads the leading dot of the read line information, and counts the value of the spatial counter counted until the read dot becomes a black dot in the leading space of the RAM 19. It is stored in the memory 19f. The name both ends space extracting unit 17k clears a space counter (not shown),
The last dot of the read line information is read, and the value of the space counter counted until the read dot becomes a black dot is stored in the last space memory 19g of the RAM 19.

The two-end space comparison unit 17m compares the value of the head space memory 19f of the RAM 19 with the value of the tail space memory 19g, and sends the comparison result to the direction determination unit 17g.

The direction determining unit 17g re-stores the image information at a correct position based on the result of the comparison. If the value of the tail space memory 19g is larger than the value of the head space memory 19f, the image information stored in the read image buffer 17d is rotated 90 degrees to the left by the direction determination unit 17g and stored again. From the direction determination unit 17g to the character recognition unit 17
h.

On the other hand, when the value of the head space memory 19f is larger than the value of the tail space memory 19g, the direction determining unit 17g rotates the image information stored in the read image buffer 17d 90 degrees clockwise. Then, the signal is sent again from the direction determination unit 17g to the character recognition unit 17h.

The character recognizing unit 17h performs a character recognizing operation based on the image information in the read image buffer 17d, and stores the recognition result in the data storage unit 19a of the RAM 19.

Next, the direction determination processing of the second embodiment in the image reading processing shown in FIG. 4 will be described with reference to FIG. FIG. 8 is a flowchart illustrating a direction determination process according to the second embodiment.

First, in STEP 102-1 a line extracting process is performed to find out in which line (image) the line exists from the read image buffer 17d and store it in the line table 19e of the RAM 19. This row extraction processing will be described in detail below with reference to FIG.

Next, in STEP 102-2, the RAM 1
9 is called. STEP102-3
Then, the widest row is detected from the value of the row table 19e, and the number of the widest row is read in STEP 102-4. In STEP 102-5, STEP
Line information corresponding to the number of lines read in 102-4 is read from the read image buffer 17d. Then, in STEP 102-6, the name both ends space extracting unit 1
The space counter (not shown) existing at 7k is cleared.

Next, in STEP 102-7, in STEP 102-7,
The first dot of the line information read in 102-5 is read, and it is determined in STEP 102-8 whether the read dot is a black dot. If the read dot is not a black dot, "1" is added to the space counter in STEP 102-9, and STEP 102-1
At 0, the next dot is read, and STEP102-8 is executed.
Return to Here, STEP is performed until a black dot is detected.
The processing from 102-8 to STEP 102-10 is repeated.

On the other hand, if the read dot is a black dot, the value of the space counter is stored in the head space memory 19f of the RAM 19 in STEP 102-11, and
At 102-12, the value of the space counter is cleared.

Next, in STEP 102-13, the STE
The last dot of the line information read in P102-5 is read, and it is determined whether or not the dot read in STEP102-14 is a black dot. If the read dot is not a black dot, STEP102-15
In step 10, "1" is added to the space counter.
In step 2-16, the next dot is read out, and the
Return to 2-14. Here, the processing from STEP 102-14 to STEP 102-16 is repeated until a black dot is detected.

On the other hand, if the read dot is a black dot, the value of the space counter is stored in the last space memory 19g of the RAM 19 in STEP 102-17.

Next, in STEP 102-18, RA
The value of the head space memory 19f of M19 and the tail space memory 1
The value of the head space memory 19f is compared with the value of the head space memory 19g to determine whether or not the value of the head space memory 19f is larger than the value of the tail space memory 19g. If the value of the tail space memory 19g is larger than the value of the head space memory 19f, STEP10
In step 2-19, the image information stored in the read image buffer 17d is rotated 90 degrees to the left and stored again, and the process ends.

On the other hand, if the value of the head space memory 19f is larger than the value of the tail space memory 19g, the STE
In P102-20, the image information stored in the read image buffer 17d is rotated rightward by 90 degrees and stored again, and the process ends.

Next, the row extracting process according to the second embodiment will be described with reference to FIG. FIG. 9 is a flowchart illustrating a row extraction process according to the second embodiment.

