JP2000251010A - Document readout method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain the method for reading out a bar code printed at an arbitrary position as to various readout objects on a document by searching the entire document image of a surface image on the document and detecting the position of a bar code line in the form of the vertex coordinates of the quadrangle surrounding the bar code line. SOLUTION: An image input part 102 binarizes the image on the document surface into black-and-white data to generate an original binary image and reduces the original binary image to generate images which differ in resolution. Then the original binary image or the images of different resolution are sent out as document images to be processed to a bar code position detection part 103 and a character extraction and character recognition part 105. The bar code position detection part 103 searches the entire surface of the inputted document image to detect the position of the bar code line. Namely, black runs in a process area are used for the inputted binary image data to extract the circumscribed rectangle surrounding connecting components as a group of black and the vertex coordinates of the four corners of the bar code line are extracted.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a form having various formats, and more particularly to a form reading method for reading character data or bar code data from a form having a bar code printed at an arbitrary position.

[0002]

2. Description of the Related Art Japanese Patent Laid-Open No. 61-75477 discloses a filing method for automatically identifying the type of a document with respect to a fixed document and providing the same classification code as that of a document already stored.
No. pp. 147-64. However, this conventional method does not describe reading a barcode printed at an arbitrary position and storing the barcode by using the barcode reading result.

Japanese Patent Application Laid-Open No. Hei 7-114616 discloses a method of printing a barcode on a slip and reading the barcode to search for a specified slip. However, in this conventional method, the target barcode is limited to the one printed at a fixed position on the voucher, and with respect to reading the barcode printed at an arbitrary position on the form, it is described in this conventional method. Not.

When a document is input after being rotated at an arbitrary angle with respect to the direction set by the scanner, the character line is recognized, the rotation angle is detected based on the recognition result, and the input image is displayed in the correct direction. The correction method is disclosed in Japanese Patent Laid-Open No. 6-1034.
No. 11 discloses this. However, in this conventional method, the rotation angle is detected using character recognition,
No consideration is given to detecting the rotation angle using the direction of the printed barcode.

Also, a form identification number is written or printed in a specific portion of the form, and the form identification is performed by reading the form identification number, and based on the format information (format information) corresponding to the form identification number. For example, Japanese Patent Application Laid-Open No.
-117366, and JP-A-11-008746. However, this conventional method does not consider reading a barcode printed at an arbitrary position and performing character recognition of a form based on format information corresponding to a barcode read result corresponding to a form identification number.

[0006]

SUMMARY OF THE INVENTION In a method for reading characters and barcodes written on a form from a surface image of the form collected by a scanner, an object of the present invention is to print at an arbitrary position on a form in various formats. Reading a barcode. Conventionally, a form is read on a form on which a barcode is printed at a fixed position. However, it has been difficult to read a barcode from a form in which the barcode printing position is arbitrary. Since it is necessary to print a barcode at a fixed position on a form, it has conventionally been difficult to freely design the form of the form. The first object of the present invention is to
An object of the present invention is to propose a method of reading a barcode printed at an arbitrary position in a reading target having various forms.

Conventionally, when a variety of forms are to be processed, a fixed position is set in common for these forms, and a form identification number is written or printed in the fixed position. Read the form identification number in
The corresponding format information is selected, and characters are read according to the format information. For this reason, conventionally, there was a problem that a reading target is limited to a form in which a form identification number is written in a fixed position. A second object of the present invention is to read a barcode printed at an arbitrary position in a reading target having various types of forms, select a format of the form according to the reading result of the barcode, and read characters. Is to propose a method.

Conventionally, for various types of forms,
To store the form image and search for the desired form,
A method is known in which a form identification number printed at a fixed position is read and used as an index for search. In this conventional method, there is a problem that a reading target is limited only to a form in which a form identification number is written in a fixed position. Further, a method of automatically identifying the type of a form by extracting a frame line of the form can be considered. However, it is difficult to automatically identify the type of the form in a form having a similar frame line. A third object of the present invention is to read a barcode representing a type of a form, and a barcode printed at an arbitrary position in a reading target having various forms, and read a result of reading the barcode. It is to propose a method of adding an index for search to the form based on the search.

