JP5249353B2 - Paper sheet identification device and paper sheet identification method - Google Patents

Description

  The present invention relates to a paper sheet identification apparatus and a paper sheet identification method for identifying a paper sheet by reading the conveyed paper sheet with an optical line sensor oriented in a direction orthogonal to the conveyance direction, and in particular, per unit time. The present invention relates to a paper sheet identification apparatus and a paper sheet identification method capable of improving the number of processed sheets and reducing the size of the apparatus.

  Papers such as banknotes, securities such as checks, forms such as payment slips are transported using a transport mechanism, and papers are identified using an optical line sensor that receives and emits light such as visible light and infrared light. A paper sheet identification device is known.

  When performing a paper sheet identification process using such an optical line sensor, it is common to perform a type determination process using image data for the entire paper sheet after scanning of the entire paper sheet is completed. .

  For this reason, when the paper sheets are continuously identified, the identification process for the next paper sheet cannot be started until the type determination process is completed. Therefore, since a conveyance path having a length that is the sum of the sheet length and the conveyance distance during the type determination process is required, a longer conveyance path for conveying the sheets is secured or is conveyed continuously. It was necessary to take measures such as increasing the interval between paper sheets.

  However, if the above-mentioned measures are taken, problems such as enlargement of the apparatus size and sluggish increase in the number of processed sheets per unit time (processing speed) occur. For this reason, an attempt has been made to shorten the time required for the type determination processing performed after the completion of scanning of the entire paper sheet.

  For example, in Patent Document 1, the optical line sensor performs a data acquisition process for paper sheets in parallel with a blocking process for blocking the pixels of acquired data, and after the scan for the entire paper sheet is completed, the blocking process is performed. A technique for performing type determination processing for later data is disclosed.

JP-A-9-237361

  However, the technique of Patent Document 1 is merely a technique for performing a blocking process corresponding to a preparation stage for performing the type determination process during scanning of paper sheets, and shortens the time required for the type determination process itself. is not. For this reason, there is a problem that the time required for each sheet of paper cannot be sufficiently shortened.

  In addition, since the processing time to be secured after scanning the paper sheet is long, it is necessary to keep a gap between the paper sheet being acquired and the next paper sheet, and it is necessary to secure a long conveyance path. There was also a problem. In particular, in the case of a paper sheet identification device that transports paper sheets in the longitudinal direction, it is necessary to make the transport path longer than when transporting in the short direction. Will become remarkable.

  Therefore, how to realize a paper sheet identification apparatus or a paper sheet identification method that can improve the number of processed sheets per unit time and reduce the size of the apparatus is a major issue. It has become. Such a problem is also a problem that occurs in the same way in the paper sheet identification device that switches the transport direction of the paper sheet between the forward and reverse directions.

  The present invention has been made to solve the above-described problems of the prior art, and is capable of improving the number of processed sheets per unit time and reducing the size of the apparatus. It is another object of the present invention to provide a paper sheet identification method.

In order to solve the above-described problems and achieve the object, the present invention provides a paper sheet identification for identifying the paper sheet by reading the conveyed paper sheet with an optical line sensor oriented in a direction orthogonal to the transport direction. A correction data generation unit that generates predetermined correction data by correcting read data obtained by the optical line sensor; a storage unit that stores a template defined in advance for each type of paper sheet; A first evaluation value that calculates a difference between the correction data up to the first distance and the corresponding area in the template is calculated at a timing when the paper sheet is transported to the first distance. The correction value calculation unit and the correction data and the template from the first distance to the second distance at a timing when the paper sheet is transported to a second distance larger than the first distance. If the second evaluation value is calculated by the second evaluation value calculating means for calculating the second evaluation value obtained by quantifying the difference from the corresponding area in the mail, and the second evaluation value calculating means. Extraction means for extracting a candidate template corresponding to the type of the paper sheet from the template based on the first evaluation value and the second evaluation value; and the optical line sensor And a final discriminating unit for final discriminating the type of the paper sheet based on the candidate template extracted by the extracting unit and the correction data of the entire paper sheet when reading is completed for the whole sheet In the paper sheet identification apparatus, the correction data generation unit is configured to detect each of the read data for each line by the optical line sensor for each channel including a predetermined number of continuous pixels. The channel value stored in the channel is compared with a binarization threshold value to convert it to a binarized channel value, and then the paper sheet in the line direction of the optical line sensor based on the binarized channel value. Calculating the center position of the sheet, and correcting the skew of the paper sheet by shifting the channel that stores the channel value so that the center position of each line is the same. To do.

  Further, the present invention is the above invention, wherein the template includes information indicating another template similar to the template, and the extracting means includes information indicating that the extracted candidate template indicates the other template. If it is included, the other template is also extracted as the candidate template.

  Further, the present invention is the above invention, wherein the extraction means uses the other template in which any one of the first evaluation value, the second evaluation value, and the comprehensive evaluation value exceeds a predetermined threshold value. It is excluded from the candidate template.

  The present invention further includes a conveyance direction switching unit that switches the conveyance direction to a normal direction or a reverse direction in the above invention, wherein the correction data generation unit is configured to read data for each line read by the optical line sensor. A channel value of each channel composed of a predetermined number of continuous pixels is calculated as an average value of pixel values included in the channel, and the channel value is determined based on whether the transport direction is the forward direction or the reverse direction. The storage order of the channel values is reversed.

  In the invention described above, the correction data generation unit performs the skew correction every time the read data for each line by the optical line sensor reaches a predetermined number of lines of 1 or more. And

  Also, in the present invention according to the above invention, the correction data generation means adds the channel value to the data in which the skew is corrected within a block composed of a predetermined number of channels and a predetermined number of lines. And generating the correction data.

  In the invention described above, the correction data generation unit may generate the correction data by normalizing all the blocks when all the blocks are prepared for a predetermined area. It is characterized by.