First, in STEP 102-21, R
The pixel number table 19b of the AM 19 is called. Next, S
In the TEP 102-22, the variable A is set to the pixel number table 19
b holds the number of pixels counted in the first line of the
In 102-23, the count pixel number of the next one line is held in the variable B.

In STEP 102-24, the values of the variables A and B are compared to determine whether the value of the variable A is less than 10% of the variable B. And the value of variable A is 1 of variable B
If it is 0% or more, the value of the variable B is held in the variable A in STEP 102-25, and the process returns to STEP 102-23. Here, until the line where the line in which the character is present starts is detected, the steps from STEP 102-23 to STEP 102-
The processing up to 25 is repeated.

On the other hand, if the value of the variable A is less than 10% of the variable B, it is determined that the line where the line in which the character exists starts is detected, and the line extracting unit 17 is executed in STEP102-26.
i is set to "1". And ST
In EP102-27, the value of variable B is stored in variable A,
In STEP 102-28, the count pixel number of the next one line is held in the variable B.

In the next STEP 102-29, the value of the variable A is compared with the value of the variable B, and it is determined whether or not the value of the variable B is less than 10% of the variable A. And the value of variable B is variable A
If it is 10% or more, "1" is added to the width counter in STEP 102-30, and the value of variable B is stored in variable A in STEP 102-31, and
Return to 28. Here, until the detection of one line in which a character is present is completed, the processing from STEP 102-28 to STEP 102 is completed.
The process up to -31 is repeated.

On the other hand, if the value of the variable B is less than 10% of the variable A, it is determined that the detection of one line in which the character exists has been completed, and the value of the width counter is stored in the line table 19 e of the RAM 19 in STEP 102-32. Remember, STEP10
In step 2-33, the number of lines currently being counted is stored in the row table 19e. Then, the width counter is cleared in STEPs 102-34.

In STEP 102-35, the value of the variable B is stored in the variable A, and in STEP 102-36, the count pixel number of the next one line is stored in the variable B.

Next, in STEP 102-37, it is determined whether or not the processing has been performed on all the lines of the read image buffer 17d. If the processing has not been performed on all the lines, the process returns to STEP 102-23, and the processing from STEP 102-23 to STEP 102-37 is repeated until the processing on all the lines in the read image buffer 17d is completed.

(Embodiment 3) Next, an embodiment according to claims 4 and 8 of the present invention will be described with reference to FIG. 1 and FIGS.

FIGS. 10A and 10B are diagrams for explaining the direction determination of the business card image according to the third embodiment.

In the present embodiment, the number of pixels of the read image is calculated, the line corresponding to the name is specified by detecting the line having the largest width, and then the end of the name line and its end are detected. By comparing the ends of the adjacent rows, it is determined that the aligned one is the upward direction of the character written on the business card.

Normally, in a business card, the character of the name portion is described in the largest size, and the post name may be described in the vicinity of the name with the name and the first character aligned. From this, in the case of vertical writing and vertical business cards,
Comparing the end of the name line with the end of the adjacent line of the name, it can be seen that the direction in which the ends coincide (or the direction in which the difference in position is small) is the upward direction of the business card written character.

As shown in FIG. 1A, the main body cabinet unit 1 and the camera unit 3 are rotated by 90 degrees to the right and photographed, and the read image is stored in the state shown in FIG. 10A. . Here, the name line is detected by finding the line having the largest width in the image, and a line adjacent to the name line is detected. Next, the difference between the position of the end of the name row and the position of the end of the adjacent row is checked. Since the left part of the image is coincident with the end, it is assumed that the left part is the upper side of the business card, The information is rotated 90 degrees clockwise and stored again, and character recognition is performed based on the stored image information.

On the other hand, as shown in FIG. 1B, the main body cabinet 1 and the camera 3 are moved 90 degrees to the left.
The image taken by rotating the image by a degree and read is stored in a state shown in FIG. Here, the name line is detected by finding the line having the largest width in the image, and a line adjacent to the name line is detected. Next, the difference between the position of the end of the name line and the position of the end of the adjacent line is examined. The information is rotated 90 degrees to the left and stored again, and character recognition is performed based on the stored image information.

FIG. 11 is a block diagram showing a detailed operation according to the third embodiment. In this figure, parts that perform the same operations as in FIG. 3 described above are given the same reference numerals, and description thereof is omitted.

