JP2001167222A - Error correcting method, method and device for reading two-dimensional code and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the efficiency of correction when correcting an error in reading a two-dimensional (2D) code. SOLUTION: Error detection using the RS code word of redundant data for error correction is not performed but on the basis of the feature of a 2D code image, a defective code word is estimated as the error position of an encoded block (S1000). By specifying the error position of a code like this way, afterwards, elimination correction is performed for correcting the error only by calculating the size of the error (S1010). On the assumption that a QR code is made random, for example, the defective code word can be selected as a code word composed of only white or black cells.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a technique for performing error correction using an error correction code such as a Reed-Solomon (RS) code when reading a two-dimensional code.

[0002]

2. Description of the Related Art Conventionally, a two-dimensional code reader fetches a two-dimensional code as an image, executes a decoding process based on the fetched two-dimensional code image, and reads information recorded as a two-dimensional code.

[0003] First, a two-dimensional code will be described. 2
As shown in an example of FIG.
Although it has a dimensional spread and can record a much larger amount of information than the barcode shown in FIG. 13A, the structure is complicated. The two-dimensional code in FIG.
An R code, as shown in FIG. 14, is used to specify the position of the two-dimensional code 500 in the captured image.
Positioning symbols 510a, 510b, and 510c combining squares of a specific size ratio, and data cell positions provided between the three positioning symbols 510a, 510b, and 510c, where white and black are alternately combined. And timing cells 520a and 520b, which are reference patterns serving as indices of the above.

The inside of the two-dimensional code 500 is divided into n × n square cells (corresponding to “cells”), and positioning symbols 510 of the QR code are provided.
a, 510b, and 510c are, for example, a black square 512 having a side length of 7 cells, a white square 514 having a side length of 5 cells, and a black square 516 having a side length of 3 cells. Are concentrically superimposed. Therefore, this positioning symbol 51
When crossing the vicinity of the centers of 0a, 510b, and 510c linearly, black, white, black, white, and black patterns are 1: 1: 3: 1:
It is detected at a ratio of 1. Therefore, by utilizing this property, when black and white are alternately detected at the above-described ratio, the pattern is determined by positioning symbols 510a, 510b,
It is determined that the two-dimensional code 500 is a promising candidate, and the two-dimensional code 500 is inspected preferentially to determine the existing position.

The shape of the two-dimensional code 500 is 3
Positioning symbols 510a, 510b, 510c
It can be estimated that this is a range of a parallelogram uniquely determined by. The data is represented by cells (that is, data cells) of the information recording area 530 excluding the positioning symbols and the reference pattern, and each data cell is color-coded white (bright) or black (dark). Each data cell corresponds to 1-bit data. However, FIG.
In FIG. 3, the black and white pattern of the data cell is omitted for easy understanding. The position of each data cell is the center of three positioning symbols 510a, 510b, 510c and 2
The two timing cells 520a and 520b can be obtained by simple calculations as indices of vertical and horizontal coordinates, respectively. It is determined whether the position near the center of each data cell determined in this way is black or white,
By associating black with, for example, 1 and white with, for example, 0,
It can be recognized as binary data and decoded.

In the two-dimensional code reader, decoding processing based on the captured two-dimensional code image
Generally, error correction is performed. This is a function that enables data decoding even when a part of the two-dimensional code cannot be read due to damage or dirt.

As the two-dimensional code, an error correction code having a high correction capability has been used. Q
Taking the R code as an example, as shown in FIG.
Is divided into a data block and an error correction block, and encoded by Reed-Solomon (hereinafter, referred to as “RS”). Here, the error correction block is redundant data for RS encoding. Note that the data block and the error correction block are configured with one unit of 8-bit information called a codeword. D1 shown in FIG.
Each of D46 and E1 to E88 corresponds to a codeword. Hereinafter, the error correction block is described as an “RS block”, and the codewords constituting the RS block are described as an “RS codeword”. On the other hand, a code word that forms a data block is described as a “data code word”.

Here, in the QR code illustrated in FIG.
Data code words of D1 to D23 and R of E1 to E44
Block 1 consisting of S code words and D24 to D24
46, which is a block 2 composed of 46 data codewords and RS codewords E45 to E88.
S encoding is performed, and error correction is performed. Less than,
These blocks 1 and 2 will be referred to as “encoded blocks”.

Next, general error correction will be described. There are two types of error correction: a method of detecting an error position using a polynomial (hereinafter referred to as “detection correction”) and a method called erasure correction. The detection and correction roughly includes the following three steps.

[0010] A syndrome polynomial is obtained from the code. An error position polynomial and an error numerical polynomial are obtained from the syndrome polynomial. The error location is found from the error location polynomial by the Chien search method, and the magnitude of the error for that location (error pattern)
Error correction.

In the example of the QR code described above, a syndrome polynomial is obtained from the coding block, and an error position polynomial and an error value polynomial are obtained from the syndrome polynomial. Then, a codeword as an error position is specified from the error locator polynomial, and error correction is performed by determining the magnitude of the error of the codeword.

On the other hand, the erasure correction is performed by obtaining only the magnitude of the error, assuming that the position of the erroneous code (this is referred to as the erasure position), that is, the erroneous codeword is known in advance. Is what you do. Erasure correction is roughly composed of the following three steps.

A syndrome polynomial is obtained from the code. Using the erasure polynomial obtained from the erasure position, the syndrome polynomial is corrected to obtain an erasure numerical polynomial. Error correction is performed for the erasure position by determining the magnitude of the error.

That is, in the erasure correction, only the magnitude of the error is obtained on the premise that the error position is known. Therefore, unlike the detection correction, only one erasure numerical polynomial is obtained from the syndrome polynomial. Therefore, the correction capability is twice that of the detection correction.