Further, the present invention is a paper sheet identification method for identifying the paper sheet by reading the conveyed paper sheet with an optical line sensor oriented in a direction orthogonal to the transport direction, the read data by the optical line sensor A correction data generation step for generating predetermined correction data by correcting the data, a storage step for storing a template predefined for each type of the paper sheet in the storage unit, and the paper sheet from the previous position. An evaluation value calculating step for calculating an evaluation value obtained by quantifying the difference between the correction data from the previous position to the latest position and the corresponding area in the template each time the sheet is conveyed by a predetermined distance; and the evaluation value calculation Each time a new evaluation value is calculated by a process, a candidate template corresponding to the type of the paper sheet is selected from the templates based on the calculated evaluation value. When the reading of the entire paper sheet is completed by the extraction step of extracting and the optical line sensor, the paper is based on the candidate template extracted by the extraction step and the correction data of the entire paper sheet. In the paper sheet identification method including a final determination step of performing final determination on the type of leaf, the correction data generation step includes a predetermined number of continuous pixels for the read data for each line by the optical line sensor. The channel value stored for each channel is converted into a binarized channel value by comparing with a binarization threshold value, and then the paper in the line direction of the optical line sensor based on the binarized channel value. Calculate the center position of leaves and store the channel value so that the center position of each line is the same And correcting the skew of the paper sheet by shifting the channels to be.

  Further, the present invention is characterized in that, in the above invention, the extraction step further extracts the candidate template corresponding to the type of the paper sheet from the candidate templates already extracted up to the previous time. .

According to the present invention, for the read data for each line by the optical line sensor , the binarized channel is obtained by comparing the channel value stored for each channel composed of a predetermined number of continuous pixels with the threshold for binarization. The center position of the paper sheet in the line direction of the optical line sensor is calculated based on the binarized channel value after being converted into a value, and the channel value is stored in such a manner that the center position of each line is the same. Since the skew of the paper sheet is corrected by shifting the channel, the processing speed is improved by correcting the skew of the paper sheet without performing complicated image processing such as image rotation processing. There is an effect that can be.

  Also, according to the present invention, each template includes information indicating another template similar to the own template, and when the extracted candidate template includes information indicating another template, the other template is included. Is also extracted as a candidate template, so that similar templates can be extracted together by performing a process of extracting one template by associating similar templates in advance. Thus, it is possible to achieve both improvement in identification accuracy and reduction in processing time.

  In addition, according to the present invention, since one of the first evaluation value, the second evaluation value, and the comprehensive evaluation value exceeds the predetermined threshold, the other template is excluded from the candidate templates. There is an effect that the template corresponding to the type of the paper sheet can be selected from the candidate templates by the processing.

  Further, according to the present invention, the transport direction is switched between the forward direction and the reverse direction, and the channel value of each channel composed of a predetermined number of continuous pixels is read from the read data for each line by the optical line sensor. When the transport direction is different because the channel value storage order is reversed for the channel according to whether the transport direction is the forward direction or the reverse direction. Even so, it is possible to make the processing of the acquired data common by making the memory storage state of the acquired data the same.

  Further, according to the present invention, the skew correction is performed every time the read data for each line by the optical line sensor reaches a predetermined number of lines of 1 or more, so that the skew correction process is performed sequentially. There is an effect that the processing speed can be improved.

  In addition, according to the present invention, correction data is generated by adding channel values to a block composed of a predetermined number of channels and a predetermined number of lines for the data whose skew has been corrected. It is possible to improve the identification accuracy of paper sheets by removing noise by simple processing.

  Further, according to the present invention, when all the blocks for a predetermined area are prepared, the correction data is generated by normalizing all the blocks, so that the paper sheet can be obtained by appropriately performing the normalization process. There is an effect that the accuracy of class identification can be improved.

FIG. 1 is a diagram showing an outline of a paper sheet identification method according to the present invention. FIG. 2 is a block diagram illustrating a configuration of the paper sheet identification apparatus. FIG. 3 is a diagram illustrating a configuration example of a line sensor and names of banknote orientations. FIG. 4 is a diagram illustrating the activation timing of each activation target process. FIG. 5 is a diagram showing an outline of the memory storing process. FIG. 6 is a diagram showing an outline of the averaging process. FIG. 7 is a diagram showing an outline of the skew correction preparation process. FIG. 8 is a diagram showing an outline of skew correction processing and blocking processing. FIG. 9 is a diagram showing an overview of each evaluation value calculation process. FIG. 10 is a diagram showing an outline of exception processing in template extraction. FIG. 11 is a flowchart illustrating a processing procedure executed by the paper sheet identification apparatus. FIG. 12 is a flowchart illustrating a processing procedure of candidate template extraction processing.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Paper sheet identification apparatus 11 Line sensor 11a Upper unit 11aa Light emitting element 11b Lower unit 11ba Light emitting element 11bb Light emitting element 11bc Light receiving element 12 Control part 12a Data acquisition part 12b Operation instruction part 12c Correction data generation part 12ca 1 line correction part 12cb Line correction unit 12cc Blocking unit 12d First evaluation value calculation unit 12e Second evaluation value calculation unit 12f Total evaluation value calculation unit 12g Template extraction unit 12h Final discrimination unit 13 Storage unit 13a Correction data 13b Template 13c First evaluation value 13d First 2 evaluation value 13e Overall evaluation value 100 banknotes

  Hereinafter, preferred embodiments of a paper sheet identification method according to the present invention will be described in detail with reference to the accompanying drawings. In the following, after describing the outline of the paper sheet identification method according to the present invention, an embodiment of the paper sheet identification apparatus to which the paper sheet identification method according to the present invention is applied will be described. In the following, a case where a banknote is identified among paper sheets will be described.

  First, an outline of the paper sheet identification method according to the present invention will be described with reference to FIG. FIG. 1 is a diagram showing an outline of a paper sheet identification method according to the present invention. As shown in the figure, in the paper sheet identification method according to the present invention, an optical line sensor (hereinafter simply referred to as “line sensor”) provided in a direction orthogonal to the paper sheet transport direction is used. Identify paper sheets. In addition, in the same figure, it has shown about the case where the banknote 100 is conveyed in the direction (longitudinal direction) in which the long side of the banknote 100 becomes parallel to the conveyance direction 1. FIG.

  As shown in the figure, in the paper sheet identification method according to the present invention, the banknote 100 is divided into, for example, three parts in the longitudinal direction, and the first division area (see “first area” in the figure) and the next division. When the scan for the area (see “second area” in the figure) is completed, a candidate template is extracted, and when the scan for the third area (see the hatched rectangle in the figure) is completed. The main feature is that the denomination of the banknote 100 or the direction of the banknote 100 is determined.