When reading both a barcode and a character from a form image, for example, when reading a barcode with a form image collected by a scanner having a resolution of 200 dpi,
Depending on the printing accuracy of the barcode, the bar pattern may be deteriorated due to thickening of the pattern or the like, and the barcode may not be read at the resolution. For this reason, it is conceivable to simply increase the resolution of the scanner and, for example, read a barcode from a form image collected at a resolution of 400 dpi. However, if characters are read at the resolution of 400 dpi, image data increases. However, there is a problem that the processing time for character reading becomes long. A fourth object of the present invention is to reduce the processing time required for scanning by selecting and inputting an image resolution appropriate for each processing when performing both barcode reading processing and character reading on a form image. It is to be.

In reading a barcode, an image on the form surface is composed of a large number of pixels (dots). In a simple method of detecting a barcode while sequentially searching for such a large number of pixel data, a pixel There is a problem that the number of data is enormous, and it takes a long processing time to detect the position of the barcode. A fifth object of the present invention is to reduce a processing time required for detecting a barcode position from a form image.

[0011] Further, in a form in which a table is printed on a form to be read conventionally, the form is a limited form in which a barcode is printed at a predetermined fixed position. The correspondence between the frame and the barcode in the frame is a known fixed form, and the form cannot be freely designed. This is a variety of forms including tables, and it is conventionally difficult to output the barcode reading result in association with the read frame information in the form where the barcode is printed at an arbitrary position. Because there was. A sixth object of the present invention is to propose a method for associating a table structure extracted from various types of forms including a table with a barcode printed in a table frame.

When a form is input after being rotated at an arbitrary angle with respect to the setting direction of the scanner, the character line is recognized, the rotation angle is detected based on the recognition result, and the input image is converted in the correct direction. Although a correction method is known, there is a problem that a correct form direction cannot be detected depending on the recognition accuracy of a character line. A seventh object of the present invention is to propose a method for detecting a correct form direction by using a direction of a read barcode line.

[0013]

Means for Solving the Problems In order to achieve the first object, the entire surface of the form image is searched for the surface image of the form on which the barcode is printed at an arbitrary position, and the position of the barcode line is determined. Detected as the coordinates of the vertices of the rectangle surrounding the barcode line.

In order to achieve the second object, the format information for character reading is switched according to the result of reading the barcode, and the characters in the front image of the form are read.

In order to achieve the third object, the form is stored for each type using the bar code reading result as an index.

In order to achieve the fourth object, a plurality of images having different resolutions are generated, and the resolutions of the images are switched so that images having different resolutions are subjected to bar code reading processing and character reading processing. Enter for reading.

In order to achieve the fifth object, a connected component of an input image is extracted, a circumscribed rectangle of the connected component is extracted, a candidate rectangle of a bar is selected according to the size of the circumscribed rectangle, and the candidate rectangle is replaced with surrounding candidates. A barcode line candidate is extracted by fusing with the rectangle, and the barcode line is determined based on the size of the barcode line candidate and the number of rectangles in the line.

In order to achieve the sixth object, the table frame information is extracted, and the bar code lines printed in the table frame correspond to the table frame information and the barcodes in the frame. Attached
A barcode in a frame whose logical arrangement is designated in advance is selected and output.

In order to achieve the seventh object, the bar code lines are extracted assuming that the bar code lines are printed in the horizontal direction, and then the bar code lines are assumed to be printed in the vertical direction. A bar code line is extracted, and both of these extraction results are merged and output as a bar code line position, a form direction is estimated from the direction of the extracted bar code line, and a surface image of the form is determined according to the form direction. Rotate.