For example, when this is specifically explained by the above-mentioned QR code, the above-described detection correction and erasure correction are as follows.
The number of RS code words (redundant data) required for correction changes. That is, if the detection and correction are performed, it is necessary to obtain two polynomials as described above. Therefore, a correction of one codeword requires a total of two correction codewords (redundant data). Therefore, when performing detection and correction,
The total number t of codewords that can be corrected is represented by the following equation, where RS is the number of RS codewords (redundant data). t = RS / 2 ... [Equation 1] On the other hand, if erasure correction is performed, only one polynomial expression may be obtained as described above. Therefore, one RS codeword is corrected for one codeword. If there is enough. Therefore, when performing erasure correction, the total number t of codewords that can be corrected is represented by the following equation, where the number of RS codewords (redundant data) is RS. t = RS ... [Equation 2] Thus, even if the number of RS codewords is the same,
If an erroneous codeword can be identified in advance, the total number of codewords that can be corrected by performing erasure correction doubles, and the correction efficiency increases. In other words,
Even if the number of correctable codewords (correction capability) is the same, if only erasure correction is performed, the required number of RS codewords is half.

In the erasure correction, if some of the error positions of the code are not known, the known error position is corrected as the erasure position, and the error position is detected for the unknown error position. Error correction. In this case, the correction efficiency is not twice as high as the detection and correction, but only a few erasure positions are specified.
The correction efficiency is higher than the detection and correction. In this specification, the term “erasure correction” is used to include this correction method, and when it is particularly distinguished, this correction method is referred to as “erasure detection correction”.

[0017]

As described above, although the erasure correction processing has a better correction capability than the detection correction processing, it is premised that the error position is known in advance, so that the same position is always used. Although it is an effective means when it becomes dirty, there is a problem that it cannot be used when it is not possible to predict in which area the stain will occur. Therefore, conventionally, only the detection and correction processing was performed. However, in this detection and correction processing, only half the codeword of the correction codeword can be corrected as in the above equation 1, so that the correction capability is reduced and the correction efficiency is reduced. There is a problem that it becomes.

The present invention has been made to solve the above-described problem, and has as its object to improve the correction efficiency in error correction in reading a two-dimensional code.

[0019]

According to the first aspect of the present invention, there is provided an error correction method for capturing a two-dimensional code representing information by a cell distribution pattern as an image. In reading information based on the two-dimensional code image, if there is a defect in the two-dimensional code image, the error caused by the defect in the two-dimensional code image is corrected by using error correction data included in the two-dimensional code image. Is the way.

The error correction data referred to here means redundant data added as a part of information of a two-dimensional code,
Using this redundant data, encoding is performed for the entire two-dimensional code or for each encoded block as described above. In addition, the defect of the two-dimensional code image refers to a situation in which a part of the two-dimensional code image cannot be correctly read due to dirt or damage of the two-dimensional code recorded on the optical information recording surface. Such a defect is not limited to dirt or damage to the two-dimensional code itself recorded on the optical information recording surface, but may occur in a process of capturing as an image, for example. For example, when a part of the two-dimensional code to be read out of the field of view of the CCD area sensor protrudes, and a part of the two-dimensional code is not captured as an image in the image, or when specular reflection occurs on the optical information recording surface. That is it.

In the present invention, first, a defective area estimation process is executed. In the defect area estimation processing, a unit information area having the above-described defect is estimated as a defect area based on the characteristics of the two-dimensional code image. The unit information area is an area which is a unit constituting information recorded as a two-dimensional code, and is formed of predetermined bits. FIG. 4 above
In the example of the QR code shown in (1), a code word composed of 8 bits corresponds to this. The defective area corresponds to a code error position. That is, this defective area estimation processing replaces the detection of an error position using error correction data.

Then, an erasure correction process is executed. In the erasure correction process, the error correction is performed on the unit information area estimated in the defective area estimation process by using the error correction data and calculating the amount of error. That is, in the present invention, the defective area is estimated based on the features of the two-dimensional code image without specifying the defective area using the error correction data (error detection). Then, erasure correction for correcting the error by calculating only the magnitude of the error is performed. As a result, if the defect area is reliably estimated by the defect area estimation processing, the error correction efficiency is doubled when the number of error correction data is the same as the conventional error correction by the conventional error correction by detection and correction. Becomes Conversely, if the correction capability is the same as in the conventional case, the number of error correction data is half that in the conventional case. Thereby, the correction efficiency can be improved.

The reason that "if the defective area is reliably estimated by the defective area estimation processing" is that there is a unit information area that is not estimated as a defective area even though there is actually a defect. It is because sex is also considered. Therefore, it is desirable to perform the erasure detection and correction described in the related art. That is, as described in claim 2, the erasure correction processing is performed on a unit information area other than the unit information area estimated in the defect area estimation processing by using error correction data to detect the presence or absence of a defect and to determine an error amount. May be calculated to correct the error. in this case,
For the unit information area not estimated as the defective area, error detection and correction are performed using the error correction data. Therefore, if there is a unit information area that is not estimated as a defective area, the correction efficiency is reduced by that much, but the correction efficiency is higher than the conventional detection and correction because the defective area is estimated in the defect estimation processing. Become.

By the way, the above-mentioned defective area estimation processing may be any processing as long as the defective area is estimated based on the characteristics of the two-dimensional code image. A specific example of the defective area estimation processing will be described below. However, it is needless to say that the processing is not limited to the processing exemplified below. (1) In a two-dimensional code, white and black cells are often randomly distributed. Therefore, white and black cells are also randomly distributed in the captured two-dimensional code image. Under such a premise, it is conceivable to adopt the following configuration.

That is, the above-described problem area estimation processing is performed by
A process of detecting an area having a predetermined area or more in which one of white or black cells is unevenly distributed at a predetermined rate or more, and estimating a unit information area including at least a part of the area as a defective area; (Claim 3).

Here, the predetermined ratio and the predetermined area can be experimentally determined based on the type of the two-dimensional code. Generally, when the two-dimensional code is viewed locally, it is highly likely that one of the white and black cells is unevenly distributed. Since the degree of the cell bias varies depending on the type of the two-dimensional code,
It is sufficient to set an area and a ratio enough to discriminate a cell bias generated under an appropriate situation from an improper cell bias caused by contamination of a two-dimensional code or the like. For example, in the case of a two-dimensional code in which the deviation of white or black cells rarely exceeds 70% when looking at an area of 1/4 of the entire code area when there is no defect, the deviation is biased at a predetermined ratio of 80% or more. In other words, a region of １ ／ or more of the distributed code area is detected.

In this way, cells that should be white due to dirt on the two-dimensional code are captured as black, and cells that should be black due to damage to the two-dimensional code are captured as white. It is effective when doing. (2) In the QR code, as a device for making it easier to read, there is a case where randomization is performed by taking an exclusive OR with an appropriate random pattern so that white and black cells are randomly arranged. Therefore, in the two-dimensional code in which the white and black cells are intentionally randomized, the following configuration may be adopted instead of the configuration (1).

That is, it is conceivable that the defective area estimation processing is processing for estimating a unit information area composed of only one of white and black cells as a defective area (claim 4). For example, in the randomized QR code, the probability that the unit information area, which is the code word shown in FIG. 4, is composed of only one of white and black cells is about 0.01%. Therefore, such a unit information area composed of only one of the white and black cells is extremely likely to be a defective area. By doing so, processing can be performed for each unit information area, and it is not necessary to detect an area that satisfies certain conditions as shown in (1) above, so that the defective area estimation processing is simplified. Is advantageous.

(3) Also, there is a high possibility that the unit information area adjacent to the unit information area detected in the above (2) has a defect. For example, when a unit information area consisting of only black cells is generated due to dirt of a two-dimensional code, white cells may be captured as black cells due to the dirt in a black cell part of an adjacent unit information area. It is because there is.

Therefore, the defective area estimation processing may be performed as processing for estimating a unit information area adjacent to the unit information area as a defective area in addition to a unit information area composed of only one of white and black cells. Good (claim 5).
This increases the possibility of reliably estimating a unit information area that is likely to have a defect.

(4) By the way, the above (1) to (3)
It is conceivable that the processing is based on a two-dimensional code image captured as binary data. In this case, for example, C
The scanning signal from the CD area sensor is converted into a signal of “0” or “1” by a binarizing circuit or the like, and is converted into image data. On the other hand, processing based on a two-dimensional code image captured as multi-valued data may be considered. When a two-dimensional code image is taken in as multi-valued data, the multi-valued data is binarized by performing a differentiation process or the like on the multi-valued data in the above-described problem area estimation processing.

[0032] However, a situation may occur where proper binarization cannot be performed due to a defect in the two-dimensional code image. For example, when the two-dimensional code is stained with gray ink or the like, it is conceivable that the white cell portion cannot be determined as white or black. This is because the amplitude (level change) of the multi-value data becomes small.

Therefore, when the two-dimensional code image is multi-valued data, the defect region estimation processing detects a region corresponding to an amplitude portion where binarization of the multi-valued data cannot be properly performed, and at least detects the region corresponding to the amplitude portion. It is conceivable that the unit information area including a part is estimated as a defective area (claim 6).

In this way, a unit information area including an area where it is highly likely that proper binarization cannot be performed is selected as a defective area. (5) As described above, the defect of the two-dimensional code image occurs not only in the case where the two-dimensional code itself is stained or damaged, but also in the process of capturing the two-dimensional code as an image. As an example, a situation can be considered in which a part of the two-dimensional code to be read protrudes from the field of view of the CCD area sensor and a part of the two-dimensional code is not captured as an image in an image.

In consideration of such a situation, when the defect area estimation processing does not include a part of the two-dimensional code in the two-dimensional code image, the unit information area including at least a part of the two-dimensional code is determined. It is conceivable to perform a process of estimating it as a defective area (claim 7).

In this case, by specifying a portion that is reflected in the image, a region that is not reflected can be determined, and a unit information region including at least a part of the region can be determined. Since the unit information area is at least partially not shown in the image, it may be a defective area.

(6) In general, a two-dimensional code includes:
It includes functional patterns such as a positioning symbol for specifying a code region and a cell position in an image, and a timing cell. For example, in the case of a QR code, the positioning symbols 510a, 510b, and 510c and the timing cells 520a and 520b shown in FIG.

Therefore, the defect area estimation processing determines whether or not there is a defect in the function pattern included in the two-dimensional code image. If there is a defect in the function pattern, a unit located around the function pattern is determined. The information area may be estimated as a defective area (claim 8).

The function pattern is composed of fixed cells. For example, the positioning symbols 510a, 510b, and 510c included in the QR code are, for example, black squares 512 each having a side length of 7 cells, and each side having a length of 5 cells.
This is a figure formed when a white square 514 corresponding to a cell and a black square 516 having a side length of 3 cells are superimposed concentrically (see FIG. 14).

Therefore, when this function pattern is not composed of fixed cells, it may be determined that there is a defect in the function pattern. Such a defect in the functional pattern may be caused by, for example, dirt or breakage of the two-dimensional code. Therefore, there is a high possibility that the unit information area located around the functional pattern also has a defect. Therefore,
A unit information area located around the function pattern is estimated as a defective area. The unit information area located in the “periphery” may be, for example, adjacent to the function pattern. Further, a unit information area outside the area may be included. How far the unit information area should be located in the periphery may be determined according to the use environment of the two-dimensional code, for example. For example, in an environment in which dirt or the like of a two-dimensional code occurs in a wide range, the “periphery” is a relatively large area. Conversely, if the environment is such that dirt or the like of the two-dimensional code occurs in a narrow range, the “periphery” is a relatively small area.

In this way, it is possible to reliably inspect the unit information area around the function pattern for a defect caused by dirt or the like. (7) Furthermore, in the unit information area of the two-dimensional code,
In addition to the data indicating the arbitrary information, there is an area where a certain fixed value or a certain range of values is stored. For example, QR
Examples of codes include an area for storing a recording format (mode) such as numbers, alphabetic characters, and kanji, and an area for storing a character number indicator indicating the number of data characters.

Therefore, if the value or a value within the range is not recorded in the unit information area in which a certain value or a certain range of values is stored, the defective area estimation processing is performed. A process for estimating an area as a defective area may be employed.

In this case, it can be reliably determined that there is a defect in the specific unit information area. Above, the failure estimation process has been described with a specific example. However, since the processes described in the above (4) to (7) are all processes assuming a special situation, the above (1) to (3) It is effective to use in combination with any of the above.

Incidentally, it is conceivable that there is no defect in the unit information area estimated in the defect area estimation processing. If the number of unit information areas free of such a problem increases, the correction efficiency may be worsened, and errors may not be corrected. This will be described.

As described above, the 23 data code words D1 to D23 of the QR code shown in FIG.
Consider a coded block of block 1 composed of 44 44 code words. When error correction in the encoded block 1 is performed by erasure correction, 44
Since 44 RS codewords are added, 44 codewords can be corrected. However, for example, when 45 codewords (unit information area) are estimated as the defective area in the above-described defective area estimation processing,
Errors cannot be corrected. At this time, actually, there may be 20 defective codewords and 25 codewords have no defective. In consideration of such a situation, it is preferable to correct an error by the method described in claim 10.

Further, in addition to the defect area estimation processing and the error correction processing, a detection and correction processing for detecting an error in the unit information area using the error correction data and for performing an error correction based on the error detection result is performed. The following (a)
Or (b) the order of execution.

(A) First, the defective area estimation processing and the erasure correction processing are executed. If the error cannot be corrected, the detection and correction processing is executed next. (B) First, the detection and correction processing is executed, and if the error cannot be corrected, the defective area estimation processing and the erasure correction processing are next executed.

That is, in consideration of the possibility that a unit information area that does not actually have a defect in the above-mentioned defect estimation area is estimated as a defect area, there are cases where it is better to perform error correction by erasure correction, There are cases where it is better to perform error correction by inspection correction.

For example, if the coding block of the QR code has 44 correction code words as shown in FIG. 4, the result of specifying the defective code word is 44
If it is up to the number, the error can be corrected by erasure correction. On the other hand, when error correction is performed by detection and correction, errors in up to 22 codewords can be corrected.

Therefore, in a situation where 25 defective areas are specified in the defective area estimation processing in spite of the fact that there are 20 defective code words,
Although the erasure correction cannot correct the error, the detection correction can correct the error. On the other hand, under the situation where there are actually 23 defective codewords and 40 defective areas are specified in the defective area estimation processing, errors cannot be corrected by detection and correction. Errors can be corrected.

Therefore, as described in (a) or (b) above, by using the detection correction and the erasure detection correction together, the correction efficiency can be further improved. Although the invention has been described as an error correction method, the error correction method as described above captures a two-dimensional code as an image and is positioned as a part of a decoding process based on the two-dimensional code image.

Therefore, the present invention is realized as a two-dimensional code reading method characterized in that an error is corrected using the above-described error correction method, and information recorded as a two-dimensional code is read based on a two-dimensional code image. You can also. Further, the present invention can be realized as an invention of a two-dimensional code reading apparatus characterized by reading information recorded as a two-dimensional code using the two-dimensional code reading method.

The function of executing the above-described error correction method or two-dimensional code reading method can be provided, for example, as a program activated on a digital circuit or a computer system. In the case of such a program, for example, it is recorded on a computer-readable recording medium such as a floppy (registered trademark) disk, a magneto-optical disk, a CD-ROM, and a hard disk, and is loaded into a computer system as needed and activated. Can be used. In addition, the ROM or the backup RAM is used to record the program as a computer-readable recording medium, and the ROM or the backup RAM is stored.
The AM may be incorporated in a computer system and used.

[0054]

Embodiments of the present invention will be described below with reference to the drawings. [First Embodiment] FIG. 1 is a block diagram of a two-dimensional code reader 1 according to a first embodiment. The two-dimensional code reader 1 according to the present embodiment includes a control circuit 10, a light emitting diode (illumination LED) 11, a CCD area sensor 12, an amplification circuit 13, an A / D conversion circuit 14, an address generation circuit, 17, a synchronization pulse generating circuit 16, and an image memory 20.
, Switch group 31, liquid crystal display 32, communication I / F
It is configured centering on the circuit 33.

The control circuit 10 includes a CPU, ROM, RA
It is configured as a computer system having M, I / O, and the like. The control circuit 10 executes a reading process and the like to be described later according to a program stored in the ROM. The control circuit 10 includes a switch group 31 and a liquid crystal display 3.
2 and the communication I / F circuit 33, and also performs input / output control.

The illumination LED 11 illuminates red light for illumination on a two-dimensional code which is optical information to be read. The CCD area sensor 12 has a plurality of two-dimensionally arranged CCDs, which are light receiving elements, captures the outside world, and outputs the two-dimensional image as a horizontal scanning line signal. In this embodiment, the scanning line signal from the CCD area sensor 12 is amplified by the amplifier circuit 13 and output to the A / D conversion circuit 14.

The amplification circuit 13 amplifies the scanning line signal output from the CCD area sensor 12 with an amplification factor corresponding to the gain control voltage from the control circuit 10.
The A / D conversion circuit 14 converts an analog scanning line signal into a digital signal and outputs the digital signal to the image memory 20. A / D
The data converted into a digital signal by the conversion circuit 14 is multi-valued data, and the multi-valued data is stored in the image memory 20. Further, in the multi-valued data, each pixel has a gradation value of 8 bits (“0” to “255”).

The synchronizing pulse generating circuit 16 outputs a synchronizing pulse sufficiently finer than the two-dimensional image data pulse output from the CCD area sensor 12. The address generation circuit 17 counts this synchronization pulse, and
Generate an address for 0. The multilevel data is written in 8-bit units for each address.

The switch group 31 includes a reading switch for instructing a user to start reading processing, a numeric keypad, and various function keys, and is used for inputting information. The liquid crystal display 32 is configured as, for example, a two-tone LCD, and is used for displaying read optical information and the like.

The communication I / F circuit 33 communicates with a host computer or the like. For example, the communication I / F circuit 33 transmits data to an external device via a communication light emitting element (not shown), or transmits light to a communication light receiving element (not shown). (For example, a program for operating the system, a command for instructing data transmission of the decoded two-dimensional code, etc.) from the external device. Needless to say, a configuration for connecting to the host computer by a wire may be adopted.

In the two-dimensional code reader 1 of this embodiment having such a configuration, the scanning line signal output from the CCD area sensor 12 is amplified by the amplifier circuit 13, and the amplified scanning line signal is output to the A line. The data is converted into multi-value data by the / D conversion circuit 14 and stored in the image memory 20. Then, the control circuit 10 decodes the two-dimensional code based on the multi-value data stored in the image memory 20.

In the two-dimensional code reader of this embodiment,
It is characterized by so-called error correction when reading a dimensional code. In the QR code of this embodiment, information is recorded in units of 8-bit code words. More specifically, as shown in FIG. 4, data blocks 1, E1 to E4 each including 23 data code words D1 to D23.
4, RS block 1 consisting of 44 RS code words, data block 2 consisting of 23 data code words D24 to D46, and RS block 2 consisting of 44 RS code words E45 to E88. . Then, the RS encoding is performed in units of a block 1 composed of the data block 1 and the RS block 1 and a block 2 composed of the data block 2 and the RS block 2. Error correction is performed. In the QR code of the present embodiment,
Randomization has been performed to make the cells easier to read.
The randomization means that white and black cells are intentionally distributed so as not to be biased. For example, randomization is performed by taking an exclusive OR with an appropriate random pattern corresponding to a two-dimensional code. Of course, at the time of decoding, decoding is performed using this random pattern.

On the premise of such a QR code, the reading process executed by the control circuit 10 will be described with reference to the flowchart of FIG. First step S1000
Then, a failure estimation process is performed. This process is a process of estimating a defective codeword among the codewords of the two-dimensional code image. Here, a flag is set for a code word having a defect. This S1000
Corresponds to the process of detecting the error position of the encoded blocks 1 and 2.

Here, the failure estimation processing will be described with reference to the flowchart of FIG. In a first step S2000, one codeword is selected. In this defect estimation process, codewords are selected one by one in order to determine whether or not there is a defect for each codeword.
When the failure estimation processing is repeated and the processing of S2000 is repeated, D2 → D3 →.
Codewords are selected one by one in the order of D46 → E1 →... → E88. At S2010, the selected codeword is decoded.

Then, in the next S2020, it is determined whether or not the codeword is a specific codeword. Data code word D
In addition to the so-called data having an arbitrary value, 1 to D46 include a mode having a certain fixed value or a certain range of values, a character number indicator, a terminator, and a data code word called dummy data. here,
Thus, a data codeword having a certain value or a certain range of values is determined as a specific codeword. If it is determined that the codeword is a specific codeword (S2020: YES), the process proceeds to S2030.
On the other hand, if it is determined that the codeword is not a specific codeword (S
2020: NO), and proceeds to S2050.

At S2030 to which the process proceeds when the selected codeword is a specific codeword, at S2010
It is determined whether the decoded value of is normal. For example, when the value of this specific codeword is shown for a certain QR code, the mode is “01”, “02”, “0” in hexadecimal notation.
3 "," 04 ", or" 08 ". The character number indicator has a value of 1 to 106 in decimal notation. The terminator is “00” in hexadecimal notation, and the dummy data is “EC” and “1” in hexadecimal notation.
1 ". Therefore, if the specific codeword takes any other value, a negative determination is made here. If it is determined that the decoded value of the codeword is normal (S2030: YES), the process proceeds to S2070 without executing the process of S2040. On the other hand, when it is determined that the decoded value of the codeword is not normal (S
2030: NO), the flag corresponding to the code word is set in S2040, and then the flow shifts to S2070.

At S2050 to which the process proceeds when the selected codeword is not a specific codeword, it is determined whether or not the codeword is composed of only one of white and black cells. Here, when it is determined that the code word is composed of both white and black cells (S20)
50: NO), the process of S2060 is not performed, and S207
Move to 0. On the other hand, if it is determined that the image data is composed of only white or black cells (S2050: YES), S20
At 60, the flag corresponding to the code word is set, and thereafter, the flow shifts to S2070.

For example, it is highly possible that a stain on the code as shown in FIG. 5A or a damaged portion of the code as shown in FIG. 5B is constituted only by black or white cells. for that reason,
As described above, the flag is set on the assumption that there is a defect in the code word composed of only one of the white and black cells. Thereby, for example, the codewords D33 to D37, E5 to E9 hatched in FIG.
The flag is set corresponding to.

In S2070, it is determined whether all codewords have been selected. D1 to D46 and E1 to E8
If all of the 134 codewords No. 8 have been selected, a positive determination is made here. Here, when it is determined that all codewords have been selected (S2070: YE
S), the fault estimation processing ends. On the other hand, if there is a codeword that has not been selected (S207).
0: NO), and repeat the processing from S2000.

The description returns to the flowchart of FIG. In S1010, erasure detection and correction are performed. The erasure detection and correction is to perform error correction by calculating the magnitude of the error on the assumption that the error position of the code is specified as described above. In consideration of the possibility that there is an unspecified error position, error positions are detected and corrected at positions other than the specified error position (erasure position). Here, S
For the code word for which the flag has been set at 1000, the error magnitude of this code word is calculated and error correction is performed. On the other hand, for a code word for which a flag is not set, error position detection and correction are performed.

For example, D33 to D37 and E5 shown in FIG.
When the flag is set in the code word of E9, the error position of the encoded block 1 is specified as the code words of E5 to E9. The error position of the encoded block 2 is specified as a codeword of D33 to D37.

Therefore, for block 1, error correction is performed as follows. A syndrome polynomial is obtained from a code composed of data code words D1 to D23 and RS code words E1 to E44. Using the erasure polynomial obtained from the code words E5 to E9 (erasure positions), the syndrome polynomial is corrected to obtain the erasure numerical polynomial.

Code words E5 to E9 (erasure positions)
And error correction is performed by calculating the magnitude of the error. The same applies to block 2. In the next step S1020, it is determined whether or not the correction was successful. QR code 500 of this embodiment
Are encoded in units of blocks 1 and 2, and 44 RS code words are added to each of blocks 1 and 2. Therefore, if the number of codewords in blocks 1 and 2 estimated in S1000 is up to 44, erasure detection and correction can be performed. If it is determined that the correction has been made (S1020: YES), the process proceeds to S1050. On the other hand, when it is determined that the correction could not be performed (S1020:
NO), and it transfers to S1030.

In S1030, detection and correction are performed. The detection and correction is to detect an error position of a code and to calculate the magnitude of the error to correct the error. For example, if the error cannot be corrected for the block 1, the error is corrected as follows. A syndrome polynomial is obtained from a code composed of data code words D1 to D23 and RS code words E1 to E44.

An error locator polynomial and an error numerical polynomial are obtained from the syndrome polynomial. An error codeword (error position) is determined from the error locator polynomial by the Chien search method, and the magnitude of the error with respect to the codeword is determined to perform error correction. Note that S
The calculation methods of the erasure detection and correction processing and the detection and correction processing of 1010 and S1030 have been well known among those skilled in the art, and thus detailed description thereof will be omitted. For example, when it is realized by a computer system, the Euclidean method, the Peterson method, or the like is used. For example, JP-A-6-77
No. 844 discloses a calculation method using the Euclidean method.

In the next step S1040, it is determined whether or not the correction has been made. In the QR code according to the present embodiment, encoding is performed in units of blocks 1 and 2, and 44 RS code words are added to each encoded block. In detection and correction, two error code polynomials and an error numerical polynomial are obtained, so that two RS codewords are required to correct one codeword. Therefore, errors in up to 22 codewords in an encoded block can be corrected. If it is determined that the correction has been made (S1040: YES),
The process moves to S1050. On the other hand, if it is determined that the correction has failed (S1040: NO), the process proceeds to S1060.

Step S105 when error correction is successful
At 0, the decoding process is performed, and thereafter, the main reading process ends. The information recorded as a two-dimensional code is read by this decoding process. On the other hand, in S1060 to which the process proceeds when error correction has not been performed, error processing is performed, and then the main reading processing ends.

That is, in this embodiment, a defective code word area is estimated based on the features of the two-dimensional code image (S1000 in FIG. 2, see FIG. 3). Then, erasure detection and correction for correcting the error by calculating the magnitude of the error are performed (S1010 in FIG. 2). As a result, if a faulty codeword is reliably estimated in the fault estimation process, the error correction efficiency is doubled as compared with the conventional error correction by detection and correction. In detection and correction, two RS codewords are required to correct one codeword. However, in erasure detection and correction, it is assumed that an error position is specified, and there is no need to specify an error position in correction processing. Therefore, if there is one RS codeword, error correction is possible. Conversely, if the correction capability is the same as the conventional one, the number of RS code words is half that of the conventional one. Thereby, the correction efficiency can be improved.

In the erasure detection and correction, some of the defective codewords may not be able to be estimated by the defect estimation process. Make corrections. When a defective codeword that cannot be estimated is detected, the correction efficiency is lower than when all the defective codewords are estimated. However, even in such a case, since only the magnitude of the error is calculated for the estimated defective codeword and the error is corrected, the correction efficiency is improved as compared with the conventional method.

Further, even if the error cannot be corrected by the erasure detection and correction in S1010, that is, even if the estimated number of faulty codewords exceeds the number of RS codewords, the actual faulty Codeword is R
If it is less than half the S code word, the error can be corrected by conventional detection and correction. Therefore, in the present embodiment, it is determined whether or not the correction has been made by the erasure detection and correction (S102 in FIG. 2).
0), if it cannot be corrected (S1020: N
O), the same detection and correction as in the prior art is performed (S10).
30). As a result, the correction efficiency can be further improved.

In the defect estimating process of the present embodiment, as a device for making it easier to read, randomization is performed by taking an exclusive OR with an appropriate random pattern so that white and black cells are randomly arranged. We focused on what has been done for QR codes. Randomized QR code 50
At 0, the probability that each codeword shown in FIG. 4 is composed of only one of white and black cells is about 0.01%.

Therefore, it is highly probable that a code word composed of only white or black cells has a defect. In the present embodiment, it is determined whether or not the code word is composed of only white or black cells (S205 in FIG. 3).
0), if configured (S2050: YE
S), a flag is set as having a defect in the code word (S2060). As a result, a codeword as an error position can be reliably selected. Further, the failure estimation processing of the present embodiment is processing for each codeword (S2000, S2070), and the codeword position information and the like are stored in advance as information for decoding, so that the design can be made relatively easily.

Further, in the defect estimation processing of the present embodiment, the data code word is decoded for a data code word in which a fixed value or a fixed range value included in the data code words D1 to D46 is stored. Then, it is determined whether the value or a value within the range is stored (S2030).
If the decoded value is not normal (S203
0: NO), and sets a flag assuming that the data code word has a defect (S2040). As a result, the defect can be reliably determined for a specific data code word.

In this embodiment, it is assumed that a code word composed of only white or black cells has a defect. However, a code word adjacent to a code word composed of only white or black cells has a defect. There is likely to be. This is because the dirt or damage of the QR code is likely to have spread to the periphery of the code word as indicated by the two-dot chain line in FIG.

Therefore, as a modification of the first embodiment, in S2060 in FIG. 3, not only the flag corresponding to the selected code word but also the flag corresponding to the adjacent code word is set. It may be.
Thereby, the code word D33 as shown in FIG.
~ D37, codewords D19 ~ adjacent to E5 ~ E9
D25, D32, D38, E4, E10, E21-E2
7, the flag is set. In this way, it is more likely that a code word that is likely to have a defect can be reliably estimated.

The failure estimation processing of FIG. 3 in this embodiment corresponds to the failure area estimation processing.
The processing of 0 corresponds to the erasure correction processing, and the processing of S1040 corresponds to the detection correction processing. [Second Embodiment] The second embodiment differs from the first embodiment in the defect estimation processing. Therefore, only the failure estimation processing will be described as the description of the present embodiment.
Since the hardware configuration is the same as in the first embodiment, the same reference numerals are used.

The failure estimating process of this embodiment is as shown in the flowchart of FIG. First step S3
In 000, the code area is calculated. This process calculates the area of the two-dimensional code area in the image. Then, in subsequent S3010, 2 calculated in S3000 is used.
The area of one codeword is calculated from the area of the dimensional code area.

In the next step S3020, an area larger than the area of one code word calculated in S3010 and in which white or black cells are biased is detected. In this embodiment,
An area consisting only of white or an area consisting only of black is detected. For example, an area blackened due to dirt or the like of the two-dimensional code as shown in FIG. 5A or an area whitened due to damage or the like as shown in FIG. 5B is detected. .

At S3030, one of the detected areas is selected. Then, in S3040, a code word at least partially included in the area is specified, a flag corresponding to the code word is set, and then, the process proceeds to S3040.
Move to 50. For example, when a region indicated by a two-dot chain line in FIG. 7 is selected, code words including at least a part of the region are D19 to D25 and D32 to D3.
8, E4 to E10 and E21 to E27, the flags are set for these codewords.

In S3050, it is determined whether or not all the detected areas have been selected. If all the detected areas have been selected (S3050: YES), the failure estimation processing ends. On the other hand, if there is an area that has not been selected (S3050: NO), the processing from S3030 is repeated.

With such a defect estimating process, a code word having a defect can be estimated as in the first embodiment. Therefore, an effect similar to that of the first embodiment can be obtained in terms of improvement in correction efficiency. In the present embodiment, the area is equal to or larger than one code word.
An area consisting of only white or black cells was detected. that is,
As described above, the QR code 500 of the present embodiment is randomized, which is based on the fact that it is extremely unlikely that a code word is composed of only white or black cells.

In the defect estimating process shown in FIG. 8, the proportion of cells to be distributed and the area of the area to be detected are determined, for example, experimentally based on the type of the two-dimensional code. I just need. In other words, an area and a ratio are set so as to be able to distinguish between the cell bias generated under an appropriate situation and the inappropriate cell bias generated due to contamination of the two-dimensional code, etc.
What is necessary is just to detect the area | region which meets this condition, and it is not limited to the detection conditions of S3010 and S3020 mentioned above. For example, when there is no defect, 1 /
In the area of No. 4, if the two-dimensional code is such that the deviation of the white or black cells rarely exceeds 70%, the processing proceeds to S301 in FIG.
0, the area of 1/4 of the code area is calculated, and S30
At 20, it is possible to detect a region of ８０ or more of the code area where white or black cells are unevenly distributed at 80% or more.

In the failure estimating process as in this embodiment, a processing step for detecting an area where white or black cells are unevenly distributed is required. By changing the area, various 2
It is advantageous in that it can be applied to dimensional codes.

In this embodiment, the failure estimation processing shown in FIG. 8 corresponds to the failure area estimation processing. [Others] In addition to or instead of the defect estimating process shown in the first or second embodiment, a code word defect may be estimated as follows.

(I) There may be a situation where a two-dimensional code image as multi-valued data taken into the image memory 20 cannot be properly binarized. For example, when the two-dimensional code is stained with gray ink or the like, it is conceivable that the white cell portion cannot be determined as white or black. This is because, as shown in FIG. 9, the amplitude (level change) of the multivalued data corresponding to the stained portion becomes small.

Therefore, in S3020 of the failure estimating process of the second embodiment, an area corresponding to an amplitude portion where multi-valued data cannot be properly binarized may be detected. For example, the area is detected by determining whether the amplitude is equal to or less than a predetermined value.
By doing so, a flag is set in S3040 assuming that there is a defect in the code word at least partially included in this area. As a result, a codeword including an area where it is highly likely that proper binarization cannot be performed can be selected.

In this case, similarly to the second embodiment, the failure estimation processing shown in FIG. 8 corresponds to the failure area estimation processing. (R) The defect of the two-dimensional code image occurs not only in the case where the two-dimensional code itself is dirty or damaged, but also in the process of capturing the two-dimensional code as an image. For example, as shown in FIG.
It is conceivable that a part of the QR code 500 to be read protrudes from the field of view of the CCD area sensor 12 and part of the QR code is not captured in the image.

Therefore, in this case, it is conceivable to execute an estimation process based on a code area as shown in FIG. That is, four vertices of the two-dimensional code area are detected (S4000), and it is determined whether the detected four vertices are in the image (S4010). If it is within the image (S4010: YES), the process ends. On the other hand, if it is not in the image (S4010: NO), an area not shown in the image is calculated (S4020). The area not shown in the image can be specified by calculating the area shown in the image. Then, a flag is set assuming that there is a defect in a code word at least partially included in an area not shown in the image (S4030, see FIG. 12).

In this case, the estimation processing based on the code area shown in FIG. 10 corresponds to the defective area estimation processing. (I) In addition, a two-dimensional code generally includes a functional pattern such as a positioning symbol for specifying a code area or a cell position in an image, and a timing cell.
For example, in the case of a QR code, the positioning symbols 510a, 510b, and 510c and the timing cells 520a and 520b shown in FIG.

Therefore, an estimation process based on the function pattern shown in FIG. 11 may be executed as follows. That is, FIG.
As shown in (5), it is determined whether or not there is a defect in the specific pattern (S5000), and if there is no defect (S5000:
NO) The process ends, and if there is a problem (S5000: Y
ES), a flag corresponding to a code word adjacent to the specific pattern is set (S5010).

The function pattern is composed of fixed cells. For example, the positioning symbols 510a, 510b, and 510c included in the QR code are, for example, black squares 512 each having a side length of 7 cells, and each side having a length of 5 cells.
This is a figure formed when a white square 514 corresponding to a cell and a black square 516 having a side length of 3 cells are superimposed concentrically (see FIG. 14).

Therefore, when the function pattern is not composed of fixed cells, for example, when the positioning symbol and the timing cell are dirty as shown in FIG. 12B, the function pattern has a defect. It may be. At this time, there is a high possibility that a code word located around the function pattern has a defect. Therefore, it is presumed that a code word located around the function pattern has a defect. For example, if there is a defect in the upper left positioning symbol in FIG. 7, E63 to E68, E78, E8
It is assumed that there is a defect in the code word of 3, E88. The code word located in the “periphery” may be adjacent to the function pattern as described above, but may be determined according to the use environment of the two-dimensional code. For example, in an environment in which dirt or the like of a two-dimensional code occurs in a wide range, the “periphery” is a relatively large area. Conversely 2
In an environment in which dirt on a dimensional code or the like occurs in a narrow range, the "periphery" is a relatively small area. In this way, it is possible to reliably detect a defect in the code word around the function pattern caused by dirt or the like.

In this case, the estimation processing based on the function patterns shown in FIG. 11 corresponds to the defective area estimation processing.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a two-dimensional code reader according to an embodiment.

FIG. 2 is a flowchart showing a two-dimensional code reading process.

FIG. 3 is a flowchart illustrating a defective area estimation process according to the first embodiment.

FIG. 4 is an explanatory diagram showing the relationship between codewords constituting code information and encoded blocks.

FIG. 5 is an explanatory diagram showing a defect of a two-dimensional code image due to dirt and breakage of a code.

FIG. 6 is an explanatory diagram showing a state where a defective codeword is specified;

FIG. 7 is an explanatory diagram showing a defect occurrence area due to dirt.

FIG. 8 is a flowchart illustrating a failure estimating process according to the second embodiment.

FIG. 9 is an explanatory diagram showing the relationship between code contamination and the amplitude of multi-value data.

FIG. 10 is a flowchart illustrating an estimation process based on a code area.

FIG. 11 is a flowchart illustrating an estimation process based on a function pattern.

12A is an explanatory diagram illustrating a situation where a part of a code area is not captured as an image, and FIG. 12B is an explanatory diagram illustrating a state in which a function pattern has a defect.

FIG. 13A is an explanatory diagram illustrating a barcode,
(B) is an explanatory view showing a two-dimensional code.

FIG. 14 is an explanatory diagram showing a configuration of a QR code.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Two-dimensional code reader 10 ... Reading control circuit 11 ... Light emitting diode 12 for illumination 12 ... CCD area sensor 13 ... Amplification circuit 14 ... A / D conversion circuit 16 ... Synchronous pulse generation circuit 17 ... Address generation circuit 20 ... Image memory 31 ... Switch group 32 ... Liquid crystal display 33 ... Communication I / F circuit

Claims (13)

[Claims] 1. A two-dimensional code representing information by a cell distribution pattern is taken in as an image, and the information is read based on the two-dimensional code image.
In the case where there is a defect in the two-dimensional code image, in an error correction method for correcting an error caused by the defect of the two-dimensional code image using error correction data included in the two-dimensional code, the information is a unit constituting the information. Executing a defect area estimation process of estimating a unit information area having the defect as a defect area corresponding to a code error position based on characteristics of the two-dimensional code image without using the error correction data. Then, an erasure correction process of calculating an error amount and correcting an error is performed on the unit information area estimated in the defect area estimation process using the error correction data. Error correction method. 2. The error correction method according to claim 1, wherein the erasure correction processing is performed on the error correction data for a unit information area other than the unit information area estimated in the defect area estimation processing. An error correction method comprising: detecting whether there is a defect and calculating an error amount to correct the error. 3. The error correction method according to claim 1, wherein the defective area estimating process is performed on an area having a predetermined area or more in which one of white and black cells is unevenly distributed at a predetermined ratio or more. And correcting the unit information area including at least a part of the area as the defective area. 4. The error correction method according to claim 1, wherein said defective area estimating process estimates said unit information area composed of only one of white and black cells as said defective area. An error correction method characterized by processing. 5. The error correction method according to claim 4, wherein the defective area estimating process is performed in addition to the unit information area configured by only one of white and black cells and an adjacent area to the unit information area. An error correction method characterized by a process of estimating a unit information area to be performed as the defective area. 6. The error correction method according to claim 1, wherein said two-dimensional code image is multi-valued data, and said defect area estimating process is performed by using an appropriate two-dimensional data of said multi-valued data.
An error correction method, comprising: detecting a region corresponding to an amplitude portion that cannot be converted into a value; and estimating the unit information region including at least a part of the region as the defective region. 7. The error correction method according to claim 1, wherein the defective area estimation processing is performed when a partial area of the two-dimensional code is not shown in the two-dimensional code image. An error correction method, wherein the unit information area including at least a part of the area is estimated as the defective area. 8. The error correction method according to claim 1, wherein the defect area estimating process determines whether there is a defect in a function pattern included in the two-dimensional code image,
If there is a defect in the function pattern, the error correction method is a process of estimating the unit information area located around the function pattern as the defect area. 9. The error correction method according to claim 1, wherein the defective area estimation processing includes the step of determining the value or a value in a unit information area in which a fixed value or a fixed range of values is stored. An error correction method characterized in that when a value within the range is not recorded, the unit information area is estimated as the defective area. 10. The error correction method according to claim 1, further comprising: performing error detection of said unit information area using said error correction data, and performing error correction based on said error detection result. An error correction method including a detection and correction process to be performed, and performing the process in any of the following order (a) or (b). (A) First, the defective area estimation processing and the erasure correction processing are executed. If the error cannot be corrected, the detection correction processing is executed next. (B) First, the detection and correction processing is executed, and if an error cannot be corrected, the defective area estimation processing and the erasure correction processing are next executed. 11. A two-dimensional code reading method, comprising: correcting an error using the error correction method according to claim 1; and reading the information based on the two-dimensional code image. 12. A two-dimensional code reading apparatus for reading information recorded as a two-dimensional code by using the two-dimensional code reading method according to claim 11. 13. An error correction method according to claim 1 or a two-dimensional code reading method according to claim 11, recorded as a program to be executed by a computer system. recoding media.