JP2011067796A - Sectioning apparatus, method for recognizing sectioning information, and program for recognizing sectioning information - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a sectioning apparatus, a method for recognizing the sectioning information, and a program for recognizing the sectioning information by which the sectioning information can be recognized at a high accuracy even if the bar code information is incorrect. <P>SOLUTION: A recognition part 14 reads the images of paper sheets by a scanner 21, recognizes the information which the bar code detected from the images of the paper sheets read by the scanner 21 by a bar code recognition part 22, extracts a correct candidate of the sectioning information corresponding to the recognition result of the bar code from a correct candidate table 26 in which correct candidates of the sectioning information relevant to the information described with bar codes on the paper sheets are stored, confirms whether the sectioning information of each extracted correct candidate is correct or not by letter recognition processing, and sections the paper sheets off based on the sectioning information of the correct candidate determined to be correct by the confirmation. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention recognizes classification information by means of bar codes and characters written on paper sheets, such as a mail sorting machine that sorts mail items with address information indicated by customer barcodes and letters. The present invention relates to a sorting apparatus, a sorting information recognition method, and a sorting program for performing sorting information recognition processing.

  Conventionally, in a sorting apparatus, a character recognition unit for recognizing characters indicating a sorting destination such as a postal code or address information written on a paper sheet, and a sorting information such as a sorting destination are set in advance. Some have a customer barcode recognition unit that reads barcode information (customer barcode). In such a sorting apparatus, paper sheets are sorted based on the recognition results by the character recognition unit and the customer barcode recognition unit.

  Conventionally, an attempt has been made to increase the recognition accuracy of the classification information by using the recognition result by the character recognition unit and the recognition result by the customer barcode recognition unit. For example, Japanese Patent Application Laid-Open No. 2002-52366 describes that a description position of an address is estimated based on a barcode detection position. However, what is disclosed in Japanese Patent Application Laid-Open No. 2002-52366 does not assume that the barcode information is incorrect, and it is considered that sufficient recognition accuracy may not be obtained.

  Japanese Patent Laid-Open No. 2000-15184 discloses that the classification information as the recognition result of the customer barcode is collated with the address database, and if the recognition result does not exist as the classification information in the address database, the recognition result by the character recognition unit. Is used to correct a customer barcode recognition result. However, although what is disclosed in Japanese Patent Laid-Open No. 2000-15184 assumes that the barcode information is incorrect, it is determined whether or not it is incorrect by collating with the address database. Even if the recognition result is present in the address database, it cannot be determined as a misrecognition.

JP 2002-52366 A JP 2000-15184 A

  An object of one aspect of the present invention is to provide a sorting apparatus, a sorting information recognition method, and a sorting information recognition program that can recognize sorting information with high accuracy.

  A sorting apparatus according to one aspect of the present invention sorts a paper sheet in which at least a part of the sorting information is described with a barcode and characters, and associates the information with the barcode described on the paper sheet. A table for storing the correct candidates for the classified information, a reading unit for reading the image of the paper sheet, and a barcode recognition unit for recognizing information indicated by the barcode detected from the image of the paper sheet read by the reading unit And character recognition means for recognizing characters detected from the image of the paper sheet read by the reading means, and correct answer candidates of the category information corresponding to the barcode recognition result by the barcode recognition means are extracted from the table And character recognition by the character recognition means to determine whether or not the classification information of each correct candidate extracted by the extraction means is correct Has a sure means, and dividing means for dividing the paper sheet based on the sorting information of the correct candidate determined is correct by said confirmation means.

  A classification information recognition method according to an aspect of the present invention is a classification information recognition method for paper sheets in which at least a part of the classification information is described by barcodes and characters, and reads a paper sheet image. The information indicated by the barcode detected from the read image of the paper sheet is recognized, and the correct answer candidate of the classification information corresponding to the barcode recognition result is converted into the information described in the barcode on the paper sheet. Extracted from the table storing the correct answer candidates of the associated category information, confirmed whether or not the extracted category information of each correct answer candidate is correct, and determined that it was correct by the confirmation The paper sheet is classified based on the classification information of correct answer candidates.

  A classification information recognition program according to one aspect of the present invention is a classification information recognition program for paper sheets in which at least a part of the classification information is described by barcodes and characters, and reads an image of the paper sheets. Function, function for recognizing information indicated by the barcode detected from the image of the read paper sheet, and correct answer candidate of the classification information corresponding to the recognition result of the barcode are described on the paper sheet as a barcode A function of extracting from the table storing the correct answer candidates of the category information associated with the information to be performed, a function of confirming whether or not the extracted category information of each correct answer candidate is a correct answer, The computer is caused to execute a function of classifying the paper sheet based on the classification information of the correct answer candidates determined to be correct by confirmation.

  According to one aspect of the present invention, it is possible to provide a sorting apparatus, a sorting information recognition method, and a sorting information recognition program that can recognize sorting information with high accuracy.

