JP5530895B2 - Symbol code recognition apparatus and symbol code recognition program - Google Patents

Description

  The present invention relates to a symbol code recognition apparatus for recognizing a symbol code such as a barcode, and a symbol code recognition program for causing a computer to recognize the symbol code.

  2. Description of the Related Art Conventionally, a symbol code recognition device such as a barcode reader that reads a symbol code such as a barcode and recognizes the contents of the symbol code is known. Conventionally, various techniques for correctly recognizing the read symbol code have been proposed.

JP 2006-330906 A JP 10-49622 A JP-T 9-506197 JP 2008-21090 A JP 2000-357205 A

  Conventionally, the existence of a symbol code is known in advance, and a proposal for correct recognition when only the symbol code area is read can be found. However, for example, when a symbol code is printed or pasted somewhere on an A4 size form and various characters and figures other than the symbol code are recorded on the form, the position of the symbol code is changed. The problem is how to specify it quickly and accurately. In particular, it is necessary to specify the position of the symbol code at high speed and accurately even from an image with poor image quality that is soiled due to crushing or rubbing due to noise. It is also necessary to accurately recognize the contents of such a bad image quality symbol code.

  Therefore, the problem of the symbol code recognition apparatus and the symbol code recognition program of the present case is to accurately extract the symbol code from the image.

  The symbol code recognition device of the present case has an extraction unit and a recognition unit. Moreover, this extraction part has a 1st extraction part and a 2nd extraction part.

  Here, the first extraction unit converts the symbol code on each scanning line into the symbol code on the basis of the change pattern of the pixels on each scanning line having the first pitch parallel to each other among the pixels constituting the image in which the symbol code is recorded. Extract overlapping one-dimensional regions.

  Further, the second extraction unit integrates the one-dimensional regions on the adjacent scanning lines based on the similarity of the change patterns of the pixels in the one-dimensional region on the adjacent scanning lines extracted by the first extraction unit. As a result, a two-dimensional area in which the symbol code on the image is recorded is extracted.

  The recognizing unit recognizes the content of the symbol code recorded in the two-dimensional area extracted by the extracting unit.

  According to this case, the two-dimensional area in which the symbol code is recorded is accurately extracted from the image, and the content of the symbol code recorded in the extracted two-dimensional area is recognized.

It is a system outline figure containing a notebook PC which operates as one embodiment of this case. It is an internal block diagram of the system shown in FIG. It is a block diagram of the symbol code recognition device of the first embodiment. It is the figure which showed an example of the symbol code. It is the schematic diagram which showed an example of the form. It is the figure which showed the scanning line. It is a figure which shows the 2nd scanning line employ | adopted by the recognition part. It is a block diagram of the symbol code recognition apparatus of 2nd Embodiment. It is the figure which showed the first half part of the flowchart which shows the process performed with the symbol code recognition apparatus shown in FIG. It is the figure which showed the second half part of the flowchart which shows the process performed with the symbol code recognition apparatus shown in FIG. It is a figure which shows a scanning line. It is a schematic diagram which shows a candidate confirmation process. It is explanatory drawing of the confirmation process whether a line | wire width ratio satisfy | fills a regulation value. It is explanatory drawing of a key detection process. It is explanatory drawing of an integrated confirmation process. It is a schematic diagram of integrated information. It is explanatory drawing of an inclination calculation process. It is explanatory drawing of the process in an inclination candidate integration part. It is a schematic diagram of integrated information when a symbol code is recorded obliquely. It is a figure which shows an example of symbol data. It is the figure which showed an example of the symbol code before symbol data correction | amendment. It is the figure which showed an example of the symbol code after symbol data correction | amendment. It is explanatory drawing of the processing content in a matrix process part. It is explanatory drawing of the processing content in a matrix process part. It is explanatory drawing of the processing content in a line sort process part. It is explanatory drawing of the processing content of a recognition result determination part. It is the figure which showed the priority of the comparison in a recognition result determination part.

  Hereinafter, embodiments will be described.

  FIG. 1 is a system outline diagram including a notebook personal computer (hereinafter abbreviated as “notebook PC”) operating as an embodiment of the present case.

  FIG. 1 shows a notebook PC 1 and a scanner 2. The notebook PC 1 includes a main body device 10 and a display device 20 that can be opened and closed with respect to the main body device. The main body device 10 performs arithmetic processing by executing a program such as a CPU (Central Processing Unit), a memory, and a storage unit. The necessary elements are provided. In addition, the storage unit stores a symbol code recognition program as an embodiment for operating the notebook PC 1 as an embodiment of the present symbol code recognition apparatus. Furthermore, a keyboard 110 that receives a user operation is provided on the upper surface of the main body device 10.

  The display device 20 has a display screen 210, and an image corresponding to an instruction from the main body unit 10 is displayed on the display screen 210.

  The notebook PC 1 is connected to a mouse 3 which is a kind of pointing device.

  Further, the scanner 2 reads the image from a form or the like in which an image including a symbol code is partially recorded, and generates image data. In the following, “image data” is also expressed as “image” when there is no need to explain the concept separately. The scanner 2 is connected to the notebook PC 1, and an image read by the scanner 2 is input to the notebook PC 1.

  In the notebook PC 1, the symbol code recognition program stored in the storage unit is executed, and when an image is input from the scanner 2, the symbol code area on the image is extracted and the contents of the symbol code are extracted in the notebook PC 1. Is recognized.

  FIG. 2 is an internal configuration diagram of the system shown in FIG.

  Here too, a notebook PC 1 and a scanner 2 are shown. The notebook PC 1 includes a CPU 11, a memory 12, a storage unit 13, an operation unit 14, a display unit 15, and an interface 16, which are connected to each other via a bus 17.

  The CPU 11 is a central processing unit that executes a program.

  The memory 12 is a memory in which a program executed by the CPU 11 is expanded and stored in an execution format. The memory 12 is also used as a work area for a program being executed by the CPU 11.

  The storage unit 13 is a large-capacity storage unit such as an HDD (hard disk drive) that stores various programs and data in a nonvolatile manner.

  The operation unit 14 is an element that receives an operation from the user, and collectively refers to the keyboard 110 and the mouse 3 shown in FIG.

  The display unit 15 is provided in the display device 20 shown in FIG. 1 and is an element that has the display screen 210 shown in FIG. 1 and displays an image on the display screen 210 in response to a command from the CPU 11. .

  Further, the interface 16 is an element that is connected to the scanner 2 and receives an image (image as image data) obtained by the scanner 2.

  FIG. 3 is a block diagram of the symbol code recognition apparatus of the first embodiment.

  The symbol code recognition device 30 represents a function realized in the notebook PC 1 by executing a symbol code recognition program in the notebook PC 1 shown in FIGS.

  The symbol code recognition device 30 includes an extraction unit 31 and a recognition unit 32. The extraction unit 31 further includes a first extraction unit 311 and a second extraction unit 312, and the recognition unit 32 includes a correction unit 321. , First recognition unit 322, second recognition unit 323, and integration unit 324.

  The extraction unit 31 extracts a two-dimensional area in which the symbol code 100 on the image is recorded.

  The recognizing unit 32 recognizes the content of the symbol code recorded in the two-dimensional area extracted by the extracting unit 31.

  Here, a symbol code that is a recognition target in the symbol code recognition device 30 shown in FIG. 3 will be described.

  FIG. 4 is a diagram illustrating an example of a symbol code.

  The symbol code 100 is a one-dimensional symbol code composed of an alternating arrangement of black bars and white bars each having a plurality of line widths. A start code 101 is recorded at the top of the symbol code 100, one or more character codes 102 are recorded following the start code 101, and a stop code 103 is recorded at the end.

