JP2010079725A - Magnetic ink character reader - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve the recognition rate of magnetic ink characters, and increase processing speed. <P>SOLUTION: The magnetic ink character reader is provided with: a magnetic data acquiring means for acquiring magnetic data included in the magnetic ink characters printed in a document; a magnetic character recognizing part for recognizing the magnetic ink characters from the magnetic data; an optical character reading part for optically reading the magnetic ink characters, and acquiring image data; and an optical character recognizing part for recognizing the magnetic ink characters from the image data. The optical character recognizing part recognizes the magnetic ink characters on the basis of recognition candidate characters obtained by the magnetic character recognizing part. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a magnetic ink character reading device that includes at least a magnetic ink character reading unit and an optical character reading unit, and performs reading processing of magnetic ink characters printed on a processing target medium such as a check.

  In commercial transactions, shopping at stores, payments at restaurants, etc., payments may be made using checks or the like. Generally, an account number, a monetary amount, and the like are printed with magnetic ink characters at a predetermined place (lower column) of a check. In the check-out process by check, the validity of the check is confirmed by reading the magnetic ink characters and referring the read data to a predetermined organization.

  There is MICR (Magnetic Ink Character Recognition) as a technique for reading magnetic ink characters. In this technique, magnetic information contained in a magnetic ink character is read by a magnetic reader, and the character is recognized based on the read magnetic information (Patent Document 1).

  As described above, since the magnetic ink information includes an account number, a monetary amount, etc., it is important to read magnetic ink characters in the check settlement process. Therefore, a device for accurately reading magnetic ink characters has been devised.

In the magnetic ink character reading device described in Patent Document 2, in addition to magnetic character recognition processing in the magnetic ink character reading unit, optical character recognition processing (OCR: Optical Character Recognition) is performed, and the magnetic ink character reading unit can read the character. The recognition rate is improved by optically recognizing the missing characters.
JP 49-49545 A JP 7-182448 A

  However, there are cases where a character that cannot be recognized by the magnetic character recognition process does not succeed even if an attempt is made to recognize the character by the optical character recognition process.

  FIG. 16 is an enlarged view around the magnetic ink character of an actual check. In FIG. 16, the character “0” surrounded by the frame line 100 in the magnetic ink character string “7680” is covered with a part of the signature character handwritten by the customer. Due to the influence of the sign character, the character “0” surrounded by the frame line 100 has a shape that looks close to “8”.

  The character “0” surrounded by the frame line 100 may have a recognition result of “8” when a simple optical character recognition process as described in Patent Document 2 is performed. There is concern about the deterioration of the recognition results.

  In general, in optical character recognition processing, there is a problem that the more dictionary data to be compared with the recognition target, the longer the processing time. In order to increase the speed of the optical character recognition process, it is desirable that the process be performed with only the minimum necessary dictionary data.

  The present invention has been made in view of the above-described problems, and an object of the present invention is to improve the recognition rate of magnetic ink characters and to increase the processing speed.

  In order to solve the above-described problems and achieve the object, a magnetic ink character reading device according to the present invention includes magnetic data acquisition means for acquiring magnetic data included in magnetic ink characters printed on a document, and the magnetic data. The magnetic character recognition means for recognizing the magnetic ink character, the optical character reading means for optically reading the magnetic ink character to obtain image data, and the optical for recognizing the magnetic ink character from the image data Character recognition means, wherein the optical character recognition means recognizes the magnetic ink character based on a recognition candidate character that is a recognition result of the magnetic character recognition means.

  The magnetic ink character reading device control method according to the present invention includes a magnetic data acquisition step of acquiring the magnetic data by using magnetic data acquisition means for acquiring magnetic data included in the magnetic ink characters printed on the document. A magnetic character recognition step for recognizing the magnetic ink characters from the magnetic data, and an image data acquisition step for acquiring image data using an optical character reading means for reading the magnetic ink characters printed on the document, An optical character recognition step for recognizing the magnetic ink character from the image data, the optical character recognition step based on a recognition candidate character that is a recognition result of the magnetic character recognition step. It is characterized by recognizing characters.

  According to the present invention, it is possible to improve the recognition rate of magnetic ink characters and to increase the processing speed.

  Hereinafter, an embodiment of a magnetic ink character reading device according to the present invention will be described in detail with reference to the drawings.

  The present invention relates to a magnetic ink character reader and can be applied to processing of a medium to be processed such as a check as well as a check. As a most typical example, in the following description, a medium to be processed ( A description will be given using a check as a manuscript.

  In addition, although the present invention does not limit the font of the magnetic ink character to be read, the target of the magnetic ink character to be read in the present embodiment is E13B for easy understanding. FIG. 1 shows a list of characters in the E13B font.

  FIG. 2 is a perspective view showing an appearance of the magnetic ink character reading device 1 according to the embodiment.

  The magnetic ink character reading device 1 is provided with a hopper portion 2 for stacking checks. A conveyance path 3 conveys the check in a standing state, and is formed in a groove shape from the hopper portion 2 to the discharge port 4. The checks stacked on the hopper 2 are conveyed one by one in the direction of the arrow along the conveyance path 3 by a conveyance mechanism (not shown). Note that the discharge port 4 is provided at two locations. For example, if the distribution condition is set as to whether or not the magnetic ink characters can be read normally, the check is distributed to the two discharge ports 4 according to the reading result.

  FIG. 3 shows an example of a check. On the check 90, magnetic ink characters are printed in a predetermined area 91 as shown in FIG. The area 91 of the check 90 includes, from the left side of FIG. 3, “auxiliary own bank column II”, “auxiliary own bank column I”, “exchange office column (exchange number, institution code)”, “own bank data column (store No. bill no. Account No) ”and“ Money amount column ”, and magnetic information is recorded in each column. The description will be made assuming that the magnetic ink characters are read from the right side to the left side in the reverse order of the characters.

  FIG. 4 is a block diagram showing the internal configuration of the magnetic ink character reading device 1 of this embodiment.

  As shown in the schematic internal configuration diagram of FIG. 4, the magnetic ink character reading device 1 controls the magnetic reading unit 10, the optical reading unit 20, the memory 34, and these units that read information from a check conveyed on the conveyance path 3. The main control unit 31 is connected to the host device 40 via the external interface 32. Each unit is connected via a bus line 33 and can receive a command from a main control unit 31 such as a CPU. An arrow D written along the conveyance path 3 indicates the conveyance direction of the check.

  The magnetic reading unit 10 is a magnetic head 6 (magnetic data acquisition means) that reads magnetic data in magnetic ink characters printed on a check, and A / D conversion or smoothing is performed on the magnetic data read by the magnetic head 6. A magnetic data processing unit 11 that performs processing such as the above, a character cutout unit 15 that detects position information of each character from the magnetic data, a magnetic data dictionary table 12 that is used in magnetic recognition processing described later, and a character cutout unit 15 includes a magnetic recognition processing unit 13 (magnetic character recognition means) that performs recognition processing on each character data clipped by 15 using the magnetic data dictionary table 12. Although the character cutout unit 15 and the magnetic recognition processing unit 13 of the present embodiment are configured by hardware, some or all of them can be configured as CPU software such as the main control unit 31. In the magnetic head 6, a portion for magnetizing the magnetic ink is arranged on the upstream side where the check flows, and a reading head for reading the magnetized magnetic ink is arranged on the downstream side.

  Further, the optical reading unit 20 optically reads the magnetic ink character in the check and the image data read by the image reading sensor 7 (optical character reading means) is an image suitable for optical recognition processing. An optical data processing unit 21 that performs various processes to achieve the above, an optical data dictionary table 23 that is used in an optical character recognition process to be described later, and optical data dictionary table 23 for image data. An optical recognition processing unit 22 (optical character recognition means) that performs character recognition processing is included. Although the optical data processing unit 21 and the optical recognition processing unit 22 of the present embodiment are configured by hardware, some or all of them may be configured as CPU software such as the main control unit 31.

  Next, a flow of a series of character recognition processing in the magnetic ink character reading device 1 will be described with reference to the flowchart of FIG.

  In step S101, magnetic data in the magnetic ink printed on the check is acquired using the read signal output from the magnetic head 6.

  Magnetic material is mixed with the magnetic ink characters printed on the check. Therefore, when the check is conveyed by the conveyance path 3 and passes in front of the magnetic head 6, the magnetic ink character is magnetized at a portion that magnetizes the magnetic ink. When the magnetized magnetic ink character passes in front of the reading head, the magnetic flux changes according to the shape of the magnetic ink character, so that a waveform corresponding to the shape of the character is output from the magnetic head 6.

  FIG. 6 shows a waveform of a read signal when magnetic information is read by a magnetic head for “0” and “1” which are one of E13B fonts. The left side is the waveform when “0” is read by the magnetic head, and the right side is the waveform of “1”. The vertical axis represents the output voltage value (128 = 0V), and the horizontal axis represents the passage of time.

  The read signal (magnetic data) output from the magnetic head 6 is sent to the magnetic data processing unit 11.

  The magnetic data processing unit 11 performs A / D conversion on the acquired magnetic data, and converts the magnetic data into digital data. Since A / D conversion is a known technique, detailed description thereof is omitted here.

  In step S102, the image reading sensor 7 performs optical reading (image data acquisition) of the magnetic ink character portion of the check. The read image is processed by the optical data processing unit 21 through various image processing such as shading processing and gamma correction. Since the shading process and the gamma correction are known techniques, a detailed description thereof is omitted here.

  In step S103, a magnetic character portion is cut out from the magnetic data (hereinafter referred to as a sampling waveform) converted into digital data in step S101.

  The character cutout is determined by detecting the character start position. Specifically, the character start position is detected using the tip of the sampling waveform as the right end data of the check. For this purpose, first, the first peak of the sampling waveform is detected. Next, the character start position is obtained based on the peak position found first, and a predetermined range from the character start position is cut out as data for one character.

  A series of flow of character cutout processing in step S103 will be described with reference to the flowchart of FIG. In the description, D represents sampling data, and D (X) represents sampling data at the position of X. R in the description is a constant uniquely determined by the character font and sampling rate used to cut out one character.

  First, in order to scan data from the right end of the check, X is reset to 0, and at the same time, the character count N is also reset to 0 (step S201). Next, D (X) is compared with a predetermined threshold value K (step S202). If D (X) is larger than K (step S202: Yes), D (X) is compared with D (X + 1). (Step S204). If D (X) is larger than D (X + 1) (step S204: Yes), it can be determined that D (X) is the first vertex position in the waveform data. Therefore, position information indicating that the data from X to X + R is the Nth character data is stored in the memory 34 (step S205). Thereafter, X is advanced by R, and N is counted by 1 (step S206). If D (X) is smaller than K (step S202: No) or D (X) is smaller than D (X + 1) (step S204: No), X is advanced by one (step S203). Then, it is determined whether or not X is less than the length Length of the sampling data from the right end to the left end of the check (step S207). If it is smaller (step S207: Yes), the process returns to step S202, and is equal to or exceeds. In the case (step S207: No), the series of character cut-out processing ends, and the process proceeds to step S104.

  In step S104, the character count N is reset to zero.

  In step S105, the N-th magnetic character recognition process (MICR recognition process) is performed by the magnetic recognition processing unit 13.

  FIG. 8 is a flowchart showing a series of MICR recognition processing in step S105.

  First, in step S301, the magnetic recognition processing unit 13 acquires magnetic data of the Nth character. That is, with reference to the position information data stored in the memory 34, the waveform of the portion corresponding to the Nth magnetic data is cut out from the magnetic data.

  In step S302, the magnetic recognition processing unit 13 detects the peak position and creates peak position data for the Nth character magnetic data.

  When the E13B font, which is the font of magnetic ink characters in this embodiment, is divided into eight equal parts with respect to the time axis, the peak of the read signal obtained when the character is read with the magnetic head is Each block is configured to exist in a different block. Therefore, each character is divided into eight equal parts, and the peak position is detected by examining where the peak is in the magnetic signal. Further, the amplitude level for each peak position is examined, and peak position data including the time and amplitude level of the peak position is created.

  As a peak detection method, for example, a method in which a point at which the amplitude level of the sampling waveform has changed from an increase to a decrease is detected as a positive peak, and a point at which the amplitude level of the sampling waveform has changed from a decrease to an increase is detected as a negative peak. is there. The peak detection method may be other modes.

  For example, for a digital signal as shown in FIG. 9A, a signal representing a peak position by the above method is shown in FIG. 9B, which indicates that the number of peaks is six. When the peak position data is illustrated, it is as shown in FIG.

  In step S303, the magnetic recognition processing unit 13 creates a relationship between the peak position time, amplitude level, and polarity as a table based on the peak position data obtained in step S302. This table is called a sample table. FIG. 10 shows an example of the sample table.

  In FIG. 10, the sample table is created on the assumption that the number of peaks is 4, but this is not the case. In the present embodiment, a case where four peaks are detected from the magnetic data of the Nth character will be described for easy understanding.

  In step S304, the magnetic recognition processing unit 13 compares each data table in the magnetic data dictionary table 12 with the sample table created in step S303.

  The magnetic data dictionary table 12 stores in advance individual dictionary tables indicating the peak position time, amplitude level, and polarity for each character. FIG. 11 shows a dictionary table of one character as an example, but a dictionary table is created for all MICR characters (14 types).

  In the example of the dictionary table of FIG. 11, the dictionary table is a character having four peaks, but the number of peaks is not limited to this.

  In step S304, the magnetic recognition processing unit 13 compares the sample table created in step S303 with all dictionary tables included in the magnetic data dictionary table 12. That is, the two-dimensional distance between each peak in the sample table and each peak in all dictionary tables is calculated.

  FIG. 12 is a diagram in which the waveform data of the dictionary table and the waveform data of the sample table are displayed in an overlapping manner. The vertical axis represents amplitude, and the horizontal axis represents time. The waveform represented by the solid line is the waveform data represented by the sample table, while the waveform represented by the dotted line is the waveform data represented by the dictionary table to be compared.

  In the following description, a case where the sample table shown in FIG. 10 is compared with the dictionary table shown in FIG. 11 will be described. Further, the polarities included in the tables of FIGS. 10 and 11 are not used in this embodiment.

  The magnetic recognition processing unit 13 calculates the two-dimensional distances L0, L1, L2, and L3 of the four terms in the dictionary table with respect to the first term (St_0, Sl_0) of the sample table. The distance L0 is the distance from the first peak S-peak0 of the waveform represented by the sample table to the first peak D-peak0 of the waveform represented by the dictionary table as shown in FIG. 12, and the distance L1 is the first peak of the waveform represented by the sample table. The distance from the first peak D-peak1 of the waveform represented by the sample table to the third peak D-peak2 of the waveform represented by the dictionary table from the distance S2 from the first peak D-peak1 of the waveform represented by the sample table. The distance L3 is a distance from the first peak S-peak0 of the waveform represented by the sample table to the fourth peak D-peak3 of the waveform represented by the dictionary table. These distances L0, L1, L2, and L3 are obtained by the following calculation.

L0 = √ {A × (St — 0−Dt — 0) ² + B × (Sl — 0—Dl — 0) ² }
L1 = √ {A × (St — 0−Dt — 1) ² + B × (Sl — 0—Dl — 1) ² }
L2 = √ {A × (St — 0−Dt — ² ) ² + B × (Sl — 0—Dl — ² ) ² }
L3 = √ {A × (St — 0−Dt — 3) ² + B × (Sl — 0—Dl — 3) ² }
Here, A and B are predetermined constants.

  Next, based on the obtained distance, the magnetic recognition processing unit 13 determines that the first term of the sample table is the four terms in the dictionary table of characters to be compared in the magnetic data dictionary table 12. Which is closest is determined by brute force, and the distance to the nearest term is stored. The term determined to be the closest in the dictionary table of the characters to be compared is subsequently omitted. Subsequently, the distance between the second term (St_1, Sl_1) of the sample table and the remaining three terms is calculated, the distance between the closest terms is stored, and the closest term is stored. Is omitted from the next comparison. Next, the distance between the remaining two terms is calculated for the third term (St_2, Sl_2), the distance between the closest terms is stored, and the closest term is Omit from comparison. Finally, the distance between the fourth term (St_3, Sl_3) and the remaining terms is calculated and stored. Thus, the two-dimensional distance from the dictionary table term of the character to be compared in the dictionary table 12 is calculated, and the closest peak is determined. When calculating the two-dimensional distance by adjusting the unit of time axis and the unit of amplitude axis so that the aspect ratio of the waveform for one character on the two-dimensional drawing does not become extremely long or wide. Is considered. Alternatively, the constants A and B are adjusted to balance weighting when calculating time and amplitude.

  As described above, when the determination is made up to the end, that is, the fourth term (St_3, S1_3), the distances determined to be closest to each other are added, and the sum value is stored. Similarly, the sum of distances from the sample table shown in FIG. 10 to the next character in the dictionary table 12 and the term determined to be closest to the next character is calculated.

  In step S305, recognition candidate characters are determined from the sum of the distances calculated in step S304.

  Specifically, the character in the dictionary table 12 in which the sum of the distances calculated in step S304 is minimized is set as the first candidate character, and the second candidate, the third candidate ( Multiple candidates). In this embodiment, the third to the third smallest in the sum are used as recognition candidate characters.

  Further, the sum calculated when compared with the dictionary table of the character is synonymous with the certainty (score) of the MICR character reading result. If the sum is small, the match rate with the dictionary data is high, so the recognition result is highly reliable. Conversely, if the sum is large, the match rate with the dictionary data is low and the recognition result is judged to be low in reliability. be able to. In the present embodiment, the identification information of the first to third recognition candidate characters determined in step S 305 and the respective certainty factors are paired and stored in the memory 34. Note that the identification information of the first to third recognition candidate characters is used as an example, and the number of recognition candidate characters may be more or less than three. When the series of processing in step S305 is completed, the magnetic reading unit 10 ends the series of MICR data recognition processing, proceeds to step S106, and continues the processing.

  In step S106, it is determined whether or not character recognition by OCR is further required for the MICR recognition result obtained in step S105.

  FIG. 13 is a flowchart showing a series of steps in step S106 for determining whether or not character recognition by OCR is necessary.

  In step S401, the MICR recognition score of the first candidate character obtained in step S105 is compared with a predetermined threshold G. If the score of the first candidate character is equal to or smaller than the predetermined threshold G (step S401: Yes), the process proceeds to step S402, and it is determined that the first candidate is sufficiently reliable and that further character recognition by OCR is unnecessary. Then, the first candidate character is set as a final recognition result, and the process proceeds to step S109. If the score of the first candidate character is greater than the predetermined threshold G (step S401: No), the process proceeds to step S403.

  In step S403, it is determined whether the MICR recognition score of the first candidate character is equal to or greater than a predetermined threshold B (a value greater than the threshold G). If the MICR recognition score of the first candidate character is equal to or greater than the predetermined threshold B (step S403: Yes), the process proceeds to step S404, the first candidate is determined to be too unreliable, and is treated as an unreadable character. Proceed to S109. If the MICR recognition score of the first candidate character is smaller than the predetermined threshold B (step S403: No), it is further determined that character recognition by OCR is necessary, and the process proceeds to step S107.

  In step S107, dictionary data used in OCR recognition is selected.

  Similar to the magnetic data dictionary table 12, the optical data dictionary table 23 has dictionary data for all E13B font characters, that is, 14 characters. An example of dictionary data included in the optical data dictionary table 23 is shown in FIG. As shown in FIG. 14, each dictionary data that the optical data dictionary table 23 has is a template of binary image data of each character. Hereinafter, the dictionary data of each character included in the optical dictionary table is referred to as a template. FIG. 14A is an example of the dictionary for each character of the E13B font shown in FIG. 3, and FIG. 14B is an enlarged view of the template for the character “1”. As shown in the figure, the height direction of the character is the x direction, and the direction orthogonal to the x direction is the y direction.

  In step S107, only the first to third candidate character templates recognized in step S305 are selected as templates to be used in the subsequent step S108. For example, if it is determined in step S305 that the first candidate character is “2”, the second candidate character is “5”, and the third candidate character is “8”, the template selected in step S107 is “ 2 ”,“ 5 ”, and“ 8 ”.

  In step S108, an OCR recognition process for the Nth character is performed using the template selected in step S107.

  FIG. 15 is a flowchart showing a flow of a series of processes of the OCR recognition process.

  In step S501, the image data at the position of the Nth character is cut out from the image data obtained in step S102. Since the position information of each character stored in the memory 34 is position information regarding magnetic data, it is necessary to convert this into image data (number of pixels). Specifically, the position information regarding the magnetic data is converted into the position information in units of pixels using the following formula.

(SPC) × (CPS) × (DPI) ÷ 2.54
SPC: Number of magnetic data samples per character CPS: Centimeters per sample DPI: Image resolution (dots / inch)
The number of dots in the horizontal direction per character is calculated by the above formula, and the image data of the Nth character is cut out.

  In step S502, the binary image data cut out in step S501 is compared with the template selected in step S107. Specifically, the comparison area of 24 dots in the x direction and 19 dots in the y direction in the recognition target range is compared with each template selected in step S107, and among the 456 dots, there is a mismatched dot. Count the number and keep it as a discrepancy rate.

  In step S503, a recognition result character is determined from the result of the mismatch rate with each template counted in step S502.

  First, the template having the lowest mismatch rate among the mismatch rates with each template is determined. Next, if the mismatch rate of the template having the lowest mismatch rate is equal to or smaller than a predetermined constant P, the character of the template is taken as the recognition result, and the process proceeds to step S109. On the contrary, when the mismatch rate of the template having the lowest mismatch rate is higher than a predetermined constant P, it is determined that the Nth character cannot be read, and the process proceeds to step S109.

  In step S109, it is determined whether the recognition process has been completed up to the last character. If the recognition process is not completed up to the last character (step S109: No), the process proceeds to step S110, N is incremented by 1, and the process returns to step S105 to continue the process. When the recognition process has been completed up to the last character (step S109: Yes), the magnetic ink character reading device 1 transmits a character recognition result to the host device 40, and ends a series of processes.

  In the present embodiment, the polarity data of the dictionary data table is not used. However, for example, when the amplitude is from +127 to -128, the two-dimensional distance can be calculated using the polarity. If the third candidate character has a too bad score, the dictionary information of the third candidate character may not be used for OCR. Further, if the second candidate character has a too bad score, the dictionary information of the second candidate character and the third candidate character may not be used for OCR. Alternatively, only the dictionary information of the first candidate character may be used for OCR at any time, and it may be determined that recognition is not possible if the OCR recognition result is bad. In addition, the image data template provided in the optical data dictionary table is a binary image. However, if the image data acquired from the optical reading unit is a multi-value image, the template can be applied to the multi-value image. It is desirable to keep it. Further, the OCR method is not limited to the method disclosed in the present embodiment. As described above, in the present embodiment, the configuration is such that the dictionary information of the characters to be compared with the image data is changed or limited according to the result of the magnetic recognition processing in the optical recognition processing in the optical reading unit. It has become. By utilizing the information of the previous magnetic recognition process during the optical recognition process, the magnetic ink character recognition result can be improved and the processing time can be increased.

  More specifically, there are the following effects.

  FIG. 17 is an example of waveform data when magnetic information is read by a magnetic reader for magnetic ink characters “0” and “8”. FIG. 17A shows the waveform data of the magnetic ink character “0”, and FIG. 17B shows the waveform data of the magnetic ink character “8”. The vertical direction represents the signal intensity of magnetic information, and the horizontal direction represents the passage of time. As shown in FIG. 17, the waveform data of the magnetic ink characters “0” and “8” are very different in their shapes. Therefore, it is considered that the magnetic ink character reading unit hardly recognizes “0” as “8”.

  Therefore, in this embodiment, even if there is a possibility of misreading in the optical character recognition processing as shown in FIG. 16, or the magnetic character recognition result in the magnetic ink character reading unit is insufficient. In the optical character recognition process, the result of the magnetic character recognition process is utilized to secure an excellent recognition rate. By doing so, the magnetic character recognition step and the optical character recognition processing step compensate for each other's weaknesses, and it can be expected that the magnetic ink character recognition result is improved and the processing time is increased.

It is a figure which shows the list of the character of E13B font. It is a perspective view which shows the external appearance of the magnetic ink character reader of one Embodiment. It is a figure which shows an example of a check. It is a block diagram which shows the internal structure of the magnetic ink character reader of one Embodiment. It is a flowchart which shows the flow of a series of processes of character recognition in a magnetic ink character reader. It is a figure which shows the waveform at the time of reading magnetic information with a magnetic head about "0" and "1" which are a part of E13B font. It is a flowchart which shows a series of flows of a character cut-out process. It is a flowchart which shows a series of flows of a MICR recognition process. It is a figure which shows the example of a digital signal, a peak position signal, and peak value data. It is a figure which shows an example of a sample table. It is a figure which shows an example of a dictionary table. It is the figure which displayed the waveform data of the dictionary table, and the waveform data of the sample table in an overlapping manner. It is a flowchart which shows a series of flows which judge whether the re-recognition by OCR is required. It is a figure which shows an example of the template of each character of E13B font. It is a flowchart which shows the flow of a series of processes of OCR recognition. It is an enlarged view around the magnetic ink character of an actual check. It is a figure which shows the example of the waveform data of a magnetic ink character.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Magnetic ink character reader 2 Hopper part 3 Conveyance path 4 Outlet 6 Magnetic head 7 Image reading sensor 10 Magnetic reading part 11 Magnetic data processing part 12 Magnetic data dictionary table 13 Magnetic recognition process part 15 Character cutout part 20 Optical reading part 21 Optical data processing unit 22 Optical recognition processing unit 23 Optical data dictionary table 31 Main control unit 32 External interface 33 Address bus 34 Memory 40 Host device 90 Check 91 Magnetic ink character part 100 Magnetic ink character “0”

Claims (3)

Magnetic data acquisition means for acquiring magnetic data contained in magnetic ink characters printed on a document;
Magnetic character recognition means for recognizing the magnetic ink characters from the magnetic data;
Optical character reading means for optically reading the magnetic ink characters and acquiring image data;
Optical character recognition means for recognizing the magnetic ink characters from the image data;
With
The magnetic ink character reading device, wherein the optical character recognition unit recognizes the magnetic ink character based on a recognition candidate character that is a recognition result of the magnetic character recognition unit.   The magnetic ink character reader according to claim 1, wherein the magnetic character recognition unit determines a plurality of recognition candidate characters. A magnetic data acquisition step of acquiring the magnetic data using magnetic data acquisition means for acquiring magnetic data included in the magnetic ink characters printed on the original;
A magnetic character recognition step for recognizing the magnetic ink characters from the magnetic data;
An image data acquisition step of acquiring image data using an optical character reader that reads magnetic ink characters printed on a document;
An optical character recognition step for recognizing the magnetic ink characters from the image data;
With
The method of controlling a magnetic ink character reader, wherein the optical character recognition step recognizes the magnetic ink character based on a recognition candidate character that is a recognition result of the magnetic character recognition step.

Images