FIG. 1 is a diagram showing a schematic configuration of a sorting system as a sorting apparatus according to the present embodiment. FIG. 2 is a diagram illustrating an example of paper sheets to be processed by the sorting system. FIG. 3 is a diagram illustrating a configuration example of a correct candidate table. FIG. 4 is a flowchart for explaining classification information recognition processing by the first classification method. FIG. 5 is a flowchart for explaining a correct candidate table update process by the first classification method. FIG. 6 is a flowchart for explaining the update process of the correct candidate table based on the division information determined by the VCS. FIG. 7 shows a configuration example of a sorting system to which the third sorting method is applied. FIG. 8 is a flowchart for explaining the update process of the correct answer candidate table based on the information on the moving person.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a diagram showing a schematic configuration of a sorting system 1 as a sorting device according to the present embodiment.
The sorting system 1 is a paper sheet processing apparatus that sorts paper sheets based on sorting information written in characters on the paper sheets.
As shown in FIG. 1, the sorting system 1 includes a sorting machine main body (main body) 2 and a VCS (video coding system) 3. The sorter body 2 sorts paper sheets based on address information or the like. The VCS 3 inputs address information of paper sheets whose address information could not be discriminated by the sorter body 2 through an operator's coding operation.

As shown in FIG. 1, the sorting machine main body 2 is provided with an operation panel 10 on the front surface. The operation panel 10 includes a display device with a built-in touch panel. In the operation panel 10, an operator (operator) designates a processing mode and a processing start, and displays an operation state of the sorting machine.
The sorter body 2 includes a supply unit 11 that supplies paper sheets. The supply unit 11 supplies paper sheets to the conveyance path 12 one by one at a predetermined interval.

  In the paper stored in the supply unit 11, information indicating a destination (classification destination) such as a zip code, an address, and an address (hereinafter also simply referred to as classification information or destination information) is written in characters. In addition, some paper sheets stored in the supply unit 11 are printed with a barcode indicating classification information (destination information) called a customer barcode. Further, the paper sheet supplied to the supply unit 11 has a destination bar code indicating classification information of the paper sheet or an ID bar code indicating identification information (ID information) for identifying the paper sheet. Some may be printed with special ink (fluorescent ink) that is not directly visible to the human eye. The destination barcode or ID barcode is printed by this sorter or another sorter. Further, the ID barcode is given to a paper sheet for which character recognition of the classification information has not been performed by OCR processing described later.

  A BCR (bar code reader) 13, a recognition unit 14, a BCW (bar code writer) 15, and a sorting unit 16 are provided on the conveyance path 12. Each of these units operates based on the control by the sorting control unit 17 according to the processing status such as the conveyance state for each paper sheet.

  The BCR 13 reads a destination barcode or ID barcode printed with special ink (fluorescent ink) on a paper sheet conveyed on the conveyance path 12. The BCR 13 includes a reading unit that reads an image of a barcode and a recognition unit that recognizes a barcode in an image read by the reading unit. The barcode recognition result by the BCR 13 is supplied to the sorting control unit 17.

  The recognizing unit 14 recognizes paper sheet classification information. The recognition unit 14 includes a scanner 21, a barcode recognition unit 22, a character recognition unit 23, a determination unit 24, a dictionary database (DB) 25, and the like. The scanner 21 reads an image of a paper sheet on the conveyance path 12. The scanner 21 optically reads an image on a paper sheet. The scanner 21 is connected to a barcode recognition unit 22 and a character recognition unit 23. The image read by the scanner 21 is supplied to the barcode recognition unit 22 and the character recognition unit 23.

  The bar code recognition unit 22 recognizes the classification information given by the customer bar code. For example, the bar code recognition unit 22 searches for a customer bar code from an image of a paper sheet read by the scanner 21, and recognizes the classification information indicated by the customer bar code when the customer bar code is detected. . The character recognizing unit 23 recognizes the division information written in characters. For example, the character recognizing unit 23 refers to the dictionary data registered in the dictionary DB 25 and recognizes the division information indicated by the characters from the sheet image read by the scanner 21.

  The barcode recognition unit 22 and the character recognition unit 23 are connected to the determination unit 24. The recognition result by the barcode recognition unit 22 and the recognition result by the character recognition unit 23 are supplied to the determination unit 24. The determination unit 24 comprehensively determines the recognition result by the barcode recognition unit 22 and the recognition result by the character recognition unit 23, and determines the recognition result of the classification information for the paper sheet. The recognition result of the classification information by the determination unit 24 is supplied to the classification control unit 17. The determination unit 24 can be realized by, for example, an electronic computer (computer) having a CPU, various memories, and various interfaces having a similar hardware configuration. In this case, each process by the determination unit 24 can be realized by the CPU executing the program. The processing contents in the determination unit 24 will be described later in detail.