  There are multiple types of symbol codes. In the case of the symbol codes of the type shown here, each of the start code 101, the code of one character in the character code 102, and the stop code 103 are each composed of nine bars. These nine bars are composed of five black bars and four white bars sandwiched between the black bars. Furthermore, there are quiet zones 104 and 105 in which no bar array exists before the start code 101 and after the stop code 103. The patterns of the start code 101 and the stop code 103 are determined for each type of the symbol code 100. Therefore, by examining the start code 101 and the stop code 103, the type of the symbol code 100 can be known. Here, black bars / white bars are used, but the bars are not necessarily limited to black / white as long as they are bars of brightness or color that can be distinguished from each other. That is, the black bar / white bar is an example of each of the first bar / second bar. However, in the following description, it is expressed as a black bar / white bar as a typical example.

  Also, the black bar and the white bar shown in FIG. 4 are both composed of two types of bar widths, a thin line and a thick line. This is also an example, and according to the current standard, the black bar and the white bar. There are up to four bar widths each. In the following description, a bar having two types of line widths, ie, a thin line and a thick line, will be mainly described as an example. However, the present case is also applied to a symbol code including a bar having a line width other than two types.

  FIG. 5 is a schematic diagram showing an example of a form.

  Here, for example, a one-dimensional symbol code 100 as shown in FIG. 4 is printed or affixed somewhere on the A4 size paper 201 (lower right corner in the example of FIG. 5).

  In addition to the symbol code 100, various characters, figures, ruled lines, and the like are recorded in this form.

  The symbol code recognition apparatus 30 shown in FIG. 3 extracts an area where the symbol code 100 exists from an image (an image made up of image data) obtained by reading a form as shown in FIG. This is a device for recognizing the contents of the symbol code 100.

  Here, the symbol code 100 on the form 200 shown as an example in FIG. 5 is recorded in the lower right corner on the sheet 201, but the recording position and number of the symbol codes 100 on the sheet 201 are not known in advance. The symbol code recognition device 30 in FIG. 3 finds and extracts an area where the symbol code 100 exists from an image on the data.

  Further, the symbol code 100 on the form 200 is allowed to be affixed. That is, the symbol code 100 is not necessarily affixed parallel or perpendicular to the side of the paper 201, and may be affixed obliquely. The symbol code recognition apparatus 30 also extracts and recognizes such a symbol code 100 attached obliquely.

  FIG. 6 is a diagram showing scanning lines.

  The extraction unit 31 of the symbol code recognition device 30 shown in FIG. 3 sets scanning lines 301 having a first pitch (for example, 2.5 mm pitch) on the image read by the scanner 2 shown in FIG. However, the pixels of the image obtained by reading with the scanner 2 are arranged at a pitch (for example, 0.1 mm pitch) sufficiently finer than the first pitch (2.5 mm pitch).

  The scanning line 301 set here is set before the detection of the symbol code 100. When the symbol code 100 is recorded obliquely, the scanning line 301 is a scanning line that obliquely crosses the symbol code 100. Become.

  Returning to FIG. 3, the description will be continued.

  The first extraction unit 311 of the symbol code recognition device 30 shown in FIG. 3 is based on the change pattern of the pixels on each scanning line of the first pitch parallel to each other among the pixels constituting the image in which the symbol code is recorded. A one-dimensional area overlapping with the symbol code on each scanning line is extracted. Specifically, the first extraction unit 311 determines whether or not the change pattern satisfies the following plurality of criteria based on the change pattern of the pixels on each scanning line 301 (see FIG. 6). An area (one-dimensional area) on the scanning line 301 that satisfies the above criteria is extracted.

  (1) Quiet zones 104 and 105 (see FIG. 4) exist in front and behind.

  (2) The line width ratio between the black bar and the white bar satisfies the specified value.

  (3) The line width ratio between bars of a plurality of types of line widths satisfies a specified value.

  (4) The number of bars meets the specified value.

  (5) The start code 101 and the stop code 103 (see FIG. 4) are recognized.

  Specific examples of these (1) to (5) will be described later in the description of the second embodiment.

  The second extraction unit 312 of the symbol code recognition device 30 illustrated in FIG. 3 performs adjacent scanning based on the similarity of the change patterns of the pixels in the one-dimensional region on the adjacent scanning line extracted by the first extraction unit 311. Integrate one-dimensional regions on a line with each other. By this integration, the second extraction unit 312 extracts a two-dimensional area in which the symbol code on the image is recorded. Specifically, the second extraction unit 312 satisfies the following plurality of criteria from the plurality of one-dimensional regions extracted by the first extraction unit 301 on the plurality of scanning lines 301 (see FIG. 6). A first integration process for extracting a one-dimensional area and integrating the extracted plurality of one-dimensional areas into a two-dimensional area is executed.

  (6) The positions of the plurality of one-dimensional regions in the direction along the scanning line 301 match within the specified value.

  (7) The positions of the white bars in the plurality of one-dimensional areas match within a specified value.

  (8) The number of black bars matches within the specified value.

  (9) The black bars are continuous in the direction across the plurality of scanning lines 301, that is, in the vertical direction in which the black bars extend (see FIG. 4).

  (10) The types of symbols represented by the start code 101 and the stop code 103 (see FIG. 4) of the plurality of one-dimensional areas match.

  Specific examples of these (6) to (10) are also left to the description of the second embodiment to be described later.

  The symbol code 100 is printed or affixed in parallel to the side of the form sheet 201 (see FIG. 5), and the symbol code 100 on the image obtained by reading with the scanner 2 (see FIGS. 1 and 2) (see FIG. 4). The above-described first integration process is sufficient when there is little crushing or twisting (see FIG. 5). However, here, even when a lower quality image is obtained, the following second extraction process is performed in order to accurately extract the two-dimensional region on which the symbol code 100 is printed or pasted.

  That is, the 1st extraction part 311 extracts the one-dimensional area | region which satisfy | fills the reference | standard of (1)-(4) among the reference | standard of said (1)-(5) first as a primary candidate area | region, Furthermore, primary From the candidate areas, a primary candidate area that satisfies the criterion (5) is extracted as a secondary candidate area.

  In addition, the second extraction unit 312 is a secondary that satisfies the criteria (6) to (10) from among the plurality of secondary candidate regions extracted by the first extraction unit 311 on the plurality of scanning lines 301. Extract candidate areas. The “one-dimensional region” in (6) to (10) is a one-dimensional region that satisfies the above criteria (1) to (5), and is the “two-dimensional candidate region” itself. Further, the second extraction unit 312 also sets the primary candidate region that satisfies the following criteria as a target for integration.

  (11) A primary candidate region sandwiched between a plurality of extracted secondary candidate regions.

  (12) Meet the criteria of (6) to (9), excluding the criteria of (10) among the criteria of (6) to (10) above.

  That is, here, the second candidate region that integrates both the secondary candidate region that satisfies the above criteria (6) to (10) and the primary candidate region that satisfies the criteria (11) to (12) into a two-dimensional region. Integration processing is executed.

  When the symbol code 100 is attached obliquely on the form sheet 201 (see FIG. 5), the two-dimensional region in which the symbol code is recorded cannot be extracted by the first integration process and the second integration process. There is a case. Therefore, the second extraction unit 312 also performs the following third integration process. Here, of the primary candidate areas extracted by the first extraction unit 311, the third integration process target primary candidate that is excluded from the integration target in both the first integration process and the second integration process in the second extraction unit 312. The first candidate area for the third integration process that satisfies the following criteria is integrated into the two-dimensional area.

  (13) A third integration processing target one-dimensional candidate region that intersects a straight line extending in a direction perpendicular to the direction in which the black bar extends from the start code start point of the third integration processing target primary candidate region in which the start code is detected. There is.

  (14) The inclination of the symbol code matches within a specified value.