As in the first and second embodiments, the image signal from the photographed image buffer 17a is output to the gate 17c. When the enter button 5 is pressed during image photographing, the gate 17c is transmitted from the input control unit 17b. , The gate 17c is opened, and the image information held in the captured image buffer 17a is output to the read image buffer 17d. When the image information is output to the read image buffer 17d, the pixel count unit 17e outputs the read image buffer 17d.
The pixels of the image information held in d are counted. This is a preparation operation for detecting each line width included in the image information. The number of black dots for each line is counted to obtain a total, and the total number of dots corresponding to the number of lines is stored in the RAM 1.
9 is stored in the pixel number table 19b.

In the image information held in the read image buffer 17d, when the pixel counting for all the lines is completed, the pixel counting unit 17e switches to the row extracting unit 17i.
Is sent to

The operation of the row extracting unit 17i is described in the second embodiment.
Since the operation is the same as the operation of the row extracting unit 17i of FIG. 7 described above, the detailed description is omitted here.

When the operation of the line extracting unit 17i ends, a signal is sent from the line extracting unit 17i to the name area extracting unit 17j. The name area extracting unit 17j detects the widest line from the value in the line table 19e of the RAM 19 and reads the number of the widest line. Next, line information corresponding to the read line number is read from the read image buffer 17d, and the read line information is sent to the name area end detection unit 17n.

The name area end detecting section 17n clears a space counter (not shown), reads the first dot of the read line information, and counts the value of the space counter counted until the read dot becomes a black dot in the name of the RAM 19. It is stored in the end space memory 19h.

When the operation of the name area end detection section 17n is completed, a signal is sent to the proximity area end detection section 17p. The proximity area end detection unit 17p reads the line information of the proximity line in the name area from the read image buffer 17d, clears a space counter (not shown), and reads the leading dot of the read line information. Then, the value of the space counter counted until the read dots become black dots is stored in the near end space memory 19i of the RAM 19.

When the operation of the proximity area edge detection section 17p is completed, a signal is sent to the edge comparison section 17q. The end comparison unit 17q compares the value of the name end space memory 19h of the RAM 19 with the value of the near end air memory 19i, and sends the comparison result to the direction determination unit 17g.

The direction determining unit 17g re-stores the image information at a correct position based on the result of the comparison. If the value in the near end space memory 19i and the value in the name end space memory 19h are not the same, the direction determination unit 17g transfers the image information stored in the read image buffer 17d to the left by 90 degrees.
It is rotated and stored again, and a signal is sent from the direction determining unit 17g to the character recognizing unit 17h.

On the other hand, if the value of the name end space memory 19h and the value of the near end space memory 19i are the same, the direction determination unit 17g corrects the image information stored in the read image buffer 17d by 90 degrees to the right. It rotates and stores again, and sends a signal from the direction determination unit 17g to the character recognition unit 17h.

The character recognizing section 17h performs a character recognizing operation based on the image information in the read image buffer 17d, and stores the recognition result in the data storage section 19a of the RAM 19.

Next, the direction determination processing of the third embodiment in the image reading processing shown in FIG. 4 will be described with reference to FIG. FIG. 12 is a flowchart illustrating a direction determination process according to the third embodiment.

First, in STEP103-1, the line number from the read image buffer 17d is determined to be present, and a line extraction process for storing the line in the line table 19e of the RAM 19 is performed. Since the row extraction processing has already been described in the second embodiment with reference to FIG. 9, the description is omitted here.

Next, in STEP 103-2, the RAM 1
9 is called. And STEP1
In step 03-3, the line having the largest width is detected from the value in the line table 19e, and in step 103-4, the number of lines having the largest width is read. In STEP 103-5, line information corresponding to the number of lines read in STEP 103-4 is read from the read image buffer 17d. Then, in STEP 103-6, a space counter (not shown) existing in the name area end detection unit 17n is cleared.

Next, in STEP 103-7, STEP 103 is executed.
The head dot of the line information read in 103-5 is read, and it is determined in STEP 103-8 whether the read dot is a black dot. If the read dot is not a black dot, “1” is added to the space counter in STEP103-9, and STEP103-1
At 0, the next dot is read out and STEP103-8 is executed.
Return to Here, STEP is performed until a black dot is detected.
The processing from 103-8 to STEP 103-10 is repeated.