[0020]

FIG. 1 is a block diagram of a form reading system according to an embodiment of the present invention. A form to be read in the present invention has one or a plurality of barcodes 1113 and 1114 printed at arbitrary positions as indicated by reference numeral 1100 in FIG. As shown in FIG. 1, a form recognition unit 100 for reading a form on which a barcode is printed at an arbitrary position, an image storage unit 110 for storing an image of the form surface, and a form image are displayed to confirm read data. An image display / data input unit 120 for correction is connected by a network 130. The scanner unit 101 collects an image of a form surface. The scanner unit is formed of a one-dimensional sensor, and is commonly used for character reading and barcode reading. The image input unit 102 generates an original binarized image by binarizing the image on the form surface into black and white, and generates a plurality of images having different resolutions by reducing the original binarized image. Then, the original binary image or an image having a different resolution is sent to the barcode position detecting unit 103 and the character extracting / character recognizing unit 105 as a form image to be processed. The barcode position detecting unit 103 searches the entire surface of the input form image to detect the position of the barcode line. The barcode recognition unit 104 cuts out pixel data based on the input position coordinates of the barcode row and decodes the barcode. The decoded barcode reading result is sent to the control unit 106 and the character extraction / character recognition unit 105. The character extracting / character recognizing unit 105 selects format information indicating a character position, a character type, and the like based on the input barcode reading result, and reads characters written on a form according to the selected format information. The control unit 106 controls the entire form recognition unit, and sends out the form image, the character reading result, and the barcode reading result to the image storage unit 110. The printing unit 107 prints a character or a barcode on the surface of the form according to the result of reading the form. The image storage unit 110 is configured by a large-capacity storage device, and stores form images, character reading results, and barcode reading results.
store.

FIG. 2 is a flowchart showing the process of reading a form. In step 200, a form image is input. In step 201, the input form image is searched to detect a barcode position. Step 202 is a barcode recognition process, in which pixel data of a form image is cut out based on the detected barcode position coordinates and the barcode is decoded. Step 203
Then, it is determined whether the barcode recognition result is an accept or a reject. If the barcode recognition result is an accept, the document (form) identification information corresponding to the barcode reading result is set in step 204. On the other hand, if the barcode recognition result is reject, the type of document (form) is identified from the form image in step 205, and document (form) identification information is set.
In step 205, for example, the type of the form may be identified using a frame line as described in JP-A-61-75477. In step 207, an index serving as a search key is added to the input form image based on the barcode reading result in order to store and search the form image in the image storage unit 110. Next, the process proceeds to a character reading process, and in step 208, format information (format parameters) is set according to the document (form) identification information. Then, in step 209, character extraction and
Perform character identification. Instead of using the format information, a character line at an arbitrary position may be extracted from the form image, and characters may be cut out from the character line to perform character recognition.

FIG. 3 is a flowchart of a process for explaining another embodiment of the form reading, and shows a process of reading the form by switching the resolution. In step 300, an original binarized image which is an image having a resolution collected by the scanner unit 101 is input. Then, in step 301, the original 2
A reduced image is generated from the binarized image. The generation of the reduced image in step 301 is described in, for example, JP-A-63-131274 (Japanese Patent Application Nos. 61-276553 and 1986).
6) Filed on November 21, 2011, inventor: Yoshihiro Shima, Seiji Kashioka,
Junichi Higashino, Applicant: Hitachi, Ltd.) At step 302, a previously set recognition control flag is set. One of the three control methods is selected by the recognition control flag. The three control methods are control methods of executing both barcode recognition and character recognition, executing only barcode recognition, and executing only character recognition. In step 304, a recognition control flag is determined. As shown by the condition 320, in the case of performing both barcode recognition and character recognition, first, in step 305, an original binary image is selected,
The original binary image selected in step 306 is searched to detect the barcode position. Then, in step 307, barcode recognition is performed. Thereafter, in step 308, the reduced image generated in step 301 is selected, and in step 30
In step 9, characters are extracted, and in step 310, character recognition is performed.
On the other hand, as shown by the condition 321, in the case of executing only barcode recognition, an original binary image is selected in step 311;
In step 312, the position of the bar code is detected.
At 13 the barcode is recognized. On the other hand, as shown by the condition 322, when only character recognition is executed, a reduced image is selected in step 314, characters are extracted in step 315, and character recognition is performed in step 316.