  (15) The black bars are continuous in a direction across a plurality of scanning lines.

  (16) The symbol code type matches, or if the symbol code type is unknown, the symbol code type is sandwiched between a plurality of third integration processing target one-dimensional candidate areas.

  Specific examples of the processes (13) to (16) will also be described in the second embodiment described later.

  3 recognizes the content of the symbol code 100 (see FIG. 4) extracted by the extraction unit 31 and recorded in the two-dimensional area as follows. The In the recognizing unit 32, a scanning line having a pitch different from the scanning line 301 (see FIG. 6) having the pitch employed in the extracting unit 31 is employed.

  FIG. 7 is a diagram illustrating a second scanning line employed by the recognition unit 32.

  The second scanning line 302 is a scanning line extending in a direction perpendicular to the direction in which the bar of the symbol code 100 extends. Therefore, when the symbol code 100 is arranged obliquely on the image as shown in FIG. 7, oblique scanning lines are employed. 6 is a scanning line with a pitch of 2.5 mm, for example, but the second scanning line 302 employed here has a narrower pitch than the scanning line 301, such as a pitch of 0.5 mm. It is a scanning line.

  Returning to FIG. 3, the description will be continued.

  In the correction unit 321 configuring the recognition unit 32 illustrated in FIG. 3, the number of bars on the second scanning line 302 (see FIG. 7) in the two-dimensional region extracted by the extraction unit 31 is determined as the second scanning line 302. It is measured every time, and it is determined whether or not the measured number matches the specified number. The correction unit 321 executes a correction process for matching the number of bars on the second scanning line 302 to the specified number by dividing or integrating the black bars with respect to the scanning line 302 whose measured number does not match the specified number. . The first recognizing unit 322 and the second recognizing unit 323 constituting the recognizing unit 32 respectively execute a first recognition process and a second recognition process described below for the symbol code corrected by the correcting unit 321. To do.

  The first recognizing unit 322 performs the second recognition for each second scanning line 302 on the two-dimensional region extracted by the extracting unit 31 based on the pixel change pattern on each second scanning line 302 (see FIG. 7). A first recognition process for temporarily recognizing the contents of the symbol code in the dimension area is executed.

  The second recognizing unit 323 divides the two-dimensional region extracted by the extracting unit 31 into each character region based on the change pattern of the pixels on each second scanning line, and each one of the character regions. The character area is further divided into a plurality of sub-areas. And this 2nd recognition part 323 performs the 2nd recognition process which recognizes a character temporarily for every sub-region divided | segmented in this way.

  Further, the integration unit 324 integrates the temporary recognition result by the first recognition unit 322 and the temporary recognition result by the second recognition unit 323, and recognizes the content of the symbol code recorded in the two-dimensional area.

  Specific examples of the processes in the correction unit 321, the first recognition unit 322, the second recognition unit 323, and the integration unit 324 will be described later in the second embodiment.

  In the symbol code recognition device 30 according to the first embodiment shown in FIG. 3, after executing the above processing, the form is rotated by 90 ° on the image and the same processing is executed again. This is because the symbol code 100 (see FIG. 5) may be printed or affixed in a direction different from the direction shown in FIG. 5 by 90 °. In the symbol code recognition device 30 of the first embodiment, as described above, the area on which the symbol code is recorded is extracted and the content of the symbol code in the area is recognized.

  Next, a more specific second embodiment will be described. Also in the second embodiment, the system configuration shown in FIG. 1 and the hardware configuration shown in FIG. 2 are the same as those in the first embodiment, and those figures will be referred to as they are, and re-explanation will be omitted. In the second embodiment, the contents described with reference to FIGS. 4 to 7 are the same as those in the first embodiment, and re-explanation with reference to FIGS. 4 to 7 is also omitted.

  FIG. 8 is a block diagram of the symbol code recognition apparatus of the second embodiment.

  Similarly to the symbol code recognition apparatus 30 of the first embodiment shown in FIG. 3, the symbol code recognition apparatus 50 is also executed by executing the symbol code recognition program in the notebook PC 1 shown in FIGS. Represents the function to be realized.

  The symbol code recognition device 50 includes an image input unit 51, an image storage unit 52, an extraction unit 53, an extraction result storage unit 54, a recognition unit 55, and a result output unit 56.

  The image input unit 51 receives an image read from a form sheet by the scanner 2 shown in FIGS. The image input unit 51 corresponds to the interface 16 shown in FIG. 2 in hardware. Alternatively, the image input unit 51 is not limited to an element that receives an image input from the scanner 2. For example, an element that receives an image via a communication line or a portable storage medium (for example, a DVD) is loaded. It may be an element that receives an image from a portable storage medium. Alternatively, the image input unit 51 may be an element that reads an image directly from a form sheet, for example, the scanner 2 itself shown in FIGS.

  The image obtained by the image input unit 51 is temporarily stored in the image storage unit 52. The image storage unit 52 corresponds to the storage unit 13 illustrated in FIG. 2 in hardware.

  The extraction unit 53 extracts an area in which a symbol code existing on the image is recorded from the form image stored in the image storage unit 52. Various processes in the extraction unit 53 are realized by a combination of the CPU 11 shown in FIG. 2 and a symbol code recognition program executed by the CPU 11. The extraction unit 53 includes an extraction control unit 531, a temporary extraction unit 532, a key detection unit 533, a candidate integration unit 534, an inclination calculation unit 535, and an inclination candidate integration unit 536. The description of these units 531 to 536 will be given later. Of these units 531 to 536, the combination of the temporary extraction unit 532 and the key detection unit 533 corresponds to the first extraction unit 311 of the first embodiment (see FIG. 3). In addition, the combined element of the candidate integration unit 534, the inclination calculation unit 535, and the inclination candidate integration unit 536 corresponds to the second extraction unit 312 of the first embodiment described above.

  Information extracted by the extraction unit 53 is stored in the extraction result storage unit 54. Similarly to the image storage unit 52, the extraction result storage unit 54 corresponds to the storage unit 13 shown in FIG.

  Based on the form image stored in the image storage unit 52 and the extraction result stored in the extraction result storage unit 54, the recognition unit 55 stores the contents of the symbol code recorded in the area extracted by the extraction unit 53. Recognize Various processes in the recognition unit 55 are also realized by a combination of the CPU 11 shown in FIG. 2 and a symbol code recognition program executed by the CPU 11. However, the recognition result storage unit 552 in the recognition unit 55 corresponds to the storage unit 13 in FIG.

  The recognition unit 55 includes a recognition control unit 551, a recognition result storage unit 552, a symbol data creation unit 553, a symbol data correction unit 554, a matrix processing unit 555, a line sort processing unit 556, and a recognition result determination unit 557. Here, among these units 551 to 557, the symbol correction unit 554 mainly corresponds to the correction unit 321 of the first embodiment (see FIG. 3). Also, the line sort processing unit 556 and the matrix processing unit 555 correspond to the first recognition unit 322 and the second recognition unit 323, respectively, of the first embodiment described above. Furthermore, the recognition result determination unit 557 corresponds to the integration unit 324 of the first embodiment described above.

  The description of each of these units 551 to 557 constituting the recognition unit 55 will be given later.

  The recognition result of the symbol code obtained by the recognition unit 55 is output from the result output unit 56. The use of the recognition result output from the result output unit 56 differs depending on the use of the form, and here, only the description that the recognition result is output will be given.

  9 and 10 are diagrams showing a first half part and a second half part of a flowchart showing processing executed by the symbol code recognition apparatus 50 shown in FIG. 8, respectively. FIG. 9 shows processing executed by the extraction unit 53 in FIG. 8, and FIG. 10 shows processing executed by the recognition unit 55 in FIG.

  Hereinafter, description will be made with reference to both the block diagram of FIG. 8 and the flowcharts of FIGS.

