JP3196603B2 - Barcode recognition method and system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a recognition system for converting an image (image data) of a bar code provided on a form, document, package or other article into a character code, and more particularly to a low-quality image data. And a method and system for recognizing.

[0002]

2. Description of the Related Art Generally, there are pen type, optical type, image scanner type and other recognition devices for bar code recognition. The pen type and the optical type are recognized by inputting only the barcode image data by crossing the stripe pattern forming the barcode in one line. For this reason, in the case of low-quality barcodes, for example, the print may be scuffed and damaged (blurred) or copied or F
If the stripe pattern is not clear (crushed) due to AX transmission / reception, the value of the image data input by reading the bar code is not stable due to blurring, crushing, or a quantization error of the image. As a result, the width of the stripe pattern (black and white bar) (hereinafter, also referred to as “width value”) is erroneous, and the barcode cannot be recognized. This type of recognition system is disclosed in Japanese Patent Application Laid-Open No. 62-90776 (patent application).
This is described in the inventor Masataka Ota and the applicant Sato Corporation on June 6, 1985.

Further, the barcode image input by the image scanner device is determined to be black and white based on a preset value based on the number of dots of the input image (striped pattern). That is, the number of dots determined to be white or black is compared with a preset value, the width of the black-and-white bar is determined to be thin and wide (hereinafter, referred to as “thin determination”), and the barcode is recognized. Was.

[0004] Therefore, in the case of a low-quality barcode, the value of the stripe pattern image in dot units is not stable.
According to the above recognition, since a simple black-and-white determination or a thin determination is performed based on a preset value, the width value of the black-and-white bar is incorrect, and the barcode cannot be recognized. In addition, since the recognition is performed in dot units, the recognition speed is low. This type of recognition system is disclosed in Japanese Unexamined Patent Application Publication No.
91) February 12, 1990, inventor Peter John Somerville, applicant International Business Machines Corporation), and JP-A-6-30948.
1 (Patent Application April 21, 1993, inventor Hajime Honmachi et al. And applicant Nippon Telegraph and Telephone Corporation).

[0005] A similar technique is character recognition. In character recognition, the size of characters is normalized to match with a dictionary pattern prepared in advance. Therefore, characters similar to the dictionary pattern and characters similar to some extent can be recognized. In barcode recognition, thin and thick bar information is essential, and similar shapes and similarities are not allowed. In this regard, it can be said that barcode recognition has more severe conditions.

As described above, the conventional recognition system has a problem that a low-quality barcode image cannot be recognized and the recognition speed is low.

[0007] The storage density and capacity of magnetic disk devices are increasing year by year, and the applications thereof are not limited to conventional large-sized recording devices and recording devices for personal computers, but are expected to be used in a wide range of fields. For example, a type of mobile computing in which each person records necessary data and carries it around is under study.

[0008]

First, for each input image, a value for extracting a black-and-white bar constituting a barcode, a value for determining the thinness of a black-and-white bar, and the number of recognition lines are changed. By doing so, even low-quality barcode images can be recognized.

Second, high-speed recognition is performed by changing the number of recognition lines using image shaping data (SE data: start & end data). Where S
The E data is defined as a conversion direction in a barcode reading direction when rectangular coordinates including an X axis and a Y axis are determined in advance for a dot image of a barcode in which black pixels are continuous (hereinafter, “SE”). Conversion direction "), refers to a set of data that is generated when a barcode image is scanned along the conversion direction and includes the X coordinate of the start (start point) of the black pixel and the X coordinate of the end (end point) of the black pixel.

[0010]

In order to achieve the first object, first, bar candidate SE data which is considered to constitute a barcode is sequentially extracted for each bar from SE data of an image. Next, from the bar candidate SE data sequentially extracted for each bar, a set of the width of the bar that frequently appears and a coordinate set that gives the width (hereinafter, referred to as “frequency wide coordinate”) is obtained. The bar candidate SE data is tracked based on the multi-frequency coordinates to extract “constituent black and white bar data” constituting the bar. From this black and white bar data, normalization is performed according to the specific rules of the bar code standard to obtain character bar data of one character. Based on this character bar data, based on the “thin width value” that gives information on the width of the character bar,
A thin code is determined and a character code is extracted.

In order to achieve the second object, when converting the dot data of the input recognition image data into SE data, the number of lines of the SE data is changed.

Hereinafter, the principle and operation of the present invention will be described. FIG. 1 shows an outline of a processing block for performing barcode recognition. First, the form 10 input by the image scanner is entered.
The barcode recognition area is extracted based on the number of recognition lines 11 from the image.