On the other hand, if the read dot is a black dot, the value of the space counter is stored in the name end space memory 19h of the RAM 19 in STEP 103-11, and
In EP103-12, the adjacent row and the number of the adjacent rows are read from the row table 19e.

Next, in STEP 103-13, the STE
Line information corresponding to the number of lines read in P103-13 is read from the read image buffer 17d. Then, in STEP103-14, a space counter (not shown) existing in the proximity area end detection unit 17p is cleared.

Next, in STEP 103-15, the STE
The head dot of the line information read in P103-13 is read, and it is determined whether the read dot is a black dot in STEP103-16. If the read dot is not a black dot, STEP 103-1
In step 7, "1" is added to the space counter,
At 03-18, the next dot is read out and STEP1 is executed.
Return to 03-16. Here, the processing from STEP 103-16 to STEP 103-18 is repeated until a black dot is detected.

On the other hand, if the read dot is a black dot, the value of the space counter is stored in the near end space memory 19i of the RAM 19 in STEP103-19.

Next, in STEP 103-20,
The value of the name end space memory 19h of the RAM 19 is compared with the value of the close end space memory 19i to determine whether the value of the name end space memory 19h matches the value of the close end space memory 19i. I do. And the name end space memory 19
If the value of h does not match the value of the near end space memory 19i, the image information stored in the read image buffer 17d is rotated 90 degrees to the left in Step 103-21 and is stored again, and the process ends. I do.

On the other hand, if the value in the name end space memory 19h and the value in the close end space memory 19i match, S
In the TEP 103-22, the image information stored in the read image buffer 17d is rotated 90 degrees clockwise and stored again, and the process is terminated.

[0137]

As described in detail above, according to the present invention,
The number of pixels of the read image is calculated to detect upper and lower rows (or left and right columns) in the image, and the upper and lower (or left and right) of a character written on a business card is determined by the width of this row (or column). Can be determined.

Alternatively, it is possible to detect a space from both ends of the name line (or name column) to the portion where the character exists, and determine the upper and lower (or left and right) of the character written on the business card based on the size of this space. it can.

Alternatively, the end of the name row (or the name column) is compared with the end of the adjacent row (or the adjacent column) of the name, and based on the comparison result, the direction in which the position difference is small is described on the business card. It can be determined that the character is upward (or leftward).

Conventionally, four directions (up, down, left, and right) had to be detected. However, according to the present invention, the traveling direction of the characters on the business card can be determined, so only two direction determinations (character recognition) are required. Processing time can be reduced. For example, in a horizontally long business card, if a large space is detected at the right end of the part where the name is written, or if the start position is close to the next row, the information on this business card is written from right to left. It can be seen that is printed. At this time, it is not possible to detect whether it is a vertical business card or a horizontal business card,
At least in the case of a vertical business card, it can be assumed that the right side is up, and in the case of a horizontal business card, it is possible to assume that the lower side is up. The final direction can be detected by adopting the direction with the smaller position difference). Therefore, 2
The direction of a character can be detected by performing character recognition twice. Further, when the user selects the vertical type or the horizontal type, the character direction can be automatically specified only by specifying the upper direction of the business card.

Furthermore, the pattern characters prepared in advance for the direction determination are not always present in the business card read image, and the direction determination may not be performed. Thus, more business card read images can be handled.

[Brief description of the drawings]

FIGS. 1A to 1C are diagrams showing a shooting state of a business card recognition device according to the present invention.

FIGS. 2A and 2B are diagrams for explaining direction determination of a business card image according to the first embodiment.

FIG. 3 is a block diagram illustrating a detailed operation performed by the business card recognition device according to the first embodiment.

FIG. 4 is a flowchart illustrating a business card recognition process according to the present invention.

FIG. 5 is a flowchart illustrating a direction determination process according to the first embodiment.

FIGS. 6A and 6B are diagrams for explaining direction determination of a business card image according to the second embodiment.

FIG. 7 is a block diagram illustrating a detailed operation of the business card recognition device according to the second embodiment.

FIG. 8 is a flowchart illustrating a direction determination process according to the second embodiment.