  The extraction control unit 531 of the extraction unit 53 is responsible for controlling the entire processing of the extraction unit 53.

The temporary extraction unit 532 extracts a primary candidate area from the image (step S11 in FIG. 9). That is, the temporary extraction unit 532 reads an image on one scanning line 301 (see FIG. 6) from the image storage unit 52, and extracts a primary candidate region according to the following algorithm.
(A) Black pixels are detected from the beginning of the read image for one scanning line toward the end point of the scanning line. If no black pixel is detected until the end of the scanning line, the process branches to (c) below.

  FIG. 11 is a diagram illustrating scanning lines.

Here, the pixels on the first scanning line (hereinafter may be referred to as “line 1”; the same applies to the second and subsequent scanning lines) are scanned from the left to the right, and the black pixels are scanned. Presence is detected. The same applies to each scanning line after line 2. Although only the symbol mark is shown here, as shown in FIG. 6, the entire image of the form sheet 201 is scanned, and black pixels are detected for each scanning line.
(B) When a black pixel is detected, it is confirmed whether a series of black and white patterns continuing from the black pixel can be candidates as a one-dimensional symbol code.

  When all of the following “candidate confirmation” conditions are satisfied, the extraction information is stored in the extraction result storage unit 54 as a primary candidate region.

  The conditions for “candidate confirmation” are the conditions (1) to (4) in the first embodiment as described below.

  (1) Quiet zones 104 and 105 (see FIG. 4) exist in front and behind.

  (2) The line width ratio between the black bar and the white bar satisfies the specified value.

  (3) The line width ratio between bars of a plurality of types of line widths satisfies a specified value.

  (4) The number of bars meets the specified value.

  FIG. 12 is a schematic diagram showing the candidate confirmation process according to the above (1) to (4).

  Regarding the existence of the quiet zone in (1) above, there is a predetermined length of white space before the first black pixel where the black and white pattern starts, and there is a bar after the black pixel at the end of the black and white pattern. It is confirmed whether or not there is a predetermined length blank.

  FIG. 13 is an explanatory diagram of the confirmation process of whether the line width ratio satisfies the specified value in the above (2) and (3).

  FIGS. 13A and 13B are diagrams showing histograms of the width in the scanning line direction of the black bar and the white bar, respectively. The horizontal axis represents the line width (thickness), and the vertical axis represents the number of bars of the line width (thickness).

  Here, a symbol code having two types of thin / thick line widths of the black bar and the white bar is described as an example, and there are two peaks corresponding to the thin / thick line width. These two peaks can be sufficiently discriminated even if the symbol code image is crushed or wrinkled.

  Here, assuming that each peak represents a line width, a ratio (pb1 / pw1) between thin lines of black bars and white bars and a ratio (pb2 / pw2) between thick lines are obtained. Is within the range of the specified value 1.0-α1 to 1.0 + α1.

1.0−α1 ≦ pb1 / pw1 ≦ 1.0 + α1
1.0−α1 ≦ pb2 / pw2 ≦ 1.0 + α1
Here, α1 is a predetermined value smaller than 1.0.

  Here, the line width of the bar is described as being two types of thin / thick, but a command code having three or more types of line width is also determined in the same manner.

  Further, here, for each of the black bar and the white bar, the ratio pb2 / pb1, pw2 / pw1 of the width of the thick line to the thin line is obtained, and the standard width ratio is set in advance smaller than r and r. When the value is α2, it is determined whether or not the ratio is within the range of the specified value r−α2 to r + α2 as follows.

r−α2 ≦ pb2 / pb1 ≦ r + α2
r−α2 ≦ pw2 / pw1 ≦ r + α2
In addition, in the case of a symbol code having more types than two types of thin / thick line width, it corresponds to each thickness bar when the peak (pb1, pw1 in FIG. 13) corresponding to the thinnest bar is used as a reference. Whether or not the peak to be satisfied (pb2, pw2 in FIG. 13) satisfies each specified value is a criterion.

  FIG. 13 also shows bottom thicknesses bb1 and bw1 between two peak thicknesses pb1, pb2, pw1 and pw2, which will be described later.

  Regarding (4) above, the number of black bars that repeat within a predetermined maximum interval from the first black pixel (see FIG. 11) on one scanning line is counted, and whether or not the number is equal to or greater than a predetermined value. Is determined. When the number of black bars is too small, it is excluded from the symbol code candidates because it is not a symbol code but a black line of characters and figures on the form that happens to repeat within the maximum interval. However, there is no upper limit for the large number of black bars. This is because long command codes are possible.

When the above conditions (1) to (4) are satisfied in this way, information for specifying a one-dimensional area on the scanning line that satisfies the conditions is stored in the extraction result storage unit 54.
(C) When the processes (a) and (b) are completed for one scanning line, the image on the next scanning line is read, and the processes (a) and (b) are performed for the scanning line. Done.

In the key detection unit 533 of the extraction unit 53 in FIG. 8, the primary candidate region information extracted as described above is read from the extraction result storage unit 54, and the primary candidate region is
(5) Key detection processing, that is, start code and stop code detection processing is performed (step S12 in FIG. 9).

  FIG. 14 is an explanatory diagram of the key detection process.

  When a key, that is, a start code and a stop code is detected for the line 1, information is stored in the extraction result storage unit 54 as a primary candidate area on the line 1 as a secondary candidate area. This process is performed for each primary candidate on each line (scan line).

  In the case of the symbol codes of the type shown here, the start code, the code for one character constituting the character code, and the stop code are all five black bars and four white bars sandwiched between the black bars. And a total of 9 bars.

  Here, the thickness of the bar is binarized by using the bottom thicknesses bb1 and bw1 shown in FIG. 13 as threshold values, a bar thinner than the thickness as a thin line, and a bar thicker than the thickness as a thick line. If the binarized thin line and thick line array match up to 8 out of the 9 bars that make up the start code and stop code, or 2 lines that match and do not match are adjacent to each other. If a start code or stop code is detected, it is handled. This case is taken into consideration when crushing or drowning does not completely match. The reason that two unmatched two are adjacent to each other when up to seven matches is that the two are originally thin bars and the other is a thick bar. This is because the thickness may be detected in reverse.

  Next, in the candidate integration unit 534, candidates are integrated based on a secondary candidate area, that is, a one-dimensional area in which a key (start code and stop code) is detected (step S13 in FIG. 9). Here, integrated information representing a two-dimensional area in which a symbol code is recorded is created when all the conditions of “integration confirmation” shown below are satisfied (see FIG. 16).

  The conditions for “integration confirmation” here are as follows. These conditions are the same as the conditions (6) to (10) in the first embodiment described above.

  (6) The positions of the plurality of two-dimensional candidate regions in the direction along the scanning line 301 (see FIG. 6) match within the specified value.

  (7) The positions of the white bars of the plurality of secondary candidate areas match within the specified value.

  (8) The number of black bars matches within the specified value.

  (9) The black bars are continuous in the direction across the plurality of scanning lines 301, that is, for example, the vertical direction in which the black bars in FIG. 4 extend.

  (10) The types of symbols represented by the start code 101 and the stop code 103 (see FIG. 4) of the plurality of secondary candidate areas match.

  FIG. 15 is an explanatory diagram of the integration confirmation process.

  Here, it is assumed that secondary candidate regions satisfying the above conditions (1) to (5) exist in any of the lines 1 to 3.

  At this time, as the condition of (6) above, whether or not the positions of the secondary candidate area of line 1 and the secondary candidate area of line 2 in the left-right direction along the line coincide with each other within a predetermined error range. Is determined.

  In addition, as a method for determining the position of the white bar in (7) above, the following method is employed here.

  Here, focusing on thick white bars, the number of thick white bars on line 1 is counted, and the number of thick white bars on line 2 at the same position as the white bar on line 1 is counted. Be counted. For example, if 50% or more of the thick white bar on line 1 is present at the same position as the thick white bar on line 1 with respect to the number of thick white bars on line 1, the white bar on line 1 And the white bars on line 2 are determined to match. A similar examination is performed between the secondary candidate areas on the lines 2 and 3 and other adjacent lines.

  Whether or not the number of black bars in (8) above matches, the number of black bars on each line is counted. For example, the number of black bars on line 2 with respect to the number of black bars on line 1 is counted. If they match within a range of about ± 10%, it is determined that the number of black bars on line 1 and the number of black bars on line 2 match. The same applies to the other lines.

  Regarding the continuity of the black bar in (9) above, the original image stored in the image storage unit 52 in FIG. 8 is referred to, and the black bar is adjacent to the extending direction, that is, the vertical direction in FIG. It is checked whether or not it is connected to the line. Thus, one condition for candidate integration is that black bars are continuous in a direction across a plurality of lines.

  Further, as described in (10) above, it is checked whether or not the start code / stop code matches, that is, whether or not the symbol code types on each line match. One condition for integration is that the symbol code types match.

  FIG. 16 is a schematic diagram of the integrated information.

  For example, for the lines 1 to 3 shown in FIG. 15, when the conditions (6) to (10) are satisfied, as shown in FIG. 16, integrated information representing a two-dimensional symbol area including all the lines 1 to 3 is obtained. Created. This integrated information indicates that one symbol code is recorded in this two-dimensional symbol area.

  When the above conditions (6) to (10) are not satisfied, they are handled as follows.

Here, it is assumed that the conditions (1) to (4) are satisfied for the lines 1 and 3 shown in FIG. 15, and the start code and the stop code that are the conditions of (5) are detected. On the other hand, for the line 2 sandwiched between the lines 1 and 3, the above-described (1) to (4) are satisfied, but the start code and stop code of (5) are not detected. Even in this case,
(11) It is sandwiched between two lines where the start code and the stop code are the same.
(12) Satisfy the criteria (6) to (9) except the criteria (10) among the criteria (6) to (10).
When the above two conditions are satisfied, the primary candidate area on the line sandwiched between the secondary candidate areas on the two lines where the start code and the stop code coincide with each other is also recorded on the same single symbol code. As a target of integration.

  Next, it is determined whether or not there is a primary candidate area that has not yet been integrated (step S14 in FIG. 9).

  When the symbol code is printed or pasted on the form with no inclination, the above processing ends the extraction process of the two-dimensional area in which the symbol code is recorded, and the integrated information representing the two-dimensional area is obtained. The result is stored in the extraction result storage unit 54 (step S17 in FIG. 9).

  On the other hand, if there is a primary candidate area that has not yet been integrated, the symbol code is expected to be recorded obliquely, and the following processing is executed.

  Here, first, the inclination calculation unit 535 of the extraction unit 53 in FIG. 8 performs an inclination calculation process on each primary candidate region that has not yet been integrated (step S15 in FIG. 9).

  FIG. 17 is an explanatory diagram of the inclination calculation process.

  FIG. 17 shows the primary candidate regions on each of the lines 1 to 3 where black and white patterns are continuous.

  Here, based on the form image stored in the image storage unit 52, the extending direction of the leading black bar of each one-dimensional candidate area on each of the lines 1 to 3 is examined, and from the extending direction of the black bar. When this monochrome pattern represents a symbol code, the inclination of the symbol code is calculated. The calculated inclination information is also stored in the extraction result storage unit 54.

  Next, candidate integration processing is performed on the primary candidate area in which the inclination is detected in the inclination candidate integration unit 536 in FIG. 8 (step S16 in FIG. 9).

  Here, the primary candidate areas where the inclination is detected are integrated when the following criteria (13) to (16) are satisfied, and integrated information representing a symbol area in which one symbol code is recorded is created. The following criteria (13) to (16) are the same criteria as the criteria (13) to (16) in the first embodiment.

  (13) A primary candidate region that intersects a straight line extending in a direction perpendicular to the direction in which the black bar extends from the start code start point of the primary candidate region in which the start code is detected.

  (14) The slopes of the primary candidate areas match within a specified value.

  (15) The black bars are continuous in a direction across a plurality of lines.

  (16) The symbol code types must match. Or, when the type of the symbol code is unknown, it is sandwiched between a plurality of primary candidate areas with the same symbol code type.

  FIG. 18 is an explanatory diagram of the processes (13) to (16) in the inclination candidate integration unit 536.

  Here, in order to investigate the above (13), the first black bar of the primary candidate region on each line is extended in a direction perpendicular to the extending direction of the black bar (a direction parallel to the inclined symbol code). It is examined whether or not the straight line intersects with the primary candidate region on the adjacent line. In the example shown in FIG. 18, a straight line drawn diagonally upward in parallel to the symbol code from the top black bar of the primary candidate area on line 2 intersects with the primary candidate area on line 1. In addition, a straight line drawn diagonally upward from the leading black bar of the primary candidate area on line 3 in parallel with the symbol code intersects the primary candidate area on line 2 and further the primary candidate area on line 1. Here, it is determined that the above condition (13) is satisfied when such a crossing is performed.

  Regarding the coincidence of the inclination in (14) above, the extending direction of the black bar in the primary candidate area on each line is checked based on the image stored in the image storage unit 52, and the extending direction of the black bar is the specified value. It is determined whether or not they match.

  Regarding the continuity of the black bars in (15) above, based on the images stored in the image storage unit 52, whether or not the black bars of the primary candidate areas on each line are continuous across adjacent lines. Is examined. One of the conditions for integration is that the black bars are continuous.

  With respect to the condition that the types of symbol codes in (16) above match, even in the primary candidate area in which only one of the start code and stop code is detected, the above (13) to (15) It is only necessary to satisfy the above conditions and to match the types of the symbol codes.

  Furthermore, even if it is a primary candidate area in which neither a start code nor a stop code is detected, it is sandwiched between primary candidate areas having the same symbol code type, and the above (13) to (13) If each condition of 15) is met, it becomes an integration target.

  FIG. 19 is a schematic diagram of the integrated information when the symbol code is recorded obliquely.

  When the lines 1 to 3 described with reference to FIGS. 17 and 18 satisfy the conditions (13) to (16), the primary candidate areas on the lines 1 to 3 are shown in FIG. It is integrated as one symbol area.

  Integration information (see FIG. 16) created by the candidate integration unit 534 in FIG. 8 (step S13 in FIG. 9) and integration information (see FIG. 19) created by the inclination candidate integration unit 536 in FIG. 8 (step S16 in FIG. 9). Is stored in the extraction result storage unit 54 (step S17 in FIG. 9).

  Thus, the description of the extraction unit 53 of the symbol code recognition device 50 according to the second embodiment shown in FIG. 8 is finished. Next, the recognition unit 55 will be described.

  In the recognizing unit 55, based on the image stored in the image storage unit 52, the integrated information and the inclination information stored in the extraction result storage unit 54 are referred to, and the image pattern in the symbol area represented by the integrated information Is examined. In the recognizing unit 55, a second scanning line having a finer pitch than the scanning line 301 (see FIG. 6) employed in the extracting unit 52 is employed. The recognition unit 55 employs the second scanning line 302 parallel to the inclined symbol code even if the symbol code is inclined based on the inclination information (see FIG. 7).

  The recognition control unit 551 of the recognition unit 55 is responsible for overall control of each process of the recognition unit 55.

  The recognition result storage unit 552 stores recognition results obtained by a matrix processing unit 555, a line sort processing unit 556, and a recognition result determination unit 557, which will be described later.

  The symbol data creation unit 553 creates symbol data (step S21 in FIG. 10).

  FIG. 20 is a diagram illustrating an example of symbol data.

  The symbol data creation unit 553 alternately records the width of the black bar and the width of the white bar when the second scanning line 302 is traced from the left to the right for each of the second scanning lines 302 shown in FIG. For example, symbol data as shown in FIG.

  The relationship between FIG. 20A and FIG. 20B will be described later.

  The symbol data correction unit 554 corrects the symbol data created by the symbol data creation unit 553 (step S22 in FIG. 10).

  FIG. 21 is a diagram illustrating an example of a symbol code before symbol data correction. FIG. 22 is a diagram showing an example of the symbol code after the symbol data correction.

  As shown in FIG. 21, the symbol code on the image may be crushed or twisted.

  Therefore, here, the number of black bars is counted for each second scanning line (hereinafter, the second scanning line is referred to as a “line”). As described with reference to FIG. 4, in the case of the symbol code of the type exemplified here, one character is composed of five black bars and four white bars sandwiched between the black bars. It consists of a total of 9 bars. Therefore, the number of black bars for each line should be a multiple of five. In practice, however, the number out of multiples of 5 may be counted due to crushing or twisting.

  In the example shown in FIG. 21, the number of black bars on lines 1 and 2 is counted as 34, which is one less.

  Lines 3, 4 and 5 are counted as 35, which is a multiple of 5.

  Line 6 is counted as 36, one more.

  Therefore, in the symbol data correction unit 554, as shown in FIG. 22, with respect to the line in which the number less than the prescribed number (here 35) is counted, one black bar is divided into two in order from the thickest black bar. Division processing is performed. On the other hand, the line where the number of lines more than the specified number is counted is filled with black in order from the thinnest white bar, and the black bars on both sides of the white bar are integrated into one black bar.

  The symbol correction process described above is performed on the symbol data created by the symbol data creation unit 553.

  20A shows an example of symbol data before symbol data correction, and FIG. 20B shows an example of symbol data after symbol data correction.

  In FIG. 20A, the thickness of “black” meaning the black bar in the third column from the left is recorded as “10”, and this black bar is the thickest black bar. Also, assume that one black bar is insufficient as a result of counting the number of black bars for this line.

  In FIG. 20B, this “black: 10” is divided into “black: 4, white: 2, black: 4”.

  Such symbol data correction does not always correct correctly, but there is a high possibility of correct correction. Even if the correction is not performed correctly, the symbol code is finally recognized accurately by the subsequent processing.

  In the matrix processing unit 555 of FIG. 8, the symbol data is divided into a plurality of matrixes for each character and for each character (step S23 in FIG. 10), and pattern matching is further performed on the symbol data divided in the matrix (step S24 in FIG. 10). ).

  FIG. 23 is an explanatory diagram of processing contents in the matrix processing unit.

  As shown in FIG. 23A, the symbol code is divided for each character in the horizontal direction, and here is divided into three groups in the vertical direction. Here, the division into a plurality of pieces in both the vertical and horizontal directions is referred to as matrix division.

  Next, symbol data pattern matching is performed for each group region when the matrix is divided as shown in FIG. Here, pattern matching refers to processing for recognizing which character the symbol data represents by comparing symbol data created from an image (see FIG. 20) with a correct pattern representing each character.

  FIG. 23B shows one group region when the matrix is divided. A plurality of lines also run in this group, and a character is recognized for each line. That is, in the example shown here, it is assumed that the lines 1, 2, and 3 are both recognized as the numeral “0”, and the lines 4 and 5 cannot be recognized. At this time, as shown in FIG. 23C, “0” is adopted as the recognition result of the group by majority vote.

  In matrix processing section 555, after the above recognition processing is performed for each group in FIG. 23A, recognition processing for the entire symbol code is further performed.

  FIG. 24 is an explanatory diagram of processing contents in the matrix processing unit.

  By the above recognition processing for each group, as shown in FIG. 26, the three groups I, II, and III for each character are: I: “? 1234”, II: “01234”, III: “? 1? 44 ". Here, “?” Indicates that the group could not be recognized. In this case, priorities are assigned by majority decision, and are recognized as 1: “01234”, 2: “01244”, 3: “01235”, 4: “01245” in descending order of priority.

  This recognition result is stored in the recognition result storage unit 552 of FIG. 8 together with the priority information.

  Here, although the priority order has been determined, the final recognition result has not yet been obtained.

  In the line sort processing unit 556 in FIG. 8, symbol data pattern matching is performed for each line (step S25 in FIG. 10).

  FIG. 25 is an explanatory diagram of processing contents in the line sort processing unit.

  Here, pattern matching is performed for each line.

  In the example shown in FIG. 5, for example, line 1 will be described. Line 1 is recognized as “? 1234”, and the NG number indicating the number of characters that cannot be recognized is “1”.

  Here, priority is given to each line in ascending order of the number of NGs. In the case of the same NG number, the priority is increased in the upper line here.

  The recognition result in the line sort processing unit 556 is also stored in the recognition result storage unit 552 of FIG. 8 together with the priority order information.

  Note that the processing in the matrix processing unit 555 (steps S23 to S24 in FIG. 10) and the processing in the line sort processing unit 556 (step S25 in FIG. 10) do not ask the time ahead, and any processing is performed first. May be.

  In the recognition result determination unit 557 in FIG. 8, the recognition result by the matrix processing unit 555 and the recognition result by the line sort processing unit 556 are combined to determine the final recognition result (step S26 in FIG. 10).

  FIG. 26 is an explanatory diagram of processing contents of the recognition result determination unit.

  The recognition result determination unit 557 reads the recognition result by the matrix processing unit 555 and the recognition result by the line sort processing unit 556 stored in the recognition result storage unit 552. Then, the recognition result determination unit 557 compares both recognition results in descending order of priority, and when both recognition results match, the matching recognition result is determined as the final recognition result. In the example shown in FIG. 26, since the recognition result “01234” of the first priority of the matrix processing unit and the recognition result “01234” of the first priority of the line sorting processing unit match, this “01234” is the last. It becomes the recognition result.

  FIG. 27 is a diagram showing the priority of comparison in the recognition result determination unit.

  FIG. 27 shows that the recognition result of the first priority in the matrix processing unit is first compared with the recognition result of the first priority in the line sort processing unit. If the recognition results of the first ranks do not match, the recognition result of the first priority order of the matrix processing unit is compared with the recognition result of the second priority level of the line sort processing unit. If they still do not match, the recognition result of the second priority in the matrix processing unit and the recognition result of the first priority in the line sort processing unit are compared according to FIG.

  Here, the recognition result in the line sort processing unit is first changed to the recognition result having a lower priority than the recognition result in the matrix processing unit. This is because the recognition result by the line sort processing unit has a relatively high degree of reliability, and therefore, the recognition result of the highest priority order is used here for the recognition result by the matrix processing unit.

  Also in the second embodiment, as in the first embodiment described above, all the processes described above are performed once again by rotating the form sheet by 90 ° on the image. This is because the symbol code 100 (see FIG. 5) may be printed or affixed in a direction different by 90 ° from the direction shown in FIG.

  This is the end of the description of the second embodiment.

  According to each of the embodiments described above, even if the symbol code has been crushed or twisted, the area where the symbol code is recorded is extracted from the form sheet, and the symbol code is accurately recognized.

  Further, according to each of the embodiments described above, similarly for a symbol code attached obliquely, an area in which the symbol code is recorded is extracted from the form sheet, and the symbol code is accurately recognized.

  In each of the above-described embodiments, description has been given by taking as an example the recognition of two types of symbol codes in which the line width of the black bar / white bar is thin / thick. It is also applicable to the recognition of

  Hereinafter, various forms of the present case will be added.