  Specifically, the first evaluation value is calculated based on the scan data of the first area when the banknote 100 passes through the line sensor and the scan for the first area is completed (see (1) in the figure). ). Here, the first evaluation value is a value obtained by quantifying the difference between the scan data of the first area and the first area corresponding area in each template.

  Subsequently, when the scan for the second area is completed, a second evaluation value is calculated based on the scan data of the second area (see (2-1) in the figure). Then, a comprehensive evaluation value is calculated based on the first evaluation value and the second evaluation value (see (2-2) in the figure), and a template is extracted based on the calculated comprehensive evaluation value ((2 in the figure)). -3)).

  Then, when the scan for the third area (see the hatched rectangle in the figure) is completed, that is, the scan for the entire bill 100 is completed, the specific portion in the scan data of the entire bill 100 is discriminated, whereby the final A denomination / orientation discrimination (final discrimination) is executed (see (3) in the figure).

  Thus, in the paper sheet identification method according to the present invention, after dividing the paper sheet to be identified into a plurality of partial areas, narrowing down the template while calculating the evaluation value for each partial area, Since the final determination is made regarding the type and orientation of the paper sheet when the scanning of the entire paper sheet is completed, the determination result can be obtained immediately after the scanning is completed. Therefore, it is possible to shorten the processing time required per sheet.

  Further, when a plurality of paper sheets are continuously determined, the determination result is obtained almost simultaneously with the completion of the scan for one paper sheet, so that the next paper sheet can be immediately started to be scanned. . Therefore, by shortening the interval between the sheets to be discriminated, it is possible to perform high-speed processing and shorten the conveyance path length.

  In the paper sheet identification method according to the present invention, image processing is performed so that the memory storage state of the scan data is the same regardless of whether the paper sheet is conveyed in the forward direction or the reverse direction. Therefore, the number of templates is reduced, but this point will be described later with reference to FIG.

  Below, the Example which concerns on the paper sheet identification apparatus to which the paper sheet identification method based on this invention shown in FIG. 1 is applied is described using FIGS.

  First, the configuration of the paper sheet identification apparatus 10 according to the present embodiment will be described with reference to FIG. FIG. 2 is a block diagram illustrating a configuration of the paper sheet identification device 10. In the figure, only components necessary for explaining the features of the paper sheet identifying apparatus 10 are shown, and descriptions of general components such as a transport mechanism are omitted.

  As shown in FIG. 1, the paper sheet identification device 10 includes a line sensor 11, a control unit 12, and a storage unit 13. Further, the control unit 12 includes a data acquisition unit 12a, an operation instruction unit 12b, a correction data generation unit 12c, a first evaluation value calculation unit 12d, a second evaluation value calculation unit 12e, and a comprehensive evaluation value calculation unit 12f. A template extraction unit 12g and a final determination unit 12h. The correction data generation unit 12c further includes a one-line correction unit 12ca, a skew correction unit 12cb, and a blocking unit 12cc. .

  And the memory | storage part 13 memorize | stores the correction data 13a, the template 13b, the 1st evaluation value 13c, the 2nd evaluation value 13d, and the comprehensive evaluation value 13e. In the figure, for easy understanding, the first evaluation value 13c, the second evaluation value 13d, and the comprehensive evaluation value 13e are shown separately, but each evaluation value (13c, 13d, and 13e) is 1 It is good also as storing in one table etc. collectively.

  The line sensor 11 is an optical line sensor installed in a direction in which the line direction is orthogonal to the conveyance direction of the banknote 100, and acquires optical data for each line from the banknote 100 by using a light receiving and emitting element. Then, the acquired data is output to the data acquisition unit 12 a of the control unit 12.

  Here, the configuration example of the line sensor 11 and the name of the direction of the banknote 100 will be described with reference to FIG. FIG. 3 is a diagram illustrating a configuration example of the line sensor 11 and names of directions of the banknote 100. 1A is a perspective view of the line sensor 11, FIG. 1B is a cross-sectional view of the line sensor 11, and FIG. Each name is shown.

  As shown in FIG. 3A, the line sensor 11 is composed of an upper unit 11a installed along the upper surface of the conveyance path and a lower unit 11b installed along the lower surface of the conveyance path. Yes. In addition, the paper sheet identification device 10 corresponds to the switching of the transport direction, and can transport the banknote 100 in either the forward direction 31 or the reverse direction 32.

  The line sensor 11 has N + 1 channels (ch) along the line direction. The back side of FIG. 3A is called 0ch (channel) and the near side is called Nch (channel). . The width of the line sensor 11 in the line direction is greater than the width of the short side of the banknote 100.

  Moreover, as shown to (B) of FIG. 3, the light emitting element 11aa which irradiates light with respect to the banknote 100 conveyed by the conveyance path is provided in the upper unit 11a. The lower unit 11b includes a light receiving element 11bc that receives the transmitted light transmitted from the light emitting element 11aa of the upper unit 11a by the banknote 100 and receives the reflected light reflected by the banknote 100, and a light emitting element. An element 11ba and a light emitting element 11bb are provided.

  Here, the light emitting element 11aa, the light emitting element 11ba, and the light emitting element 11bb are, for example, two light sources in which an element that emits visible light such as green light and an element that emits infrared light are arranged side by side in the transport direction or line direction. The light receiving element 11bc receives the transmitted light and reflected light for visible light and the transmitted light and reflected light for infrared light in a time-sharing manner, respectively.

  And as shown to (B) of FIG. 3, the timing sensor which detects the approach of the banknote 100 is provided in the position 33 and the position 34 equivalent to the object position centering on the line sensor 11, and the banknote When 100 is transported in the forward direction 31, the timing sensor provided at the position 33 is transported in the reverse direction 32. Detect.

  In FIG. 3B, the light emitting element 11aa is arranged in the upper unit 11a, and the light emitting element 11ba, the light emitting element 11bb, and the light receiving element 11bc are arranged in the lower unit 11b. The element 11ba, the light emitting element 11bb, and the light receiving element 11bc may be arranged, and the light emitting element 11aa may be arranged in the lower unit 11b.