Next, for SE data extraction 13, the image data (dot data) in the barcode recognition area is converted to SE data. Based on the SE data in the recognition area thus obtained, processing for each bar is performed (processing 14 for each bar). Here, the bar candidates are extracted from the SE data (bar candidate SE data extraction 15), and the bar candidates are tracked to extract the black and white data constituting the bar (constructed black and white bar data extraction 16).

In the "character-by-character processing 17" for extracting a character code from the black and white bar data, the bar is determined to be thin and thick (thin determination 18), and the bar code decode 19 is sequentially performed based on the bar code dictionary 20. Do. As a result, the character code 21 is extracted.

FIG. 2 shows an outline of the barcode recognition area extraction processing. From the input image data 30, an image is cut out based on global cutout position information (user instruction information 31) designated by a user (operator of the system). This area is converted into SE data, and cut-out SE data 32 is extracted.

Then, a process of thickening or narrowing the cut-out SE data 32 in the horizontal direction (thickening / thinning process 33) is performed to obtain “horizontal thick cut-out SE data 34” in which the horizontal direction is thickened. The barcode area 35 is extracted by tracking the outline of the black area with respect to the data 34 and selecting the maximum outline in the cutout area.

Using this area 35, the contour of only the bar code area 35 on the cut-out SE data 32 is traced, and the bar coordinates of the right end and the left end are obtained. From the bar coordinates at the right end and the left end, the bar code recognition area 37
Is extracted. The configuration of the barcode recognition area 37 includes a start point end point (xs, xe) in the horizontal direction (X coordinate), a start point end point (ys, ye) in the line direction (Y coordinate), and a barcode inclination value (dg).

FIG. 3 shows an outline of the SE data extraction processing.
Data is extracted from the recognition area image data 40 along the SE conversion direction (SE data extraction). For example, it is assumed that a part of the recognition area image data 40 is scanned from the left end along the SE conversion direction. In the first line, there are black dots whose X coordinate is 11,
12, 15, 16, 17,... In this case, when the X coordinate of the start point and the end point where black is continuous is obtained,
The result is a set such as 11-12, 15-17,... (First line of the recognition area SE data 43).

As a result, the recognition area image data 40
Are stored for each line (for each direction in which the barcode is drawn) as in the recognition area SE data 43. This recognition area SE data 43 is supposedly illustrated as 44. In FIG. 3, the image appears to be slightly rotated, but the above processing is not a rotation but a change of the image. In other words, when the width of the black bar is specified in the X-axis direction, the color becomes white from a certain X coordinate. In order to avoid this, after performing the above-described specification in the X direction by a fixed amount, a process of raising the black bar in the SE conversion direction in the Y axis direction is performed. This is the change of the image (or the alignment of the image). Since it is not a rotation process, it can be performed at high speed. In the present invention, the barcode images are aligned by performing such a change, and the recognition ability is improved.

FIG. 4 shows an outline of a black-and-white bar data extraction process forming a bar code. This processing is performed for each bar. First, from the recognition area SE data 44, SEs that have not been processed yet have been viewed for each line in the SE extraction direction.
Extract data. The following processing will be described with an example of tracking from the beginning in the SE extraction direction.

First, bar candidate SE data 51 is extracted, and a more frequent width and coordinates are extracted from the data. In the case of extracting a width with a high frequency, a width with a high frequency of white is also determined at the same time. This is performed by subtracting the E (end point) coordinate of the SE data being extracted for the frequency of black from the S (start point) coordinate of the next SE data on the same line as that being extracted, and calculating a value obtained by adding 1. Used.

Next, based on the frequent width and coordinates, S
The E data is tracked to extract basic SE data (hereinafter referred to as “basic SE”) (SE tracking basic SE extraction).
Based on the basic SE, it is determined whether the SE data of the next line (the direction in which the barcode is drawn) is OK or NG under the SE tracking condition 53. The SE tracking condition 53 is that either white or black of the value of the SE data of the next line touches the range of the position of the basic SE + the margin and is not longer than this range.

In the case of OK, after the number of tracking lines is counted up, the basic SE is updated to the SE data that has been tracked so far, and the operation shifts to the determination of the SE data of the next line. In the case of NG, the basic SE is kept as it is, a new margin is added to the SE data tracking range, and the S
Move on to E data determination.

As a result, the number of lines for which the SE data can be normally tracked is obtained. Next, the normal tracking line number determination 54 is performed, and only when the number of lines (≦) larger than the normal tracking line number is obtained, these sets of SE data are
It is recognized as the configuration black and white bar data 55 and is developed. This development method assigns ID numbers in the order in which bars are extracted, in which the width of frequent black is changed to the width of black and the width of frequent white is changed to white.
This is set together with the black and white flag indicating the black and white of the bar (structured black and white bar data).