(Appendix 1)
Based on the change pattern of the pixels on each scanning line of the first pitch parallel to each other among the pixels constituting the image in which the symbol code is recorded, a one-dimensional region that overlaps the symbol code on each scanning line is extracted. One extractor;
By integrating the one-dimensional regions on the adjacent scanning lines based on the similarity of the change patterns of the pixels in the one-dimensional regions on the adjacent scanning lines extracted by the first extraction unit, the image An extraction unit having a second extraction unit for extracting a two-dimensional region in which the above symbol code is recorded;
A symbol code recognition device comprising: a recognition unit for recognizing the content of the symbol code recorded in the two-dimensional area extracted by the extraction unit.

(Appendix 2)
The symbol code recognition device has a start code at the beginning, one or more character codes following the start code, and a stop code at the end by an alternating arrangement of first and second bars each having a plurality of types of line widths. The symbol of claim 1, further comprising: a device for recognizing the content of a one-dimensional symbol code having a quiet zone in which no bar array exists before the start code and after the stop code. Code recognition device.

(Appendix 3)
The first extraction unit includes the quiet zone before and after the change pattern of pixels on each scanning line, a line width ratio between the first bar and the second bar, and the plurality of types of line widths. The line width ratio between the bars and the number of bars satisfy each specified value, and further, a one-dimensional area in which the start code and the stop code are recognized is extracted. The described symbol code recognition device.

(Appendix 4)
The second extraction unit includes, on a plurality of scanning lines, out of a plurality of one-dimensional regions extracted by the first extraction unit, a position of the one-dimensional region along the scanning line, a bar position, And a number of one-dimensional areas where the number of bars match within each specified value, the bars continue in the direction across multiple scanning lines, and the symbol types represented by the start code and stop code match The symbol code recognition apparatus according to appendix 2 or 3, wherein the two-dimensional area is extracted by executing a first integration process for integrating the plurality of extracted one-dimensional areas with each other.

(Appendix 5)
The first extraction unit includes the quiet zone before and after the change pattern of pixels on each scanning line, a line width ratio between the first bar and the second bar, and the plurality of types of line widths. A one-dimensional area in which the line width ratio between bars and the number of bars satisfy each specified value is extracted as a primary candidate area, and the start code and the stop code are recognized from the primary candidate area Extracted primary candidate areas as secondary candidate areas,
The second extraction unit includes a position, a bar in a direction along the scan line of the secondary candidate region from among the plurality of secondary candidate regions extracted by the first extractor on the plurality of scan lines. And the number of bars coincide with each other within each specified value, the bars are continuous in a direction across a plurality of scanning lines, and the types of symbol codes represented by the start code and the stop code are the same. A secondary candidate area is extracted, and further, there is a primary candidate area sandwiched between the plurality of extracted secondary candidate areas, and a primary candidate area sandwiched between the extracted secondary candidate areas is the Direction in which the position in the direction along the scanning line, the position of the bar, and the number of bars match each other within the specified values, and the bar extends over the plurality of scanning lines, compared to the plurality of extracted secondary candidate areas If extracted, the extracted That the two-dimensional area is extracted by executing a second integration process that integrates a number of secondary candidate areas and the primary candidate areas sandwiched between the extracted secondary candidate areas. The symbol code recognizing device according to appendix 4, which is characterized by the above.

(Appendix 6)
The second extraction unit is excluded from the integration target in both of the first integration process and the second integration process in the second extraction unit in the primary candidate area extracted by the first extraction unit. The third integration process target primary candidate area intersects with a straight line extending in a direction perpendicular to the direction in which the black bar extends from the start point of the start code of the third integration process target primary candidate area in which the start code is detected. Three integration processing target one-dimensional candidate regions, the symbol code slopes in the third integration processing target one-dimensional candidate region match within a specified value, and the bars are continuous in a direction across a plurality of scanning lines. The third integration processing target one-dimensional object sandwiched between the plurality of third integration processing target one-dimensional candidate regions whose symbol code types are unknown or whose symbol code type is unknown By executing the third integration processing for integrating the auxiliary area, the symbol code recognition processing according to Note 5, wherein the and extracts the two-dimensional region.

(Appendix 7)
The recognition unit
The content of the symbol code in the two-dimensional area for each second scanning line based on the change pattern of the pixels on each second scanning line of the second pitch parallel to each other on the two-dimensional area extracted by the extraction unit A first recognition unit that executes a first recognition process for temporarily recognizing
Based on the change pattern of the pixels on each second scanning line, the two-dimensional area extracted by the extraction unit is divided into areas for each character, and a plurality of areas within each character are further included. A second recognition unit for executing a second recognition process for temporarily recognizing a character for each sub-region when the sub-region is divided;
An integration unit for recognizing a character string of a symbol code recorded in the two-dimensional region by integrating the temporary recognition result by the first recognition unit and the temporary recognition result by the second recognition unit; The symbol code recognition device according to any one of supplementary notes 1 to 6.

(Appendix 8)
The number of bars on the second scanning line in the two-dimensional region is measured for each second scanning line, and it is determined whether or not the measured number matches the specified number. A correction unit that executes a correction process for matching the number of bars on the second scanning line with a predetermined number by dividing or integrating the bars;
The first recognition unit and the second recognition unit respectively execute the first recognition process and the second recognition process for the symbol code corrected by the correction unit. 8. The symbol code recognition device according to 7.

(Appendix 9)
On the computer,
Based on the change pattern of the pixels on each scanning line of the first pitch parallel to each other among the pixels constituting the image in which the symbol code is recorded, a one-dimensional region on each scanning line that overlaps the symbol code is extracted,
A two-dimensional recorded symbol code on the image by integrating the one-dimensional areas on the adjacent scanning lines with each other based on the similarity of the change patterns of the pixels in the one-dimensional area on the adjacent scanning lines. Extract the region,
A symbol code recognition program for executing a process for recognizing the contents of a symbol code recorded in the extracted two-dimensional area.

(Appendix 10)
The symbol code recognition program causes a computer to start code at the beginning, a plurality of character codes following the start code, and stop at the end by alternating arrangement of first and second bars each having a plurality of types of line widths. A program for recognizing the contents of a one-dimensional symbol code in which a code is recorded and there is a quiet zone in which no bar array exists before the start code and after the stop code Symbol code recognition program.

(Appendix 11)
In extracting the one-dimensional region, the computer has the quiet zone before and after based on the change pattern of the pixels on each scanning line, the line width ratio between the first bar and the second bar, The line width ratio between bars of different line widths and the number of bars satisfy each specified value, and further, a one-dimensional area in which the start code and the stop code are recognized is extracted. The symbol code recognition program according to appendix 10.

(Appendix 12)
In extracting the two-dimensional region, the computer is caused to select a position of the one-dimensional region in the direction along the scanning line, a bar position, and a bar from among the extracted one-dimensional regions on the plurality of scanning lines. Each of the number of the same in each specified value, the bar is continuous in a direction across a plurality of scanning lines, and further, a plurality of one-dimensional regions in which the types of symbols represented by the start code and the stop code match are extracted, 12. The symbol code recognition program according to appendix 10 or 11, wherein the two-dimensional area is extracted by executing a first integration process for integrating the plurality of extracted one-dimensional areas with each other.