  By the way, as shown to (C) of FIG. 3, when conveying the banknote 100, since the pattern of the surface of a banknote 100 and a back surface has an up-down direction, respectively, about the direction of the banknote 100, it shows in the same figure. Names such as “A direction”, “B direction”, “C direction”, and “D direction” are given.

  For example, when the banknote 100 is transported toward the line sensor 11 in the left direction in a state where the surface of the banknote 100 is upright and the pattern is erected, the banknote 100 is transported in the “A direction”. It is expressed as

  Moreover, when the banknote 100 is conveyed toward the line sensor 11 in the right direction in the state where the surface of the banknote 100 is up and the pattern is erected, the banknote 100 is conveyed in the “B direction”. It is expressed. In addition, the direction about the back surface of the banknote 100 is also expressed as “C direction” and “D direction” as shown in FIG.

  Returning to the description of FIG. 2, the control unit 12 will be described. When the banknote 100 is scanned by the line sensor 11, the control unit 12 acquires each evaluation value, narrows down candidate templates based on each acquired evaluation value, and when the scan on the banknote 100 is completed. This is a processing unit that immediately performs the final determination of the denomination and direction.

  The operation instruction unit 12b includes a correction data generation unit 12c, a first evaluation value calculation unit 12d, a second evaluation value calculation unit 12e, a comprehensive evaluation value calculation unit 12f, a template extraction unit 12g, and a final determination unit 12h included in the control unit 12. Is a processing unit for instructing the processing start timing. Here, an outline of an instruction given by the operation instruction unit 12b will be described with reference to FIG.

  FIG. 4 is a diagram illustrating the activation timing of each activation target process. As shown in the figure, each time the line sensor 11 acquires data for one line, the operation instructing unit 12b starts one line correction processing for the one line correction unit 12ca of the correction data generation unit 12c. To instruct. Here, the time interval at which the line sensor 11 acquires data for one line is determined based on a transport speed or a transport distance by a transport mechanism (not shown).

  Further, every time the line sensor 11 acquires data for one line, the operation instructing unit 12b instructs the skew correction unit 12cb to start skew correction processing, and the line sensor 11 acquires data for three lines. Every time it is done, it instructs to start blocking processing by the blocking unit 12cc. The blocking process is executed following the completion of the skew correction process for the third line.

  Further, when the line sensor 11 acquires data for 24 lines, the operation instruction unit 12b instructs the first evaluation value calculation unit 12d to start the first evaluation value calculation process. Then, when the line sensor 11 acquires 48 lines of data, it instructs the second evaluation value calculation unit 12e to start the second evaluation value calculation process.

  In FIG. 4, the case where the start instruction for the first evaluation value calculation process is performed at the 24th line and the start instruction for the second evaluation value calculation process is performed at the 48th line has been described. This corresponds to the case where the scan in the longitudinal direction is completed in approximately 72 lines. That is, when about 1/3 of the scan for the banknote 100 is completed, a start instruction for the first evaluation value calculation process is issued, and when about 2/3 is completed, a start instruction for the second evaluation value calculation process is performed. .

  In the present embodiment, the case where each evaluation value is calculated every time when the scan of the banknote 100 is completed by 1/3 will be described. However, every other value such as every ½, every ¼, every ５, etc. The evaluation values may be calculated at equal intervals. Note that each evaluation value may be calculated not at equal intervals but at unequal intervals.

  In addition, when the second evaluation value calculation process by the second evaluation value calculation unit 12e is completed, the operation instruction unit 12b instructs the comprehensive evaluation value calculation unit 12f to start the comprehensive evaluation value calculation process, and the comprehensive evaluation value When the comprehensive evaluation value calculation process by the calculation unit 12f is completed, the template extraction unit 12g is instructed to start the template extraction process. Then, at the timing when the data for all the lines of the banknote 100 are acquired, the final determination unit 12h is instructed to start the final determination process.

  As described above, the operation instructing unit 12b performs correction data generation processing for each small number of lines (for example, one line or three lines), evaluation value calculation processing that is performed each time scanning is completed about 1/3, and scanning is about 2 / 3 Controls the start timing of the comprehensive evaluation value calculation process and the template extraction process performed when completed, and the final determination process performed when all the scans are completed.

  Returning to the description of FIG. 2, the correction data generation unit 12 c will be described. The correction data generation unit 12c is a processing unit that performs scan data correction processing performed for each small number of lines (for example, one line or three lines), and includes a one-line correction unit 12ca, a skew correction unit 12cb, and blocking. Part 12cc.

  The one-line correction unit 12ca is a processing unit that performs correction processing on the data for one line acquired by the line sensor 11, and performs processing for storing the data after one-line correction in the storage unit 13 as the correction data 13a. The one-line correction unit 12ca also performs a skew correction preparation process for generating skew correction preparation data used in advance by a skew correction unit 12cb described later. Here, processing contents executed by the one-line correction unit 12ca will be described with reference to FIGS.

  FIG. 5 is a diagram showing an outline of the memory storing process. Note that (A) in the figure shows a case where the banknote 100 is conveyed in the forward direction (see 31 in FIG. 3), and (B) in the figure shows the banknote 100 in the reverse direction (see 32 in FIG. 3). ) Are shown respectively. Moreover, in the same figure, it has shown about the case where the banknote 100 is conveyed to A direction (refer (C) of FIG. 3) about both (A) and (B).

  As shown in FIG. 5A, when the banknote 100 is conveyed in the forward direction, the image stored in the memory by the one-line correction unit 12ca is as shown in FIG. Here, when the transport direction is the positive direction, the one-line correction unit 12ca reverses the channel of the memory storage image (see (A-1) in the figure) ((A-2 in the figure)). )), And a reverse image (see (A-3) in the figure) is generated.

  On the other hand, as shown in FIG. 5B, when the banknote 100 is conveyed in the reverse direction, the memory storage image by the one-line correction unit 12ca is as shown in FIG. The image stored in the memory in the reverse direction conveyance (see (B-1) in the figure) is the same as the image after reversal in the normal direction conveyance (see (A-3) in the figure). As described above, the one-line correction unit 12ca controls the memory storage position so that the same memory storage image is obtained regardless of whether the transport direction is the forward direction or the reverse direction.