FIG. 5 shows a process of recognizing the black and white bar data. In order to recognize one character from the constituent black and white bar data, the constituent black and white bar data 61 is selected. In this selection method, normalization is performed according to a specific rule of the barcode standard. In this example, nine bars are selected starting with a black bar.

Next, the black-and-white bar data 61 is divided into white bars and black bars, and ID numbers 63 and 64 are extracted by designating ID numbers in the order of thickness. Then, based on the reference bar thin value 65 to be referred to for determining the thin and thick bar, each of the ID number data 63 and 64 is determined to be thick or thin in the order of its contents.
That is, the width of the constituent black bar and the reference bar thin value 65 (the width of the thick black bar or the width of the thin black bar), the width of the constituent white bar and the reference bar thin value 65 (the width of the thick white bar or the width of the thin white bar). Width), and determine the smaller one.

As a result, there are thin and thick judgment result ID number data 66 and 67 which are divided into white bars and black bars. Based on the data 66 and 67, barcode value data 69 is extracted in the order of ID numbers based on the relationship 68 between the thin determination result and the code value. From the barcode value data 69, the smaller the order of the ID numbers is the upper bit, and the larger the order is the lower bit, the code value becomes 70. Next, a character code 72 is extracted from the barcode dictionary 71 based on the code value 70.

When the recognition is normally performed by the above processing, the reference bar thin value 65 is set to the thin determination result ID number data 6.
The processing of 73 is performed based on the 6, 67, and the configuration black and white bar data 61, and is updated to the reference bar thin value 74.

[0029]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail. FIG. 6 shows the configuration of the recognition system according to the present invention. The system includes an image scanner device 201 for inputting a document or form as image data, an image storage device 202 for storing an input image, an image cutout device 203 for cutting an image at a global cutout position specified by a user, User instruction information storage device 204 for storing the obtained global cutout position information, SE data extraction device 205 for converting image data (dot data) into SE data, SE for storing SE data
A data storage device 206, a barcode recognition area extraction device 207 for extracting a barcode recognition area, a recognition information storage device 208 for storing the number of lines of the barcode recognition area, a barcode for storing barcode recognition area SE data Recognition area SE data storage device 209, black and white bar data extraction device 210 for extracting data constituting a bar from bar code recognition area SE data, black and white bar data storage device 211 for storing black and white bar data, black and white bar data Thinness judging device 212 for judging thinness of a bar code, and bar code decoding device 2 for extracting a code value of a bar code and extracting a character code from the code value
13. A barcode dictionary storage device 214 for storing a relationship between a code value and a character code when a character code is extracted from a code value, a recognition result storage device 215 for storing a recognized character code, and an external device for inputting an instruction from a user Instruction input device 101, interactive display when inputting an instruction from a user,
The display device 103 includes a display device 103 for displaying a recognition result, a correction display, or a display for setting the number of recognition lines, and a control device 102 for controlling these devices.

Next, FIG. 7 and FIGS. 1, 2, 3, 4, 5,
The barcode recognition flow will be described with reference to 6, 8, and 9.
First, in “image input” of FIG. 7, the form 10 is input from the image scanner 201 and the image storage 2
The image data 30 is stored in 02. In “recognition area extraction 302”, the image cutout device 203 uses the cutout SE data 32 based on the global cutout position information designated by the user in the user instruction information storage device 204 from the image data 30 in the image storage device 202.
Is stored in the image storage device 202.

Next, the SE data extracting device 205 cuts out the cut-out image data, converts the cut-out image data into SE data 32, and stores it in the SE data storage device 206. Then, the barcode recognition area extraction device 207 performs the processing of the cut-out SE data 32 of FIG. 2 after the horizontal thickening / thinning processing 33 and, based on the number of recognition lines in the recognition information storage device 208, the barcode recognition area. Extract 37. Further, the information of the barcode recognition area 37 is stored in the recognition information storage device 208.

In “SE data extraction 303”, the SE data extraction device 205 extracts the recognition area image data 401 in the barcode recognition area 37 of the recognition information storage device 208 obtained from the cut-out image data of the image storage device 202. , Dot SE conversion. As a result, the recognition area SE data 402 is stored in the barcode recognition area SE data storage device 209.

In the bar candidate SE data extraction 311 of the “process for each bar 310”, the bar candidate SE data 51 is extracted from the recognition area SE data 402 in the bar code recognition area SE data storage device 209 as shown in FIG. Extract. In the configuration black-and-white bar data extraction 312 of the “process for each bar 310”, the configuration black-and-white bar data extraction device 210 performs processing for obtaining the configuration black-and-white bar data 55 from the bar candidate SE data 51 for all the bars. The bar data 403 is obtained, and the black and white bar data storage device 211 is configured.
To memorize.