(Appendix 13)
In extracting the one-dimensional region, the computer has the quiet zone before and after based on the change pattern of the pixels on each scanning line, the line width ratio between the first bar and the second bar, A one-dimensional area in which the line width ratio between bars of different line widths and the number of bars satisfy each specified value is extracted as a primary candidate area, and from the primary candidate areas, the start code and the A process of extracting a primary candidate area where a stop code is recognized as a secondary candidate area is executed,
In extracting the two-dimensional region, the computer is caused to select the position of the secondary candidate region in the direction along the scanning line and the position of the bar from among the plurality of extracted secondary candidate regions on the plurality of scanning lines. , And the number of bars that coincide with each other within each specified value, the bars are continuous in a direction across a plurality of scanning lines, and a plurality of secondary codes with the same symbol code type represented by the start code and the stop code A candidate area is extracted, and further, a primary candidate area sandwiched between the extracted secondary candidate areas exists, and a primary candidate area sandwiched between the extracted secondary candidate areas is extracted. Compared with a plurality of secondary candidate areas, the position in the direction along the scanning line, the position of the bar, and the number of bars coincide with each other within each specified value, and the bar continues in the direction across the plurality of scanning lines. If you want to Extracting the two-dimensional area by executing a second integration process for integrating the extracted plurality of secondary candidate areas and the primary candidate area sandwiched between the extracted plurality of secondary candidate areas. The symbol code recognition program according to appendix 12, which is characterized by the following.

(Appendix 14)
In the extraction of the two-dimensional area, a third integration process target primary candidate area that is excluded from the integration target in both of the first integration process and the second integration process among the extracted primary candidate areas. A third integration processing target one-dimensional candidate region intersecting a straight line extending in a direction perpendicular to the direction in which the black bar extends from the start code start point of the third integration processing target primary candidate region in which the start code is detected The slopes of the symbol codes in the third integration processing target one-dimensional candidate region match, the bars are continuous in the direction across the plurality of scanning lines, and the symbol code types match or the symbol code types are A third integration target one-dimensional candidate region sandwiched between a plurality of third integration processing target one-dimensional candidate regions that are unknown but have the same symbol code type. By executing the integration process, Supplementary Note 13, wherein the symbol code recognition program, characterized in that in which to extract the two-dimensional region.

(Appendix 15)
In recognizing the symbol code recorded in the two-dimensional area,
Based on the extracted pixel change pattern on each second scanning line at a second pitch parallel to each other on the two-dimensional area, the content of the symbol code in the two-dimensional area is temporarily recognized for each second scanning line. ,
Based on the pixel change pattern on each second scanning line, the extracted two-dimensional area is divided into character areas and each character area is further divided into a plurality of sub-areas. Tentatively recognize a character for each sub-region when
The result of provisionally recognizing the contents of the symbol code in the two-dimensional area for each second scanning line and the result of provisionally recognizing the character for each sub-area are integrated and recorded in the two-dimensional area. 15. A symbol code recognition program according to any one of appendices 9 to 14, wherein recognition processing for recognizing a character string of a symbol code is executed.

(Appendix 16)
On the computer,
The number of bars on the second scanning line in the two-dimensional region is measured for each second scanning line, and it is determined whether or not the measured number matches the specified number. Executing a correction process for matching the number of bars on the second scanning line with the prescribed number by dividing or integrating the bars;
16. The symbol code recognition program according to appendix 15, wherein the recognition process is executed for the symbol code corrected by the correction process.

1 Notebook PC
2 Scanner 10 Main unit 11 CPU
DESCRIPTION OF SYMBOLS 12 Memory 13 Memory | storage part 14 Operation part 15 Display part 16 Interface 17 Bus | bath 20 Display apparatus 30, 50 Symbol code recognition apparatus 31,53 Extraction part 32,55 Recognition part 51 Image input part 52 Image storage part 54 Extraction result storage part 56 Result Output unit 100 Symbol code 101 Start code 102 Character code 103 Stop code 104, 105 Quiet zone 110 Keyboard 200 Form 201 Paper 210 Display screen 301 Scan line 302 Second scan line 311 First extraction unit 312 Second extraction unit 321 Correction unit 322 First recognition unit 323 Second recognition unit 324 Integration unit 531 Extraction control unit 532 Temporary extraction unit 533 Key detection unit 534 Candidate integration unit 535 Inclination calculation unit 536 Inclination candidate integration unit 551 Recognition control unit 552 Store recognition result 553 symbol data creation unit 554 symbol data correcting section 555 matrix processing unit 556 line sorting portion 557 the recognition result determining unit

Claims (6)

Based on the change pattern of the pixels on each scanning line of the first pitch parallel to each other among the pixels constituting the image in which the symbol code is recorded, a one-dimensional region that overlaps the symbol code on each scanning line is extracted. One extractor;
By integrating the one-dimensional regions on the adjacent scanning lines based on the similarity of the change patterns of the pixels in the one-dimensional regions on the adjacent scanning lines extracted by the first extraction unit, the image An extraction unit having a second extraction unit for extracting a two-dimensional region in which the above symbol code is recorded;
A symbol code recognition device comprising: a recognition unit for recognizing the content of the symbol code recorded in the two-dimensional area extracted by the extraction unit.   The symbol code recognition device has a start code at the beginning, one or more character codes following the start code, and a stop code at the end by an alternating arrangement of first and second bars each having a plurality of types of line widths. 2. The apparatus according to claim 1, further comprising a quiet zone in which a quiet zone having no bar array exists before the start code and after the stop code. Symbol code recognition device.   The first extraction unit includes the quiet zone before and after the change pattern of pixels on each scanning line, a line width ratio between the first bar and the second bar, and the plurality of types of line widths. The line width ratio between the bars and the number of bars satisfy each specified value, and further, a one-dimensional area in which the start code and the stop code are recognized is extracted. 3. The symbol code recognition device according to 2.   The second extraction unit includes, on a plurality of scanning lines, out of a plurality of one-dimensional regions extracted by the first extraction unit, a position of the one-dimensional region along the scanning line, a bar position, And a number of one-dimensional areas where the number of bars match within each specified value, the bars continue in the direction across multiple scanning lines, and the symbol types represented by the start code and stop code match 4. The symbol code recognition apparatus according to claim 2, wherein the two-dimensional area is extracted by executing a first integration process for integrating the plurality of extracted one-dimensional areas with each other. . The first extraction unit includes the quiet zone before and after the change pattern of pixels on each scanning line, a line width ratio between the first bar and the second bar, and the plurality of types of line widths. A one-dimensional area in which the line width ratio between bars and the number of bars satisfy each specified value is extracted as a primary candidate area, and the start code and the stop code are recognized from the primary candidate area Extracted primary candidate areas as secondary candidate areas,
The second extraction unit includes a position, a bar in a direction along the scan line of the secondary candidate region from among the plurality of secondary candidate regions extracted by the first extractor on the plurality of scan lines. And the number of bars coincide with each other within each specified value, the bars are continuous in a direction across a plurality of scanning lines, and the types of symbol codes represented by the start code and the stop code are the same. A secondary candidate area is extracted, and further, there is a primary candidate area sandwiched between the plurality of extracted secondary candidate areas, and a primary candidate area sandwiched between the extracted secondary candidate areas is the Direction in which the position in the direction along the scanning line, the position of the bar, and the number of bars match each other within the specified values, and the bar extends over the plurality of scanning lines, compared to the plurality of extracted secondary candidate areas If extracted, the extracted That the two-dimensional area is extracted by executing a second integration process that integrates a number of secondary candidate areas and the primary candidate areas sandwiched between the extracted secondary candidate areas. 5. The symbol code recognition apparatus according to claim 4, wherein On the computer,
Based on the change pattern of the pixels on each scanning line of the first pitch parallel to each other among the pixels constituting the image in which the symbol code is recorded, a one-dimensional region on each scanning line that overlaps the symbol code is extracted,
A two-dimensional recorded symbol code on the image by integrating the one-dimensional areas on the adjacent scanning lines with each other based on the similarity of the change patterns of the pixels in the one-dimensional area on the adjacent scanning lines. Extract the region,
A symbol code recognition program for executing a process for recognizing the contents of a symbol code recorded in the extracted two-dimensional area.

Images