  FIG. 6 is a diagram showing an outline of the averaging process. FIG. 6 shows an outline of the averaging process performed in combination with the channel inversion process (see (A-2) in FIG. 5) shown in FIG. As shown to (A) of the figure, the data for one line acquired by the line sensor 11 are represented as a set of pixels.

  Here, for example, the one-line correction unit 12ca calculates an average value for pixel values of continuous eight pixels, and stores the calculated average value in a memory area corresponding to each channel.

  Then, as shown in FIG. 5B, in the case of forward conveyance, a process of replacing channel values (average value for 8 pixels) stored in each channel is performed. Specifically, the channel values are switched so that the nth channel value is stored in the (N−n) th channel. As shown in (C) of the figure, in the case of reverse conveyance, the channel values stored in each channel are not exchanged.

  FIG. 7 is a diagram showing an outline of the skew correction preparation process. In the skew correction preparation process, a binarization process for binarizing the channel value stored in each channel, a section where the banknote 100 exists in one line of data is detected, and the sheet width (of the line sensor 11) is detected. A center channel position calculation process is performed to calculate the center channel position based on the calculated paper sheet width.

  As shown in FIG. 7A, a graph 71 representing each channel value is obtained when the horizontal axis is the channel (ch) and the vertical axis is the channel value. Here, when the graph 71 exceeds a predetermined binarization threshold, “no paper sheet” is set to “0”, and when the graph 71 is equal to or less than the predetermined binarization threshold, “paper sheet is present”. By converting each channel value to “1”, a binarized channel value is obtained.

  FIG. 7B shows the binarized channel value obtained in this way. The left end detection, right end detection, and hole detection shown in FIG. 7C are performed using such binarized channel values (see FIG. 7B). As shown in FIG. 7C, the left end 74 is detected as the direction of the arrow shown in FIG. 7C, that is, the channel number that first becomes “1” from 0ch to Nch. The In the case shown in the figure, the position of the left end 74 is “1ch”.

  Further, the right end 75 is detected as the direction of the arrow shown in FIG. 7C, that is, the channel number that finally becomes “1” from 0ch to Nch. In the case shown in the figure, the position of the right end 75 is “N−2” ch. Note that a channel whose binarized channel value is “0” in the section from the left end 74 to the right end 75 is detected as a hole 73. The hole 73 corresponds to 72 shown in FIG.

  When the left end 74 and the right end 75 are detected in this way, the sheet width is calculated as a value obtained by adding 1 to the value obtained by subtracting the channel number at the left end 74 from the channel number at the right end 75. For example, in the case shown in the figure, the paper sheet width is N−2 (= (N−2) −1 + 1).

  The center channel position is calculated as a value obtained by dividing the value obtained by adding the channel number of the left end 74, the channel number of the right end 75, and 1 by 2. For example, in the case shown in the figure, the center channel position is N / 2 (= ((N−2) + 1 + 1) / 2). The center channel position described with reference to FIG. 7 is used when the skew correction unit 12cb performs the skew correction processing. This point will be described later with reference to FIG.

  Returning to the description of FIG. 2, the description of the correction data generation unit 12c will be continued. The skew correction unit 12cb is a processing unit that is activated by an instruction from the operation instruction unit 12b at every timing when the correction data 13a for three lines is prepared, and performs a simple skew correction process of the correction data 13a. is there. Here, the skew means that the direction of the long side of the banknote 100 is deviated from the transport direction, and the skew correction means a correction process in which the direction of the long side on the image data is parallel to the transport direction. Point to. Note that the skew correction unit 12cb may be activated at every timing when the correction data 13a for n lines (n is an integer of 1 or more) is prepared.

  The blocking unit 12cc is activated by an instruction from the operation instruction unit 12b every time the skew correction process by the skew correction unit 12cb is completed, and performs a process of collecting data for a predetermined line × predetermined channel as one block. It is a processing unit.

  Here, an outline of the skew correction process by the skew correction unit 12cb and the blocking process by the blocking unit 12cc will be described with reference to FIG. FIG. 8 is a diagram showing an outline of skew correction processing and blocking processing.

  3A shows the state before skew correction, FIG. 2B shows the state after skew correction, and FIG. 3C shows an example of blocking processing. Respectively. Also, the hatched area 81 in the figure corresponds to the left end 74 to the right end 75 in FIG. 7 (3), and the black rectangle 82 corresponds to the above-mentioned center channel position.

  As shown in FIG. 8A, in the state before skew correction, the center channel position is shifted in the channel (ch) direction. The skew correction unit 12cb performs correction to shift each channel position that is a channel value storage destination so that the center channel position is aligned in the channel (ch) direction. In this way, skew can be corrected as shown in FIG.

  Then, as shown in FIG. 8C, the blocking unit 12cc calculates a block value by adding each channel value (channel value) for a block composed of a predetermined line × a predetermined channel. In the figure, the case of 3 lines × 2 channels as one block is shown, but the number of lines or the number of channels constituting one block may be set to other values.

  As described above, the skew feeding correction unit 12cb executes the skew feeding correction process every several lines, for example, every three lines. The blocking unit 12cc executes the blocking process every time the skew correction process by the skew correction unit 12cb is completed.

  Further, when the block forming unit 12cc calculates the block value of one area, the average value of the block values in the corresponding area is equal to a predetermined reference value, and a curve connecting each block value and a predetermined reference Normalization processing is performed to correct each block value so that the area between the values is constant. Specifically, the sum value is obtained by adding the absolute value of the difference between each block value and the average value within the area, and the ratio between the obtained sum value and a predetermined value is calculated. Then, the difference between each block value and the average value is multiplied by the ratio, and the normalized block value is added by adding values adjusted so that the average value after multiplying the ratio becomes equal to a predetermined reference value. And

  Returning to the description of FIG. 2, the first evaluation value calculation unit 12 d will be described. The first evaluation value calculation unit 12d is activated by an instruction from the operation instruction unit 12b when, for example, the scan for the banknote 100 is completed about 1/3, and in the first area shown in FIG. 1 and each template 13b. It is a processing unit that performs a process of calculating the first evaluation value 13c by comparing the area corresponding to the first area and storing the calculated first evaluation value 13c in the storage unit 13.