FIG. 4 is a block diagram showing a processing procedure of the bar code position detecting section 103. In the image input unit 400, 2
Upon receiving the value image data, the processing area setting unit 401 sets an area where a barcode row is searched. Then, for the binary image data input by the dot run conversion unit 402, the inside of the processing area is converted from pixels (dots) to black runs (a set of black line segments in the scanning line direction). Then, the dirt contact separation unit 403 removes dirt near the bar code or separates characters in contact with the bar. Here, the removal of the short run and the 90-degree rotation of the run are repeated to separate the dirt contact. The rotation of the run by 90 degrees is described in JP-A-63-131274 as conversion of the scan direction of the run. The circumscribed rectangle extracting unit 404 extracts connected components, which are black blocks on the image, using a black run,
Further, a circumscribed rectangle surrounding the connected component is extracted.
Next, the barcode line extraction unit 405 extracts the vertex coordinates of the four corners of the barcode line from the rectangular data.
First, as internal processing of the barcode line extraction unit 405,
A bar rectangle selection unit 405 selects a rectangle corresponding to a bar based on the height and width of the rectangle. FIG. 5 is a view for explaining the selection criteria of the bar rectangle. A circumscribed rectangle 501 surrounding the black pixel block 500 is determined. The condition of the selection criterion is that the height h510 of the circumscribed rectangle 501 is within a predetermined range, that is, the maximum value hmax511 and the minimum value hmin51.
0 and the width w50 of the circumscribed rectangle 501
3 is within a predetermined range, that is, the maximum value wmax 513 and the minimum value w
min512. A circumscribed rectangle that satisfies the above condition is extracted as a possibility of a circumscribed rectangle surrounding the bar. Then, the barcode line candidate extraction unit 412 fuses rectangles corresponding to the bars with those having a short distance to generate an elongated large rectangle, and extracts the fusion rectangle as a barcode line candidate. Next, in the left and right end detection units 413,
The coordinates of the left end and the right end of the extracted barcode line candidate (fusion rectangle) are detected. FIG. 7 is an explanatory diagram of the fusion rectangle. Bar circumscribed rectangles 720, 721, 722, 723, 724
Are fused by the barcode line candidate extraction unit 412 to generate an elongated fused rectangle 700. The coordinates (xmin, ymin) of the upper left vertex 701 and the coordinates (xmax, ymax) of the lower right vertex 702 among the four corner vertices of the fusion rectangle are output from the barcode line candidate extraction unit 412. The left and right end detection unit 413 of the row selects the left end rectangle 720, and the coordinates (xlt, ylt) of the upper left vertex 710 of the left end rectangle 720
And the coordinates (xlb, ylb) of the lower left vertex 711 are detected. Further, the rightmost rectangle 724 is selected, and the rightmost rectangle 724 is selected.
24, the coordinates (xrt, yrt) of the upper right vertex 712 and the coordinates (xrb, yrb) of the lower right vertex 713 are detected. Next, the barcode line selection unit 414 selects a barcode line from the plurality of extracted barcode line candidates based on the height and width and the number of bar rectangles included in the barcode line candidates. FIG. 6 is a view for explaining the selection criteria of the barcode row. The condition for selecting a barcode line is that the height hl601 of the barcode line candidate 600 is within a predetermined range, that is, the maximum value hlmax611 and the minimum value hlmin610.
, The width wl602 is within a predetermined range, that is, the maximum value wlmax 613 and the minimum value wlmin 612, and the number of rectangles included in the barcode line candidate 600 is larger than the predetermined value. Next, in a case where the frame information of the form has already been extracted, the associating unit 406 for associating the barcode row with the frame associates the frame with the barcode line in the frame. In addition, regarding the association with the frame information,
This will be described in detail with reference to FIGS. 8, 10 and 11 described later. The retry determination unit 410 determines whether or not the bar code line has been extracted normally. If the bar code line has not been extracted normally, it is determined that dirt or character contact has occurred, and the dirt contact separation unit 403 re-determines the bar code line. Removes dirt near the code and separates characters in contact with the bar. At retry, change the parameter to remove short runs and remove many runs. In order to detect a vertically long barcode line, a rectangular rotation unit 407 is provided, which rotates the coordinates of the circumscribed rectangle of the connected component by 90 degrees, and converts the vertically long barcode line into a horizontally long barcode line. I do. Then, similarly to the bar code line extraction unit 405 described above, a bar code line extraction unit 408 extracts a bar code line. Furthermore, the bar code row is rotated by 90 degrees in the row rectangle rotation unit 409, and is converted into a vertically long bar code. As described above, whether the direction of the barcode row is the portrait direction or the landscape direction,
The bar code lines are extracted assuming that the bar code lines are printed horizontally so that the bar code lines can be extracted, and then the bar code lines are assumed to be printed vertically. Is extracted, and these two extraction results are combined and output as a barcode line position. FIG. 9 is a flowchart for explaining the processing process of the barcode recognition unit 104. Four corners 710, 7 of the bar code line input in step 900
The vertex coordinates of 11, 712, and 713 are set, and step 9
At 01, an image corresponding to the bar code line is cut out based on the vertex coordinates. Further, the inclination of the image extracted in step 902 is corrected. The inclination angle to be corrected is calculated based on the vertex coordinates of the four corners 710, 711, 712, 713 of the barcode row. For example, Japanese Patent Application Laid-Open No. 63-1
No. 31,274. Next, in step 903, the center line in the bar code row is obtained, and the black and white length is detected from the pixel (dot) data on the center line. FIG.
FIG. 4 is a diagram for explaining a method of detecting a black and white length from pixel data in a barcode row. A center line 1201 is set at a predetermined position with respect to the image 1200 in the barcode row whose inclination has been corrected, and pixel data on the line is accessed. The pixel data on the line has a black level 120 as shown by 1204.
It has a value of 2 and a white level 1203, and is a black level while the line crosses the bar, and a white level while it crosses a blank portion between the bars. The length of black and white is detected from the pixel data 1204, and in step 904, a barcode is decoded based on the length of black and white.