FIG. 9 is a flowchart illustrating a row extraction process according to the second and third embodiments.

FIGS. 10A and 10B are diagrams for explaining a direction determination of a business card image according to the third embodiment.

FIG. 11 is a block diagram illustrating a detailed operation of the business card recognition device according to the third embodiment.

FIG. 12 is a flowchart illustrating a direction determination process according to the second embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Main body cabinet part 2 Display part 3 Camera part 3a CCD camera lens 4 Scroll button 5 Enter button 6 Return button 7 Power switch 8, 90, 91, 92 Business card 11 CCD element 12 Amplifier 13 A / D converter 14 Common circuit 15 segment Circuit 16 Liquid crystal display circuit unit 17 Central control unit 17a Captured image buffer 17b Input control unit 17c Gate 17d Read image buffer 17e Pixel count unit 17f Width detection unit 17g Direction determination unit 17h Character recognition unit 17i Line extraction unit 17j Name area extraction unit 17k Name end space extraction unit 17m End space comparison unit 17n Name area end detection unit 17p Proximity area end detection unit 17q End comparison unit 18 ROM 18a Font information area 18b Program area 19 RAM 19a Data storage unit 19b Pixel Number table 19c Start width memory 19d End width memory 19e Row table 19f Start space memory 19g End space memory 19h Name end space memory 19i Proximity end space memory 20 Power supply 21 RTC 91a, 92a Head line width 91b, 92b End space Line width 91c, 92c Space between right and end of name line 91d, 92d Space between left and end of name line

Claims (8)

[Claims] 1. A business card reading means for reading a business card as an image, an extraction means for extracting an image consisting of rows or columns included in the business card image read by the business card reading means, and an image consisting of the extracted rows or columns Determining means for determining the arrangement of the business card image using a preset condition; direction determining means for determining the direction of the business card image based on the determination result of the determining means; And a character recognition means for performing character recognition. 2. The business card recognition device according to claim 1, wherein the condition for determining the arrangement of the image is a size of a row or a column in a width direction. 3. A business card reading means for reading a business card as an image, an extraction means for extracting an image consisting of rows or columns included in the business card image read by the business card reading means, and specifying a row or column corresponding to the name. Name specifying means, and a space extracting means for extracting spaces at both ends in a business card image of a row or column image corresponding to the name specified by the name specifying means; A business card recognition device, comprising: comparison means for comparing sizes; and character recognition means for performing character recognition starting from the direction of the space determined to be large by the comparison means. 4. A business card reading means for reading a business card as an image, an extraction means for extracting an image composed of rows or columns included in the business card image read by the business card reading means, and specifying a row or column corresponding to the name. Name specifying means, end detecting means for detecting both ends of the row or column extracted by the extracting means, a row or column corresponding to the name specified by the name specifying means, and another row or column 1. A business card recognition device comprising: an end comparison means for comparing the end parts; 5. A business card reading step for reading a business card as an image, an extraction step for extracting an image consisting of rows or columns included in the business card image read in the business card reading step, and an image consisting of the extracted rows or columns. A determination step of determining the location of the business card image using a preset condition; a direction determination step of determining the direction of the business card image based on the determination result of the determination step; And a character recognition step of performing character recognition. 6. The business card recognition method according to claim 5, wherein the condition for judging the arrangement of the image is a size of a row or a column in a width direction. 7. A business card reading step for reading a business card as an image, an extraction step for extracting an image consisting of rows or columns included in the business card image read in the business card reading step, and specifying a row or column corresponding to the name. A name specifying step, a space extracting step of extracting a space at both ends in a business card image of an image of a row or a column corresponding to the name specified in the name specifying step, A business card recognition method, comprising: a comparison step of comparing sizes; and a character recognition step of performing character recognition starting from a direction of a space determined to be large in the comparison step. 8. A business card reading step for reading a business card as an image, an extraction step for extracting an image consisting of rows or columns included in the business card image read in the business card reading step, and specifying a row or column corresponding to the name. A name specifying step, an end detecting step of detecting both ends of the row or column extracted in the extracting step, a row or column corresponding to the name specified in the name specifying step, and another row or column And a character recognition step of identifying a character direction based on a comparison result of the edge comparison step and performing character recognition.