  Further, the second evaluation value calculation unit 12e is activated by an instruction from the operation instruction unit 12b when, for example, the scan on the banknote 100 is completed about 2/3, and the second area shown in FIG. It is a processing unit that performs a process of calculating the second evaluation value 13d by comparing the area corresponding to the second area in 13b and storing the calculated second evaluation value 13d in the storage unit 13.

  Then, when the processing by the second evaluation value calculation unit 12e is completed, the comprehensive evaluation value calculation unit 12f is activated by an instruction from the operation instruction unit 12b, and the first evaluation value 13c and the second evaluation value 13d are obtained. The processing unit performs a process of calculating the comprehensive evaluation value 13e by adding up the templates 13b. The comprehensive evaluation value calculation unit 12f also performs a process of storing the calculated comprehensive evaluation value 13e in the storage unit 13.

  Here, an overview of each evaluation value calculation process performed by the first evaluation value calculation unit 12d, the second evaluation value calculation unit 12e, and the comprehensive evaluation value calculation unit 12f will be described with reference to FIG. FIG. 9 is a diagram showing an overview of each evaluation value calculation process.

  As shown in FIG. 9A, the first evaluation value calculation unit 12d compares the first area of the correction data 13a with the first area of each template 13b. The second evaluation value calculation unit 12e compares the second area of the correction data 13a with the second area of each template 13b.

  Further, as illustrated in FIG. 9B, the first evaluation value calculation unit 12d calculates the first evaluation value 13c for each pattern representing the type of the template 13b. Here, the first evaluation value 13c is obtained by quantifying the difference between the correction data 13 to be compared and the template 13b. For example, the first evaluation value 13c is a value obtained by adding the difference between the block values of the corresponding blocks.

  Note that the total number of blocks in which the difference in block values between corresponding blocks exceeds a predetermined threshold may be set as the first evaluation value 13c. The second evaluation value calculation unit 12e also calculates the second evaluation value by the same method as the first evaluation value calculation unit 12d.

  Then, the comprehensive evaluation value calculation unit 12f calculates the comprehensive evaluation value 13e by adding the first evaluation value 13c and the second evaluation value 13d for each type of template (for each denomination / direction). For example, as illustrated in FIG. 9B, the comprehensive evaluation value calculation unit 12 f adds the first evaluation value 13 c and the second evaluation value 13 d for the template 13 b of the pattern 1, thereby A comprehensive evaluation value 13e for the template 13b is calculated. In this way, the comprehensive evaluation value 13e is calculated for all the templates 13b.

  Returning to the description of FIG. 2, the template extraction unit 12g will be described. The template extraction unit 12g is a processing unit that performs a process of extracting one or a plurality of candidate templates from the template 13b based on the total evaluation value 13e calculated by the total evaluation value calculation unit 12f. Specifically, the template extraction unit 12g extracts the template 13b having the smallest overall evaluation value 13e as a candidate template.

  Here, each template 13b includes information indicating another template similar to the “own template” (hereinafter referred to as “similar template”). Then, when the template 13b having the smallest comprehensive evaluation value 13e includes information indicating “similar template”, the template extracting unit 12g extracts the “similar template” as a candidate template.

  If the template 13b having the smallest comprehensive evaluation value 13e does not contain information indicating “similar template”, the candidate template is narrowed down to one at this stage.

  By the way, when the template 13b with the smallest comprehensive evaluation value 13e includes information indicating the “similar template”, the template extracting unit 12g extracts the “similar template” as a candidate template. However, when such a “similar template” satisfies a predetermined condition, exception processing to be excluded from the candidate template is also performed. Here, exception processing performed by the template extraction unit 12g will be described with reference to FIG.

  FIG. 10 is a diagram showing an outline of exception processing in template extraction. As shown in the figure, the template extraction unit 12g uses the first evaluation value 13c as the predetermined first evaluation value threshold value and the second evaluation value for the “similar template” indicated by the template 13b having the smallest overall evaluation value 13e. 13d is compared with a predetermined second evaluation value threshold, and the comprehensive evaluation value 13e is compared with a predetermined comprehensive evaluation value. If there is even one evaluation value that exceeds the threshold, the corresponding “similar template” is excluded from the candidate templates.

  For example, when the template 13b of “pattern 2” is extracted as the template 13b having the smallest comprehensive evaluation value 13e, and “pattern 4” and “pattern 5” are designated as “similar templates” similar to “pattern 2” Will be described.

  In this case, the above-described exception processing is executed for the templates 13b of “Pattern 4” and “Pattern 5”. If the “pattern 4” template 13b matches the exception processing conditions, the “pattern 4” template 13b is excluded from the candidate templates. Then, the templates 13b of “Pattern 2” and “Pattern 5” are passed to the final determination unit 12h as candidate templates.

  Returning to the description of FIG. 2, the final determination unit 12h will be described. The final discriminating unit 12h is based on one or more candidate templates extracted by the template extracting unit 12g and the correction data 13a for the entire banknote 100, as shown in the final denomination and direction (shown in FIG. 3C). Further, it is a processing unit that performs processing for determining “A direction”, “B direction”, “C direction”, and “D direction”.

  The final determination unit 12h performs an auxiliary determination process of applying one or more determination formulas for each pattern of the candidate template and inspecting the validity of the candidate template with respect to the predetermined part (the characteristic part in the banknote 100) of the correction data 13a. . Then, candidate templates that are determined to be invalid by one of the determination formulas are excluded, and if one candidate template is finally left, the denomination and direction corresponding to this candidate template are set as the final determination result.

  Note that when a plurality of candidate templates finally remain or when no candidate templates remain, the banknote 100 is handled as a reject banknote, assuming that the denomination and orientation cannot be determined.

  The storage unit 13 is a storage unit configured by a storage device such as a nonvolatile memory or a hard disk drive, and includes correction data 13a, a template 13b, a first evaluation value 13c, a second evaluation value 13d, and a comprehensive evaluation value 13e. And remember. The correction data 13a and each evaluation value (the first evaluation value 13c, the second evaluation value 13d, and the overall evaluation value 13e) may be written in the volatile memory.

  The correction data 13a is data generated by the correction data generation unit 12c, and becomes the block data (see 83 in FIG. 8) for the entire banknote 100 when the scan for the entire banknote 100 is completed.

  Moreover, the template 13b is prepared for each combination pattern of the denomination and direction of the banknote 100 (see “A direction”, “B direction”, “C direction”, and “D direction” shown in FIG. 3C). Data for discrimination. Since the first evaluation value 13c, the second evaluation value 13d, and the comprehensive evaluation value 13e have already been described with reference to FIG. 9 and the like, description thereof is omitted here.

  Next, a processing procedure executed by the paper sheet identification apparatus 10 will be described with reference to FIG. FIG. 11 is a flowchart illustrating a processing procedure executed by the paper sheet identification apparatus 10. As shown in the figure, the operation instruction unit 12b determines whether or not the generation of the correction data 13a for the first area is completed by the correction data generation unit 12c (step S101), and the correction data 13a for the first area is determined. When the generation is completed (step S101, Yes), an operation instruction is given to the first evaluation value calculation unit 12d. If the determination condition in step S101 is not satisfied (step S101, No), the process in step S101 is repeated.

  Then, the first evaluation value calculation unit 12d calculates the first evaluation value 13c for each template by performing matching with all templates for the first area, and stores the calculated first evaluation value 13c in the storage unit 13. (Step S102). Subsequently, the operation instructing unit 12b determines whether or not the generation of the correction data 13a for the second area has been completed by the correction data generation unit 12c (step S103), and the generation of the correction data 13a for the second area has been completed. In such a case (step S103, Yes), an operation instruction is given to the second evaluation value calculation unit 12e. If the determination condition in step S103 is not satisfied (step S103, No), the process in step S103 is repeated.

  Then, the second evaluation value calculation unit 12e calculates the second evaluation value 13d for each template by performing matching with all templates for the second area, and stores the calculated second evaluation value 13d in the storage unit 13. (Step S104). Subsequently, the comprehensive evaluation value calculation unit 12f calculates the comprehensive evaluation value 13e by adding the first evaluation value 13c and the second evaluation value 13d for each template (denomination / direction) and the calculated comprehensive evaluation. The value 13e is stored in the storage unit 13 (step S105), and the template extraction unit 12g extracts a candidate template that is closest to the paper sheet to be identified based on the comprehensive evaluation value 13e (step S106).

  Then, the operation instruction unit 12b determines whether the correction data generation unit 12c has completed the generation of the correction data 13a for all areas (step S107), and when the generation of the correction data 13a for all areas has been completed. (Step S107, Yes), an operation instruction is given to the final determination unit 12h. In addition, when the determination condition of step S107 is not satisfied (step S107, No), the process of step S107 is repeated.

  Subsequently, the final determination unit 12h determines whether or not at least one candidate template exists (step S108). If there is a candidate template (step S108, Yes), the correction data 13a for all areas is supported. A determination process is performed (step S109). Then, it is determined whether or not the candidate template (denomination / direction) is narrowed down to one (step S110). If the candidate template is narrowed down to one (step S110, Yes), the denomination is determined. (Step S111), and the process is terminated.

  In addition, when the determination condition of step S108 is not satisfied (step S108, No) and when the determination condition of step S110 is not satisfied (step S110, No), it is determined that there is no corresponding denomination (step S112), and the process is performed. finish. Further, in FIG. 11, the denomination is determined, but not only the denomination but also the direction of the banknote 100 (four directions shown in FIG. 3C, that is, “A direction”, “B direction”). , “C direction” and “D direction”).

  In step S104, the template matching when calculating the second evaluation value for the second area is performed for templates corresponding to all denominations and directions. It is good also as limiting to the template whose 1 evaluation value became below predetermined value. By doing in this way, the processing time of a 2nd evaluation value calculation process can be shortened. In this case, since the second evaluation value (the n-1th evaluation value when all areas are divided into n) can be used as the comprehensive evaluation value in step S105, the process of the comprehensive evaluation value calculation process Time can be shortened.

  Next, a detailed processing procedure of the candidate template extraction process shown in step S106 of FIG. 11 will be described with reference to FIG. FIG. 12 is a flowchart illustrating a processing procedure of candidate template extraction processing. As shown in the figure, the template extraction unit 12g extracts a template 13b having the smallest comprehensive evaluation value 13e as a candidate template (step S201).

  Subsequently, it is determined whether or not a similar template is specified in the extracted candidate template (step S202). If a similar template is specified (step S202, Yes), the similar template is added to the candidate template (step S202). Step S203). If the determination condition in step S202 is not satisfied (No in step S202), the process proceeds to step S204 without performing the process in step S203.

  Subsequently, the template extraction unit 12g determines whether any of the first evaluation value 13c, the second evaluation value 13d, or the comprehensive evaluation value 13e is greater than a predetermined threshold for the candidate template (step S204).

  If any of the evaluation values is larger than the predetermined threshold (Yes at Step S204), the corresponding template is excluded from the candidate templates (Step S205). If the determination condition in step S204 is not satisfied (No in step S204), the process proceeds to step S206 without performing the process in step S205.

  Subsequently, it is determined whether or not there is a next candidate template (step S206). If there is a next candidate template (step S206, Yes), the next candidate template is set as a target template (step S207). ), And repeats the processing from step S204.

  On the other hand, if there is no next template (No at Step S206), the candidate template is output (Step S208), and the process is terminated. Here, when there is no candidate template, it outputs that there is no candidate template. Note that the final determination result is the denomination, orientation, and front / back of the candidate template output in step S208.

  As described above, in the present embodiment, the correction data generation unit generates predetermined correction data by correcting the read data by the optical line sensor, and the storage unit is defined in advance for each type of paper sheet. The first evaluation value calculation unit quantifies the difference between the correction data up to the first distance and the corresponding area in the template at the timing when the paper sheet is conveyed to the first distance. When the evaluation value of 1 is calculated and the paper sheet is conveyed to a second distance that is larger than the first distance, the second evaluation value calculation unit corrects from the first distance to the second distance. When the second evaluation value obtained by quantifying the difference between the data and the corresponding area in the template is calculated, and the second evaluation value is calculated, the template extraction unit performs the first evaluation value and the second evaluation value. Inside a template based on If the candidate template corresponding to the type of the paper sheet is extracted, and reading of the entire paper sheet is completed by the optical line sensor, the final discriminating unit detects the extracted candidate template and the correction data of the entire paper sheet. Based on the above, the paper sheet identification device is configured to make a final determination on the type of paper sheet.

  Therefore, by performing the data acquisition process for paper sheets and the type determination process in parallel, the type determination process can be completed when data acquisition for the entire paper sheet is completed. The number of processed sheets can be improved. In addition, the distance between the paper sheets and the paper sheets can be reduced, and the conveyance path length can be shortened. Therefore, the apparatus size can be reduced.

  In the above-described embodiment, when the second evaluation value is calculated, a comprehensive evaluation value based on the first evaluation value and the second evaluation value is calculated, and a candidate template is extracted based on the calculated comprehensive evaluation value. Explained when to do. However, the present invention is not limited to this, and the candidate templates may be narrowed down based on the first evaluation value when the first evaluation value is calculated. By doing so, the number of candidate templates to be processed can be reduced, so that the processing speed can be further improved.

  In the above-described embodiment, the case where the paper sheet is divided into three in the transport direction has been described. However, the number of divisions may be any number. For example, when paper sheets are divided into n (n is an integer equal to or greater than 1), final candidate templates are narrowed down until the evaluation value for the (n-1) th area is calculated. That's fine. In addition, each time an evaluation value for each area is calculated, the candidate templates may be narrowed down to finally determine one candidate template. When the number of divisions (n) is large, it is possible to determine the denomination / direction before calculating the (n−1) th evaluation value, that is, the (n−2) th or less.

  As described above, the paper sheet identification apparatus and the paper sheet identification method according to the present invention are useful for paper sheet identification, particularly when it is desired to improve the number of processed sheets per unit time or the size of the apparatus. Suitable for miniaturization.

Claims (9)

A paper sheet identification device for identifying the paper sheet by reading the transported paper sheet with an optical line sensor in a direction orthogonal to the transport direction,
Correction data generation means for generating predetermined correction data by correcting read data by the optical line sensor;
Storage means for storing a predefined template for each type of the paper sheet;
A first evaluation value that calculates a difference between the correction data up to the first distance and the corresponding area in the template is calculated at a timing when the paper sheet is transported to the first distance. An evaluation value calculating means;
A difference between the correction data from the first distance to the second distance and a corresponding area in the template at a timing when the paper sheet is conveyed to a second distance larger than the first distance. Second evaluation value calculating means for calculating a second evaluation value that has been digitized;
If the second evaluation value is calculated by the second evaluation value calculating means, the type of the paper sheet is selected from the template based on the first evaluation value and the second evaluation value. Extraction means for extracting corresponding candidate templates;
If the reading of the entire paper sheet is completed by the optical line sensor, the final type of the paper sheet is finalized based on the candidate template extracted by the extracting unit and the correction data of the entire paper sheet. In a paper sheet identification device provided with a final determination means for performing determination ,
The correction data generation means includes
The read data for each line by the optical line sensor is converted into a binarized channel value by comparing a channel value stored for each channel composed of a predetermined number of continuous pixels with a binarization threshold value. Based on the binarized channel value, the center position of the paper sheet in the line direction of the optical line sensor is calculated, and the channel value is stored in such a manner that the center position of each line is the same. The paper sheet identification apparatus according to claim 1, wherein the skew of the paper sheet is corrected by shifting the channel . The correction data generation means includes
Bill discrimination apparatus according to claim 1, characterized in that the correction of the skew each time read data of each line by the optical line sensor has reached the number of one or more predetermined lines. The correction data generation means includes
Wherein the skew is corrected data, according to the channel value, in claim 2, characterized in that to generate the correction data by adding in a block consisting of a predetermined number of channels and a predetermined number of lines Paper sheet identification device. The correction data generation means includes
The paper sheet identification apparatus according to claim 3 , wherein when all the blocks are arranged for a predetermined area, the correction data is generated by normalizing all the blocks. The template is
Contains information pointing to other templates similar to the template,
The extraction means includes
The paper sheet identification apparatus according to claim 1 , wherein when the extracted candidate template includes information indicating the other template, the other template is also extracted as the candidate template. The extraction means includes
According to claim 5, characterized in that exclude the first evaluation value, the said other templates either exceeds a predetermined threshold value of the second evaluation value or the total evaluation value from the candidate template Paper sheet identification device. A transport direction switching means for switching the transport direction to a forward direction or a reverse direction;
The correction data generation means includes
Whether the channel value of each channel composed of a predetermined number of continuous pixels is calculated as an average value of the pixel values included in the channel for the read data for each line by the optical line sensor, and whether the transport direction is the positive direction The paper sheet identification apparatus according to any one of claims 1 to 6 , wherein a storage order of the channel values is reversed with respect to the channel according to whether the direction is the reverse direction. A paper sheet identification method for identifying the paper sheet by reading the conveyed paper sheet with an optical line sensor oriented in a direction orthogonal to the transport direction,
A correction data generation step of generating predetermined correction data by correcting read data by the optical line sensor;
A storage step of storing a template predefined for each type of the paper sheet in a storage unit;
Evaluation that calculates an evaluation value that quantifies the difference between the correction data from the previous position to the latest position and the corresponding area in the template each time the paper sheet is conveyed by a predetermined distance from the previous position. A value calculation process;
An extraction step of extracting a candidate template corresponding to the type of the paper sheet from the template based on the calculated evaluation value each time the new evaluation value is calculated by the evaluation value calculation step;
If the reading of the entire paper sheet is completed by the optical line sensor, the final type of the paper sheet is finalized based on the candidate template extracted by the extraction step and the correction data of the entire paper sheet. In a paper sheet identification method including a final discrimination process for performing discrimination ,
The correction data generation step includes
The read data for each line by the optical line sensor is converted into a binarized channel value by comparing a channel value stored for each channel composed of a predetermined number of continuous pixels with a binarization threshold value. Based on the binarized channel value, the center position of the paper sheet in the line direction of the optical line sensor is calculated, and the channel value is stored in such a manner that the center position of each line is the same. A paper sheet identification method , wherein the skew of the paper sheet is corrected by shifting the channel . The extraction step includes
9. The paper sheet identification method according to claim 8 , further comprising extracting the candidate template corresponding to the type of the paper sheet from the candidate templates already extracted up to the previous time.

Images