In the bar thinness determination 321 of “process 320 for each character”, the thinness determination device 212 transmits the configuration monochrome bar data 403 stored in the configuration monochrome bar data storage device 211.
From the ID number data 63 to 68 in FIG. 5, a thin determination is made to obtain the bar code value data 69, and the bar code number IDs 1 to 9 of the bar code value data 404 are obtained.

In the code extraction 322 and the character code extraction 323 of “process 320 for each character”, the barcode decoding device 213 performs processing so as to obtain the character code 72 from the barcode value data 69 in FIG. Based on the barcode ID of the value data 404, 1 to 9,
The bar code data 94 is obtained. Then, based on the relationship between the barcode value and the character code in the barcode dictionary storage device 214, the character code 5C is developed for the barcode data 94. Next, it is confirmed whether or not the bar code number ID10 is a space in the configuration black and white bar data 403.

In the "process 320 for each character", the barcode ID numbers 11 to 19 are processed next to the above process, and then the barcode ID numbers 21 to 29 are processed in that order.

In the "recognition result output 330", the barcode value data 404 and the character code are stored in the recognition result storage device 21.
Transfer to 5. However, according to the barcode standard of the present embodiment, since the character code "5C" is a start end mark, only the character code "F1" is stored.

FIG. 10 illustrates a method of determining the number of lines for recognizing a barcode. There are three types of this method: a user setting 501, a barcode density 502, and an image resolution 503.
In the user setting 501, the number of recognition lines is determined by the user setting the recognition ability, and the recognition information storage device 208
Is stored in In the barcode density 502, the number of recognition lines is determined by the user setting the code density of the barcode, and is stored in the recognition information storage device 208. In the method based on the bar code density 502, before performing bar code recognition automatically, partial black and white bar data is obtained, the code density is extracted from the number of dots of one character, and the number of basic recognition lines (constant value) and the code A recognition line is calculated based on the density and stored in the recognition information storage device 208. At the image resolution 503, the number of recognition lines is determined by the user setting the image resolution of the image scanner, and is stored in the recognition information storage device 208.

FIG. 11 illustrates a range that can be recognized by the barcode recognition method of the present invention. Usually, the ratio between the thin bar and the thick bar of the barcode (hereinafter, "barcode thin ratio"
Is 1: 2 or more and the thinnest bar (hereinafter, referred to as
The print size of the “minimum thin bar” is 1 dot or more. The most severe condition in this range is bar code condition 6
01. The print image of the barcode condition 601 becomes 602. This print image 602 is indicated by an arrow 60.
The printing is performed at a printer density 604 of 3 units. That is, a thin black bar is printed with one dot and a thick black bar is printed with two dots.

The bar code print image 602
Is input by the scanner. At this time, the scanner inputs an image at the scanner density 606 in the unit of an arrow 605 larger than the arrow 603 of the printer density 604. as a result,
For example, the print image of line 607 is input image 6
08, the thin bar is 1 dot, the thick bar is 2 dots,
The thin bar can be entered normally. An essential condition is that a thick bar having a width of 2 dots can be read as 2 dots.

As described above, when inputting at a scanner density 606 smaller than the printer density 604, that is, when the printer density 604 is 200 dpi and the scanner density 606 is
If 6 is 160 dpi, when the ratio of scanner density / printer density is 0.8, a barcode printed with the minimum printer density can be recognized.

Next, the maximum value of the ratio of scanner density / printer density will be described. A print image printed under the same barcode condition 601 is 609 (same as 602).
Like the print image 602, the print image 609 is
Printing is performed at the printer density 611 in units of arrows 610. That is, a thin black bar is printed with one dot, and a thick black bar is printed with two dots. Then, the print image 609 is input by the scanner. At this time, the arrow 61 of the printer density 611
Scanner density 613 in units of arrow 612, less than 0
Is input to the scanner.

In this case, it suffices if the thin bar is read as one dot. For example, as in the input image 615, the print image of the line 614 can be input with a thin bar of 1 dot, a thick bar with 2 dots or 3 dots, and a thin bar with normal input. The above gives the maximum value of the range that can be recognized by the scanner density 613 that is higher than the printer density 611. At this time, for example, the printer density 611 is 200 dpi,
The scanner density 613 is 280 dpi, and the ratio of scanner density / printer density is 1.4. As a result,
The barcode recognition range 616 is 0.8 or more and 1.4 or less when expressed by the ratio of scanner density / printer density. In conventional barcode recognition, the ratio of scanner density / printer density was 2 or more.