FIG. 10 shows the detection of a table frame and a barcode printed at an arbitrary position from a form having a table format and various forms, and associates the frame with the barcode in the frame. It is a flowchart which shows a process. In the process, first, at step 1000, a form image is input. Then, in step 1001, the horizontal line of the table is extracted, and in step 1002, the vertical line of the table is extracted. And step 10
At 03, a frame is extracted using the horizontal and vertical lines. FIG.
FIG. 9 is a diagram for explaining a process of extracting a frame. Form image 11
In the table 00, a table as shown in 1110 is printed, and character lines 1111 and 1112 and a barcode 11 are printed in each frame of the table.
13, 1114 are printed. The result of extracting the horizontal line from the form image 1100 is shown in 1123. 1120, 1
Horizontal lines H1, H2, and H3 indicated by 121 and 1222 are extracted. Next, the result of extracting the vertical line is set to 1133. Vertical lines V1, V indicated by 1130, 1311, 1132
2, V3 are extracted. Reference numeral 1144 denotes a result of extracting frames, and frames F1, F2, F3, and F4 indicated by 1140, 1141, 1142, and 1143 are extracted. The frames are numbered so as to indicate a relative positional relationship from left to right and from top to bottom. FIG. 8 is a diagram for explaining the format of the barcode row data and the frame data. As shown by 800, the data of the bar code line has the number 801 of the bar code lines printed on the form. And 1
The 802th barcode data is set to 820, and the 820th barcode data is set to 820. As the data of the first barcode row, a vertical / horizontal identification flag 803 indicating the direction of the barcode, a frame number 8 in which the barcode exists,
04, upper left vertex coordinates 805 and 806, lower left vertex coordinates 807 and 808, upper right vertex coordinates 809 and 81 of the barcode line
0, lower right vertex coordinates 811 and 812 are set. on the other hand,
The frame data includes the number 841 of frames present in the form as shown by 840, and the first frame information 850, the m1th frame information 852, and the m2th frame information 853 are set. . In this explanatory diagram, the first barcode row 802 is associated with the m1th frame information 852, and the nth barcode row 820 is associated with the m2th frame information 854. Have been. The result of the frame extraction in step 1004 of FIG. 10 is stored as frame data 840. Then
At step 1004, the barcode position is detected, and at step 1005, the barcode is recognized. The extracted barcode row is stored as barcode data in the format shown in 800. Then, in step 1006, the frame information is associated with the barcode. At this time, the position coordinates of the barcode line are sequentially compared with the position coordinates of the four corners of the frame in the frame information, and a frame having the barcode line inside is detected.
Is performed as shown in FIG. Specifically, 804, 8
Each barcode row is provided with a correspondence in the form of a frame correspondence number indicated by 22. In step 1007, when a specific frame is designated as a logical structure among the frames having a logical arrangement relationship, a barcode line in the frame is selected and output.

FIG. 13 is a flow chart showing another embodiment of the present invention, in which the direction of a form is estimated by using the direction of a bar code line, the direction of the form is corrected, and character reading is performed. is there. In step 1300, an image is input, and in step 1301, a circumscribed rectangle of the connected component is extracted. And
At step 1302, a horizontal bar code line is extracted, and at step 1303, the bar code is decoded. Also, in step 1304, the circumscribed rectangle of the connected component is rotated by 90 degrees,
In step 1305, a vertical barcode row is extracted, and in step 1306, the barcode is decoded. Then
A barcode presence / absence determination 1307 is performed. Here, in the case where there is no barcode 1308, the rotation angle is unknown as shown in step 1312. In the case where the barcode is present in both directions 1311, similarly, the rotation angle is unknown as shown in step 1315. On the other hand, if there is only a horizontal barcode 1309, the direction of the form is estimated to be horizontal and the rotation angle is set as shown in step 1313.
In the case where there is only a vertical barcode 1310, the direction of the form is estimated to be the vertical direction and the rotation angle is set as shown in step 1314. In step 1316, the rotation of the input image is corrected according to the set rotation angle. In step 1317, characters are extracted. In step 1318, characters are recognized, and the result of reading characters is output.

[0026]

According to the present invention, since the position of a barcode can be automatically detected, a form on which a barcode is printed at an arbitrary position can be read. In addition, since barcodes can be printed on forms in various forms, form designing is easy.

Further, according to the present invention, since the form (format parameter) of the form can be set by using the reading result of the barcode at an arbitrary position, it is necessary to print the form type identification information at a fixed position. And the degree of freedom in form setting increases.

Further, according to the present invention, the form image can be stored and searched using the bar code reading result as an index, so that the trouble of specially assigning a search keyword can be omitted.

According to the present invention, since the reading process can be performed by switching the resolution of the form image, the time required for the reading process can be reduced.

According to the present invention, the bar code line position is detected using rectangular data, and the bar code is decoded by cutting out the image data in the line. As compared with the above, the processing amount is small, and the processing time required for reading the barcode can be reduced.

Further, according to the present invention, since a barcode can be associated with a table frame in a form, only barcodes belonging to a frame whose logical arrangement relation is specified are selected and output. There are advantages.

Further, according to the present invention, the direction of a form can be corrected by using the direction of a bar code line at an arbitrary position. In both cases, normal form reading can be executed.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a form reading system according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a process of reading a form.

FIG. 3 is a flowchart of a process for explaining another embodiment of form reading, showing a process of reading a form by switching resolutions.

FIG. 4 is a block diagram illustrating a process performed by a barcode position detection unit 103;

FIG. 5 is a diagram illustrating a selection criterion of a bar rectangle.

FIG. 6 is a diagram illustrating a selection criterion of a barcode row.

FIG. 7 is an explanatory diagram of a fusion rectangle.

FIG. 8 is a diagram illustrating the format of bar code line data and frame data.

FIG. 9 is a flowchart illustrating a process performed by a barcode recognition unit 104;

FIG. 10 is a process of detecting a table frame, detecting a barcode printed at an arbitrary position, and associating the frame with a barcode in the frame, from a form having a table format and various forms. FIG.

FIG. 11 is a diagram illustrating a process of extracting a frame.

FIG. 12 is a diagram illustrating a method of detecting a black-and-white length from pixel data in a barcode row.

FIG. 13 is a flowchart showing a process of estimating the direction of a form by using the direction of a bar code line, correcting the form direction, and reading characters according to another embodiment of the present invention.

[Explanation of symbols]

103: bar code position detection unit, 204: step of setting form identification information corresponding to the bar code reading result, 412: bar code line candidate extraction unit, 407: rectangle rotation unit, 700: fusion rectangle, 860: bar code Correspondence between rows and frames, 1100... A target form, in which one or more barcodes are printed at an arbitrary position.

 ──────────────────────────────────────────────────の Continuing on the front page (72) Inventor Hiroshi Shinjo 1-280 Higashi-Koikekubo, Kokubunji-shi, Tokyo Inside the Hitachi, Ltd. Central Research Laboratory (72) Inventor Kazuki Nakajima 1-280 Higashi-Koikekubo, Kokubunji-shi, Tokyo Hitachi, Ltd. Central Research Laboratory F-term (reference) 5B072 CC04 CC24 CC38 DD16 DD21 FF00

Claims (6)

[Claims] 1. A method for reading a form on a surface image of a form on which a bar code is printed at an arbitrary position, wherein a position of a bar code line whose position is unknown is detected as coordinates of a vertex of a rectangle, and a bar within the rectangle is detected. Read the code,
A form reading method characterized by switching the format information for character reading according to the bar code reading result and reading characters in the front surface image of the form. 2. A method for reading a form on a surface image of a form on which a bar code is printed at an arbitrary position, wherein a position of a bar code line whose position is unknown is detected as vertex coordinates of a rectangle, and a bar within the rectangle is detected. Read the code,
A form reading method, wherein a form image is stored for each type using the read result of the barcode as an index. 3. A form reading method for reading a bar code and an entry character from a surface image of a form on which a bar code is printed at an arbitrary position, wherein a plurality of images having different resolutions are generated, and a bar code reading process is performed. A form reading method characterized by switching the resolution of an image so that images of different resolutions are to be processed in a character reading process and inputting the read image. 4. A method of reading a barcode at an arbitrary position in a form includes extracting a connected component of an input image, extracting a circumscribed rectangle of the connected component, selecting a candidate rectangle of a bar according to the size of the circumscribed rectangle, The candidate rectangle is fused with the surrounding candidate rectangles to extract bar code line candidates, the bar code line is determined based on the size of the bar code line candidate, and the number of rectangles in the line, and the four corners of the bar code line are determined. A bar code reading method comprising: extracting a black and white dot image in a bar code line based on the vertex coordinates of the bar code and decoding the bar code. 5. A form on which a table having a frame is printed,
The surface of a form where the geometric coordinate position of the frame is unknown, but the logical arrangement relationship between the frames is known, and a plurality or a single barcode is printed at an arbitrary position In the form reading method for the image, the table frame information is extracted, and the bar code line printed in the table frame is associated with the table frame information and the barcode in the frame,
A barcode reading method in a form, wherein a barcode in a frame whose logical arrangement is designated in advance is selected and output. 6. Extracting a bar code line assuming that the bar code line is printed in a horizontal direction, and then extracting a bar code line assuming that the bar code line is printed in a vertical direction. Combining the two extraction results and outputting the barcode line position, estimating the form direction from the direction of the extracted barcode line, and rotating the surface image of the form according to the form direction. Read method.