  The BCW 15 prints a destination barcode or ID barcode on a paper sheet with fluorescent ink. For example, the BCW 15 prints, with fluorescent ink, a destination barcode obtained by converting the classification information as a recognition result into a barcode when the recognition unit 14 has successfully recognized the classification information. Further, the BCW 15 prints, with fluorescent ink, an ID barcode obtained by converting the identification information (letter ID) for identifying the paper sheet into a barcode when the recognition unit 18 succeeds in recognizing the classification information. That is, the BCW 15 prints the destination barcode on the paper sheet for which the classification information is recognized, and prints the ID barcode on the paper sheet for which the classification information cannot be recognized. Note that the classification information (classification information input by keystroke) specified by the VCS 3 to be described later is stored in association with the letter ID. For this reason, regarding the paper sheets for which the classification information is specified by the VCS 3, it is possible to classify based on the classification information by determining the classification information corresponding to the ID barcode.

  On the downstream side of the BCW 15 in the paper sheet conveyance direction, a sorting unit 16 for sorting the paper sheets according to address information is provided. The partitioning section 16 includes a plurality of partition pockets (not shown) partitioned into a plurality of stages and a plurality of rows. Each pocket is set corresponding to each sorting destination, and sheets are sequentially accumulated in the pocket corresponding to the address information based on the address information or the machine code. Further, the sorting unit 16 is provided with a VCS exclusion pocket (not shown) in which paper sheets whose sorting destination cannot be recognized are accumulated. The paper sheets accumulated in the VCS exclusion pocket are re-supplied to the supply unit 11 after the address information is input in the VCS 3, and the ID code of the paper sheet and the address information input in the VCS 3 Based on the above, it will be reclassified.

Next, the VCS 3 will be described.
As shown in FIG. 1, the VCS 3 includes an image storage / distribution device 31, a plurality of video coding terminals (hereinafter referred to as VCD) 32, and the like.
The image storage / distribution device 31 distributes images of paper sheets whose address information could not be recognized by the sorter main body 2 to each VCD 32, and inputs the key input result of the address information by each VCD 32 to the sorter main body 2. Process to return. The image storage / distribution device 31 is connected to the sorter body 2. The image storage / distribution device 31 is composed of, for example, a personal computer (PC) having a CPU, various memories, and various interfaces. The image storage / distribution device 31 stores the identification information (letter ID) of the paper sheet in association with the image of the paper sheet whose address information could not be recognized by the recognition unit 14, and stored the stored paper sheet. The image is distributed to each VCD 32.

  The VCD 32 includes a display (display unit) or a keyboard (operation unit). The VCD 32 is configured by a PC, for example. The VCD 32 displays an image of the paper sheet distributed from the image storage / distribution device 31 on a display. In the state where the image of the paper sheet is displayed, in the VCD 32, the operator inputs the address information using the keyboard. The input result by the VCD 32 is returned to the image storage / distribution device 31 in association with the identification information (letter ID) of the paper sheet. The image storage / distribution device 31 supplies the input result from the VCD 32 to the sorter body 2 in association with the identification information (letter ID) of the paper sheet. As a result, the sorting machine main body 2 obtains sorting information corresponding to the ID barcode assigned to the paper sheet.

Next, the classification information recognized by the recognition unit 14 will be described.
FIG. 2 is a diagram illustrating an example of paper sheets to be processed by the sorting system 1.
In the example shown in FIG. 2, a barcode is printed on a paper sheet together with a character string indicating classification information. The bar code shown in FIG. 2 is called a customer bar code. The customer barcode is information in which information for identifying the classification destination (classification information) is converted into a barcode. In other words, the customer barcode is a barcode indicating classification information similar to the classification information written in characters.

  For example, the address information as the classification information is represented by a combination of “zip code”, “address”, “no.”, And “room number”. For this reason, what coded the number which shows "zip code", "address", "number", and "room number" can be used as a customer barcode. In the example shown in FIG. 2, the postal code is “123-4567”, the address is “112”, and the number is “5”. For this reason, the numeric string “1234567112-5” is barcoded as a customer barcode.

  Therefore, in the sorting system, if the customer barcode printed on the paper sheet is accurately recognized, the correct sorting information can be recognized and the sorting can be accurately performed. However, in reality, there are cases where the paper sheets cannot be accurately classified even if the customer barcode printed on the paper sheets is accurately recognized. As a case where the customer barcode does not indicate an accurate classification destination, the following cases can be considered.

(A) When the description of the destination is incorrect For example, although “1-12-5” is correct, the description may be “112-5” because “−” is omitted by mistake. . Moreover, although it is correctly “122-5”, there is a case where the notation is “112-5” because one number is wrong. If the classification information described in this way is incorrect, and the erroneous classification information (“112-5”) is also present in the database, the paper sheet is erroneously classified. In order to prevent such misclassification, it is necessary to read information such as an apartment name, a building name, a name, and a company name written in characters that are not expressed in the customer barcode for confirmation.

(B) When there are a plurality of apartments or houses at the same address There may be two apartments or houses at “112-5”. For example, “123-4567 ○○ City ○ Ward ○○○○ machi 112-5 ABC Mansion Room 203 ○○ Taro” and “123-4567 ○○ City ○ Ward ○○○○ Town 112-5 Room XYZ Mansion 203 Two of “XX Shibao” are different classification information. However, when these classification information is coded as a customer bar code, the same code “1234567112-5-203” is obtained. In order to distinguish between the two types of classification information, it is necessary to read information such as an apartment name, a building name, a name, or a company name written in characters that is not represented by a customer barcode.

(C) If you have moved or moved in If you have been notified that you have moved out, you will also need to classify the paper with the old address as the new address. In this case, the customer bar code alone cannot distinguish between the old resident and the current resident. In order to distinguish this, it is necessary to read the name or company name written in characters that are not expressed in the customer barcode.

  As described above, simply recognizing the customer bar code may cause misclassification. For this reason, in this sorting system, a sorting method is used that can reliably sort paper sheets that are misclassified only by the customer barcode. Hereinafter, an example of a sorting method applied to this sorting system will be described.

First, the first sorting method will be described.
In the first sorting method, the correct answer candidate for the recognition result of the customer barcode is stored in the correct answer candidate table 26 in the recognition unit 14 of the sorter body 2. When the recognition result of the customer barcode is obtained, the recognition unit 14 extracts the correct answer candidate corresponding to the recognition result from the correct answer candidate table 26. When the correct answer candidates are extracted from the correct answer candidate table 26, the recognition unit 14 confirms (verifies) each of the extracted correct answer candidates by the character recognition process for the segment information written in characters.

  The confirmation result for each correct answer candidate is obtained, for example, as a certainty factor for each correct answer candidate. In this case, it is possible to determine whether or not the correct recognition result is based on whether or not the maximum certainty factor is equal to or greater than a predetermined threshold. If there is a correct answer candidate that is confident that it is correct by such confirmation processing, the recognition unit 14 outputs the correct answer candidate as a recognition result of the classification information for the paper sheet. As a result, the recognition result of the customer barcode can be corrected by the information stored in the correct answer candidate table 26.

  Further, the correct answer candidate table 26 is updated as needed by the discriminating unit 24 functioning as a table updating unit. For example, when the correct classification information (recognition result) is obtained from the correct answer candidate corresponding to the recognition result of the customer barcode, the correct answer table 26 counts up the frequency of the correct answer to the correct answer candidate. The In addition, when the correct classification information (recognition result) is not obtained from the correct candidate corresponding to the recognition result of the customer barcode, the determination unit 24 performs the classification by the character recognition process on the entire classification information by the character recognition unit 23. Information (recognition result) is determined. The classification information obtained as a result is registered in the correct answer candidate table 26 in association with the recognition result of the customer barcode.

FIG. 3 is a diagram illustrating a configuration example of the correct answer candidate table 26.
As shown in FIG. 3, the correct answer candidate table 26 stores a plurality of correct answer candidates and the frequency of each correct answer candidate for one customer barcode recognition result (read result). In the example shown in FIG. 3, four correct answer candidates (“12345671-12-5”, “123456671-12-5”, “123456567112-5Z1”, “123456671-12-5”, "12334567112-5Z2") is stored. Furthermore, each correct answer candidate (“12345671-12-5”, “123456672122-5”, “123456567112-5Z1”, “123456567112-5Z2”) has a frequency (“21”, “ 12 ”,“ 398 ”,“ 431 ”) are stored.

  According to such a correct candidate table 26, correct candidates corresponding to the recognition result of the customer barcode can be easily extracted. In the first classification method, the correction result of the customer barcode is corrected using the correct candidate table 26 as described above, and the correct candidate table 26 is updated based on the recognition result of the classification information.

Next, the flow of processing as the first sorting method will be described.
FIG. 4 is a flowchart for explaining classification information recognition processing by the first classification method. In the example shown in FIG. 4, it is assumed that the processing is a classification information recognition process for a paper sheet to which a customer barcode is assigned.
That is, when the paper sheet to which the customer barcode is assigned is conveyed to an image reading position by the scanner 21 of the recognition unit 14, the scanner 21 reads an image of the paper sheet conveyed on the conveyance path 12 (step). S11). When the paper sheet image is obtained by the scanner 21, the barcode recognition unit 22 detects the customer barcode from the paper sheet image, and recognizes the classification information indicated by the detected customer barcode (step S12). The recognition result of the customer barcode is notified from the barcode recognition unit 22 to the determination unit 24.

  The determination unit 24 extracts correct answer candidates corresponding to the customer barcode recognition result from the correct answer candidate table 26 (step S13). In this correct candidate extraction process, all correct candidates corresponding to the recognition result of the customer barcode may be extracted, or a predetermined number or more of the correct candidates corresponding to the recognition result of the customer barcode may be extracted. Correct answer candidates may be extracted.

  When the correct answer candidates are extracted from the correct answer candidate table 26, the determination unit 24 performs a verification process for confirming whether each correct answer candidate is likely to be correct using the character recognition unit 23 (step S14). In the verification process, for example, the degree of certainty indicating correctness of each correct answer candidate is determined by the character recognition process of the character recognition unit 23. The certainty factor indicates the likelihood that the classification information described in characters in the paper sheet image is a correct candidate. For example, the certainty factor may be the similarity between the category information written in letters on the paper sheet and the correct answer candidate, or may be a weight obtained by weighting the similarity with the frequency of the correct answer candidate.

  When the certainty factor of each correct answer candidate is obtained, the determination unit 24 determines whether there is a correct answer candidate to be adopted as a recognition result based on the certainty factor of each correct answer candidate (step S15). For example, if the maximum certainty factor is equal to or greater than a predetermined threshold, the determination unit 24 employs the correct answer candidate of the certainty factor as the recognition result of the final classification information in the recognition unit 14 (YES in step S15). . In this case (step S15, YES), the determination unit 24 outputs the correct answer candidate of the maximum certainty factor that is equal to or greater than a predetermined threshold as the recognition result of the classification information in the recognition unit 14 (step S16). When the classification information as the final recognition result of the recognition unit 14 is output, the determination unit 24 performs the update process of the correct answer candidate table based on the recognition result (step S20).

  If the certainty factor that is maximized is less than the predetermined threshold, the determination unit 24 determines that there is no correct answer candidate that can be adopted as the recognition result of the category information (NO in step S15). In this case (NO in step S15), the determination unit 24 performs a classification information recognition process in character recognition for the entire image of the paper sheet by the character recognition unit 23 (step S17). This classification information recognition process is for character recognition of the entire classification information. The character recognition unit 23 gives a certainty factor to the recognition result of the classification information by character recognition and notifies the determination unit 24 of the certainty.

  When the classification information recognition result and the certainty factor are obtained from the character recognition unit 23, the determination unit 24 determines whether or not the certainty factor of the segment information obtained by the character recognition of the character recognition unit 23 is equal to or greater than a predetermined threshold value. Is determined (step S18). When it is determined by this determination that the certainty of the segment information obtained by character recognition is equal to or greater than a predetermined threshold (step S18, YES), the determination unit 24 uses the segment information obtained by the character recognition unit 23 as The recognition unit 14 outputs the final classification information recognition result (step S18). When the classification information as the final recognition result of the recognition unit 14 is output, the determination unit 24 updates the correct candidate table 26 based on the recognition result (step S20).

  If it is determined by the above determination that the certainty factor of the classification information obtained by character recognition is less than the predetermined threshold (NO in step S18), the determination unit 24 cannot recognize the classification information of the paper sheet. Judge that it is. In this case, the determination unit 24 outputs that the classification information could not be recognized as a final classification information recognition result in the recognition unit 14 (step S19).

Next, the update process of the correct answer candidate table 26 as step S20 will be described.
FIG. 5 is a flowchart for explaining the update process of the correct candidate table 26 in step S20.
When the classification information as the recognition result for the paper sheet in which the customer barcode is detected is determined, the determination unit 24 has a combination of the recognition result of the customer barcode and the classification information as the recognition result in the correct answer candidate table 26. Whether or not (step S21).

  When it is determined that the combination of the recognition result of the customer barcode and the classification information as the recognition result does not exist in the correct answer candidate table 26 (NO in step S21), the determination unit 24 sets the confirmed classification information to the customer barcode. The correct candidate corresponding to the recognition result is additionally registered in the correct candidate table 26 (step S22). When the classification information as the recognition result is additionally registered as a correct candidate in the correct answer candidate table 26, the determination unit 24 sets the frequency for the correct answer candidate to “1” (step S23).

  If it is determined that a combination of the recognition result of the customer barcode and the classification information as the recognition result exists in the correct answer candidate table 26 (step S21, NO), the determination unit 24 responds to the recognition result of the customer barcode. “+1” (increment) is set to the frequency of the correct answer candidate that has become the classification information as the recognition result (step S23).

As described above, in the first classification method, the correct answer candidates of the classification information corresponding to the customer barcode recognition result are stored in the table, and the correct answer candidates of the classification information corresponding to the customer barcode recognition result are stored in the table. The correct classification information is specified by checking whether or not the extracted correct answer candidate is correct by character recognition. Furthermore, frequency data for each correct answer candidate is counted up each time it is determined as correct classification information for the recognition result of the customer barcode.
According to such a first classification method, it is possible to easily determine correct classification information even if an incorrect customer barcode is given or the classification information cannot be expressed by the customer barcode alone. Can do. Furthermore, according to the first sorting method, the table is updated based on the sorting information determined by using character recognition processing or the like. Therefore, the accuracy of the table and the recognition of the sorting information are increased as the sorting system 1 is operated. The accuracy can be increased.

Next, the second sorting method will be described.
In this second classification method, the correct candidate table 26 is updated using the classification information input by the operator in the VCS (video coding system) 3 as the correct classification information.
The second sorting method is realized by a sorting system 1 having a VCS 3 as shown in FIG. In the sorting system 1 of the second sorting method, when the recognizing unit 14 of the sorting machine main body 2 cannot identify the reliable sorting information (when the sorting information having the certainty level equal to or higher than the predetermined threshold value cannot be identified), the VCS 3 When the correct answer candidate table 26 is updated based on the classification information determined in step S4, and the recognition unit 14 of the sorting machine body 2 can identify the certain classification information (when the classification information with the certainty level equal to or higher than the predetermined threshold value can be identified) ), The correct answer candidate table 26 is updated by the same processing as in the first classification method.

  That is, the image of the paper sheet in which the reliable classification information among the paper sheet images read by the scanner 21 could not be recognized is the VCS distribution information associated with the identification information (letter ID) of the paper sheet. It is transmitted from the sorter body 2 to the VCS 3. At this time, the customer barcode recognition result may be added to the VCS distribution information as supplementary information. The recognition result of the customer barcode may be stored on the sorter body 2 side (for example, a memory (not shown) in the recognition unit 14) together with the identification information (letter ID) of the paper sheet.

  In the VCS 3 to which the image of the paper sheet is supplied from the sorting machine main body 2, the operator's sorting information is input based on the image of the paper sheet. The sorting information input by the VCS 3 is supplied to the sorting machine body 2 as VCS input information associated with the paper sheet identification information (letter ID). The sorting machine body 2 re-sorts the paper sheets based on the sorting information obtained from the VCS 3, and associates the recognition result of the customer barcode of the paper sheets with the sorting information from the VCS 3, so that the correct answer candidate can be obtained. Processing to update the table 26 is performed.

Next, the flow of processing as the second sorting method will be described.
First, in the second sorting method, the recognizing unit 14 of the sorting machine body 2 performs the processing for recognizing the sorting information of the paper sheet by the flow process as shown in FIG. 4 as in the first sorting method described above. Do. In particular, in the second classification method, when the classification information (correct answer candidate) with sufficient certainty is obtained (step S15, YES), the determination unit 24 updates the correct answer candidate table 26. This is the same as the sorting method.

  However, in the second classification method, when the classification information with sufficient certainty cannot be obtained (step S18, NO), the determination unit 24 cannot determine the classification information as the recognition result as the process of step S19. In addition, VCS distribution information (paper sheet image, letter ID and customer barcode recognition results) for allowing the operator to input classification information is supplied to the VCS 3. As a second classification method, when classification information with sufficient certainty is obtained from the correct answer candidates according to the recognition result of the customer barcode (step S15, NO), destination recognition processing by character recognition (step S17). And the above-mentioned VCS distribution information may be generated and supplied to the VCS 3 without performing S18).

  In contrast to such processing in the sorting machine main body 2, the VCS 3 performs sorting information input processing by the operator. That is, in the VCS 3, the VCS distribution information from the sorter body 2 is stored in the image storage / distribution device 31. The image storage / distribution device 31 distributes the VCS distribution information to each VCD 32. The VCD 32 displays an image of the paper sheet included in the VCS distribution information supplied from the image storage / distribution device 31 on the display device, and accepts an input operation of classification information from the operator. In this case, the display device may display not only paper sheet images but also information such as customer barcode recognition results.

  When the classification information is input by the operator, the VCD 32 returns the input classification information to the image storage / distribution device 31 in association with the letter ID. The image storage / distribution device 31 transmits to the sorter body 2 VCS input information in which the sorting information input by the VCD 32 is associated with information such as the letter ID of the paper sheet. Thereby, in the sorting machine main body 2, the sorting information determined by the VCS 3 can be specified by the letter ID of the paper sheet. As a second sorting method, the recognition unit 14 of the sorter main body 2 that has acquired the VCS input information from the VCS 3 also updates the correct candidate table 26 with the sorting information determined by the VCS 3.

Next, update processing of the correct answer candidate table 26 based on the classification information determined by the VCS 3 as the second classification method will be described.
FIG. 6 is a flowchart for explaining the update process of the correct answer candidate table 26 based on the classification information determined by the VCS 3.
When the VCS input information is acquired from the VCS 3, the recognition unit 14 of the sorter main body 2 identifies the combination of the customer barcode recognition result on the sheet and the classification information input by the VCS 3. For example, the recognition result of the customer barcode on each paper sheet may be stored in a memory (not shown) in association with the identification information (letter ID) of the paper sheet, and the VCS distribution information is sent to the VCS 3. It may be supplied and returned from the VCS 3 together with the VCS input information.

  When the combination of the recognition result of the customer barcode and the classification information input by the VCS 3 is confirmed, the determination unit 24 of the recognition unit 14 determines that the classification information input by the VCS 3 is the recognition result of the customer barcode of the sheet. It is determined whether or not the correct answer candidate exists in the correct answer candidate table 26 (step S31).

  When it is determined that the classification information input in the VCS 3 by the above determination does not exist in the correct candidate table 26 as a correct candidate for the recognition result of the customer barcode of the paper sheet (step S31, NO), the determination unit 24 The category information input in step S32 is additionally registered in the correct answer candidate table 26 as correct answer candidates corresponding to the recognition result of the customer barcode (step S32). When the classification information input by the VCS 3 is additionally registered as a correct answer candidate in the correct answer candidate table 26, the determination unit 24 sets the frequency for the correct answer candidate to “1” (step S33).

  If it is determined that the classification information input by the VCS 3 exists in the correct candidate table 26 as a correct candidate for the recognition result of the customer barcode of the paper sheet (NO in step S31), the determination unit 24 determines that the correct answer is correct. The candidate frequency is incremented by “+1” (step S33).

As described above, in the second classification method, the correct answer candidates of the classification information corresponding to the customer barcode recognition result are stored in the table, and the correct answer candidates of the classification information corresponding to the customer barcode recognition result are stored in the table. The correct classification information is specified by checking whether or not the extracted correct answer candidate is correct by character recognition. If reliable classification information is not obtained by such processing, processing for determining the classification information of the paper sheet is performed by the VCS 3, and the classification information determined by the VCS 3 is used as the recognition result of the customer barcode of each paper sheet. The table is updated as the associated correct answer candidate.
According to the second classification method, even if an incorrect customer barcode is given or the classification information cannot be expressed only by the customer barcode, the correct answer candidate stored in the table is used. Correct classification information can be easily determined. Furthermore, according to the second sorting method, the table is updated based on the sorting information determined by the VCS, so that the accuracy of the table and the recognition accuracy of the sorting information are increased as the sorting system 1 is operated. Is possible.

Next, the third sorting method will be described.
In the third classification method, the recognition result of the customer barcode is corrected using a database indicating that some information included in the classification information (for example, an address due to a move or the like) has been changed. Here, it is assumed that the customer bar code recognition result is corrected by the moving person information database indicating the destination (classification information) of the person whose address has been changed due to the moving.

FIG. 7 shows a configuration example of a sorting system to which the third sorting method is applied.
The sorting system shown in FIG. 7 has a configuration in which a moving person information database 41 is added to the sorting system shown in FIG. Since the configuration other than the relocation information database 41 is the same as the configuration of each part of the sorting system shown in FIG. 1, the same reference numerals are given to the same portions, and detailed description thereof is omitted.

  The moving person information database 41 is updated at any time according to an offer from the moving person. The moving person information database 41 stores the address of the new address and the address of the new address in association with the name of the new person (name, company name, etc.). Furthermore, you may be made to store the name of the present resident in the address of the mover in the mover information database 41 in association with the name of the mover.

  In the update process based on the moving person information registered in the moving person information database 41, the correct answer candidate table 26 includes one of the correct answer candidates for the recognition result of the customer barcode indicating the address of the moving person. As a result, it becomes possible to additionally register classification information including the address of the new address. If the classification information recognition process by the first classification method is performed using such a correct answer candidate table 26, if the address of the new address is obtained as the recognition result of the customer barcode, the address of the new address is selected as the correct answer candidate. Including classification information is obtained.

  In such a case, the determination unit 24 can determine whether or not the classification information including the address of the new address is correct based on a destination name (such as a name or company name) obtained by character recognition. For example, when it is determined that the name of the destination (name or company name) is a moving person, the determination unit 24 can determine that the classification information including the address of the moving address is correct. That is, in the sorting system that recognizes the sorting information of paper sheets, it is considered that the recognition accuracy of the sorting information is improved by updating the correct answer candidate table 26 based on the moving person information.

Next, the flow of processing as the third sorting method will be described.
First, in the third sorting method, the recognition unit 14 of the sorting machine main body 2 performs the processing for recognizing the sorting information of the paper sheet by the flow process as shown in FIG. 4 in the same way as the first sorting method described above. Do. In the third sorting method, the correct candidate table 26 may be updated similarly to the second sorting method. In such a classification system to which the first and second classification methods are applied, the third classification method further checks the moving person information database 41 to appropriately store information on the moving person in the correct answer candidate table 26. The process to reflect is performed.

FIG. 8 is a flowchart for explaining the update process of the correct answer candidate table 26 based on the information on the moving person.
The update process of the correct answer candidate table 26 based on the information on the moved person as shown in FIG. 8 is performed when the moving person information database 41 is written. However, the process shown in FIG. 8 can be executed at an arbitrary timing. For example, the update process of the correct answer candidate table 26 shown in FIG. 8 may be performed periodically or in response to an instruction from the operator.

  That is, in the recognition unit 14 of the sorting machine body 2, the determination unit 24 checks whether there is new mover information in the mover information database 41 (step S41). For example, the new mover information is detected by searching the mover information database 41 for the mover information registered between the time when the correct candidate table is updated based on the previous mover information and the present time. For example, the new mover information is stored in the mover information database 41 with the registration date and time of the mover information, and whether or not the registration date and time is after the update date and time of the correct answer candidate table based on the previous mover information. Determined.

  When it is determined by the above determination that there is new mover information (step S41, YES), the determination unit 24 reads the new mover information from the mover information database 41, and correct answer candidate table based on the read new mover information. 26 is updated (step S42). In the update process of the correct answer candidate table 26 based on the moving person information, the determination unit 24 registers the address information of the new address as a correct answer candidate for the customer barcode indicating the address information of the moving address. That is, when the customer barcode recognition result is the address of the new address, the correct answer candidate table 26 is updated so that the address information of the new address can be obtained as a correct answer candidate.

  Here, it is assumed that the customer barcode reflects only the address information and does not reflect the destination name (name or company name, etc.). In such a case, it is assumed that there is a high possibility that the customer barcode indicating the address of the new address is printed on the paper sheets addressed to the new addressee. That is, it is preferable to register classification information including the address of the new address as a correct candidate for the customer barcode indicating the address of the new address. Therefore, when new mover information is extracted from the mover information database 41, the determination unit 24 sets the category information including the address of the new address corresponding to the customer barcode indicating the address of the new address as the correct answer candidate table 26. Register additional. However, in the update process of the correct answer candidate table 26 based on the moving person information, the frequency for the correct answer candidate is set as the initial value “0”.

  As described above, in the third classification method, the classification information including the address of the new address is additionally registered in the correct answer candidate table as the correct answer candidate for the customer barcode indicating the address of the new employee. In the correct answer candidate table 26 in which such information is registered, even if the customer barcode recognition result is the address of the new address, classification information including the address of the new address as the correct answer candidate for the customer barcode recognition result is obtained. It is done. As a result, the classification system can easily determine the classification information including the address of the new address as correct classification information, even for paper sheets printed with a customer barcode indicating the address of the new address. The classification information can be recognized with high accuracy.

  DESCRIPTION OF SYMBOLS 1 ... Sorting system, 2 ... Sorting machine main body, 3 ... VCS, 10 ... Operation panel, 11 ... Supply part, 12 ... Conveyance path, 13 ... BCR, 14 ... Recognition part, 15 ... BCW, 16 ... Sorting part, 17 ... Classification control unit, 18 ... recognition unit, 21 ... scanner, 22 ... barcode recognition unit, 23 ... character recognition unit, 24 ... discrimination unit, 25 ... dictionary database (DB), 26 ... correct candidate table, 31 ... image accumulation / distribution Apparatus 32 ... Video coding terminal (VCD), 41 ... Relocation information database.

Claims (7)

In a sorting apparatus that sorts paper sheets in which at least a part of the sorting information is described in barcodes and characters,
A table for storing the correct answer candidates of the category information associated with the information described in the barcode on the paper sheet;
Reading means for reading an image of a paper sheet;
Barcode recognition means for recognizing information indicated by the barcode detected from the image of the paper sheet read by the reading means;
Character recognition means for recognizing characters detected from the image of the paper sheet read by the reading means;
Extraction means for extracting the correct candidate for the category information corresponding to the barcode recognition result by the barcode recognition means;
Confirmation means for determining whether or not the category information of each correct answer extracted by the extraction means is correct by character recognition by the character recognition means;
Sorting means for sorting paper sheets based on sorting information of correct answer candidates determined to be correct by the checking means;
A sorting apparatus comprising: The table stores information indicating the frequency determined to be correct for each correct answer candidate,
Furthermore, the update means for updating the frequency corresponding to the correct answer candidate determined to be correct by the confirmation means,
The sorting apparatus according to claim 1, wherein: Further, when there is no classification information determined to be correct in each correct answer candidate extracted by the extraction means, processing means for recognizing the whole classification information described in characters by the character recognition means,
An update unit that updates the table based on a combination of the barcode recognition result by the barcode recognition unit and the classification information obtained by the processing unit, when the classification unit is obtained by the processing unit; Having
The sorting apparatus according to claim 1, wherein the sorting apparatus is characterized in that: Furthermore, when there is no classification information determined to be correct for each correct answer candidate extracted by the extraction means, coding means for inputting the classification information of the paper sheet,
Updating means for updating the table based on a combination of the barcode recognition result by the barcode recognition means and the classification information input by the coding means;
The sorting apparatus according to any one of claims 1 to 3, wherein the sorting apparatus is characterized in that: In addition, a database for storing information on the mover whose address information included in at least a part of the classification information has been changed,
Updating means for adding to the table, as correct answer candidates, segment information including address information of the new address based on the newly-registered mover information in the database;
The sorting apparatus according to any one of claims 1 to 4, wherein the sorting apparatus is characterized in that: A method for recognizing classification information for paper sheets in which at least a part of the classification information is described in barcodes and characters,
Scan paper sheets,
Recognize the information indicated by the barcode detected from the read image of the paper,
The correct candidate for the category information corresponding to the barcode recognition result is extracted from the table storing the correct candidate for the category information associated with the information described in the barcode on the paper sheet,
It is confirmed by character recognition processing whether or not the extracted category information of each correct answer candidate is correct,
Classifying the paper sheet based on the classification information of the correct answer candidates determined to be correct by the confirmation;
A classification information recognition method comprising: A classification information recognition program for paper sheets in which at least part of the classification information is described in barcodes and characters,
The ability to read paper images,
A function of recognizing information indicated by a barcode detected from the read image of the paper sheet;
A function of extracting the correct candidate for the category information corresponding to the barcode recognition result from a table storing the correct candidate for the category information associated with the information described in the barcode on the paper sheet;
A function of confirming whether or not the extracted category information of each correct answer candidate is a correct answer by a character recognition process;
A function of classifying the paper sheet based on the classification information of the correct answer candidates determined to be correct by the confirmation;
A classification information recognition program characterized by causing a computer to execute.

Images