[0044]

According to the present invention, a value for extracting a black-and-white bar constituting a barcode and a thinness determination of a bar are performed for each input image using SE data of the image. Since the value or the number of recognition lines is changed,
High-speed recognition is possible even for low-quality images.

[Brief description of the drawings]

FIG. 1 is a diagram showing an outline of a processing block for performing barcode recognition according to the present invention.

FIG. 2 is a diagram showing a barcode recognition area extraction process of the present invention.

FIG. 3 is a diagram showing an SE data extraction process of the present invention.

FIG. 4 is a diagram showing a black-and-white bar data extraction process forming a barcode according to the present invention.

FIG. 5 is a diagram showing recognition of “structured black and white bar data” of the present invention.

FIG. 6 is a diagram illustrating a configuration of a recognition system according to an embodiment of the present invention.

FIG. 7 is a diagram showing a barcode recognition flow according to one embodiment of the present invention.

FIG. 8 is a diagram showing the contents of SE data extraction data according to one embodiment of the present invention.

FIG. 9 is a diagram showing data contents up to extraction of a character code according to an embodiment of the present invention.

FIG. 10 is a diagram illustrating a recognition line determination method according to an embodiment of the present invention.

FIG. 11 is a diagram showing a barcode recognition range according to the present invention.

[Explanation of symbols]

10: Form, 11: Number of recognition lines, 12: Barcode recognition area extraction, 13: SE data extraction, 14: Processing for each bar, 15: Bar candidate SE data extraction, 16: Composition black and white bar data extraction, 17: For each character Processing, 18: Bar thinness determination, 19: Barcode decoding, 20: Barcode dictionary, 21: Character code.

──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Shozo Kadota 2880 Kozu, Odawara City, Kanagawa Prefecture, Hitachi, Ltd. Storage Systems Division (56) References JP-A-6-309481 (JP, A) JP-A-4 -225488 (JP, A) JP-A-5-324896 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06K 7/00 G06K 7/10

Claims (8)

(57) [Claims] An image scanner device for inputting an image of a document or a form having a barcode, and a barcode position is determined from the input image based on a preset number of recognition lines. A barcode recognition system having a function of extracting an image including a barcode from an image, a function of further arranging a barcode image from the cutout result, and a function of converting the aligned barcode image into a character code . 2. The barcode recognition system according to claim 1, further comprising: when reading a barcode of a density created at or derived from the printer density P with a scanner density S, 0.8 ≦.
A barcode recognition system that recognizes barcodes in the range of S / P ≦ 1.4. 3. The barcode recognition system according to claim 1, further comprising: for each input image, a value for extracting a black-and-white bar constituting a barcode or a bar constituting a barcode is thin. A barcode recognition system that has a function that allows a value for determining whether a person is thick or not to be changed. 4. The barcode recognition system according to claim 1, wherein said settable number of recognition lines can be automatically set at the time of barcode recognition. 5. The method according to claim 1, wherein an image of a document or a form having a bar code is input by an image scanner, and the input image is input based on a preset number of recognition lines.
Determining the position of the barcode from the image and cutting out the image data including the barcode from the input image; and X, the start and end X of the black pixel on the line when the barcode is scanned. When the data is converted into SE data, which is a data set consisting of coordinates or Y coordinates, and the cutout result is placed on XY coordinates, the black or white dots of the image of the cutout result are determined in a direction substantially parallel to the X axis. After several consecutive moves of the next dot in the Y-axis direction,
Or after arranging a barcode image so as to maintain the continuity by moving a next dot in the X-axis direction after a predetermined number of continuations in a direction substantially parallel to the Y-axis, A bar code recognition method including the step of converting an image of a character into a character code. 6. The bar code recognizing method according to claim 5, wherein the step of converting the character code into a character code includes the step of extracting the frequent continuous values of the coordinates and dots for each bar constituting the bar code. Image tracking for each bar, determining whether the bar is thin or thick with reference to a predetermined reference value for each bar, and converting the determination result into character code data Barcode recognition method. 7. The barcode recognition method according to claim 5, wherein the step of cutting out the image includes the steps of thickening and thinning the input image in the horizontal direction of the bar constituting the barcode; A barcode recognition method including a step of extracting those contours from the result of the thickening and thinning, and a step of cutting out the largest contour. 8. The barcode recognition method according to claim 5, wherein before the step of cutting out the image is executed, the image is cut out from the input image based on global cutout position information designated by an operator. Barcode recognition method comprising the step of: