JP6534278B2 - Paper sheet discrimination apparatus and paper sheet discrimination method - Google Patents

Description

  The present invention relates to a sheet discrimination apparatus and a sheet discrimination method capable of discriminating whether or not a tape is attached to a sheet.

  Paper sheets such as banknotes are often folded with a crease along the lateral direction. And, if a tear along the short direction occurs due to the fold, it may be reinforced by the tape.

  As a prior art that can detect a sheet having a tape attached in this manner, there is a sheet thickness detection device disclosed in Patent Document 1, for example.

  The sheet thickness detecting device disclosed in Patent Document 1 includes a sheet passing between a reference roller provided on a fixed rotating shaft and a detection roller provided in contact with the reference roller. Mechanical thickness detection is performed using a detection block that is rotationally displaced according to the thickness of the sensor.

Patent No. 4819162 gazette U.S. Pat. No. 8,089,045

  However, in mechanical thickness detection, only a tape having a thickness of 50 μm or more can be detected, and for example, a thin tape such as 30 μm or 40 μm can not be detected.

  On the other hand, the detection device disclosed in Patent Document 2 optically detects wrinkles and tape sticking of paper sheets instead of mechanical thickness detection. However, in the detection device disclosed in Patent Document 2, by irradiating light to the paper sheets from diagonally above, bright reflection of the tape side end located on the light source side than the tape center or the tape center can be obtained. Also, the shadow of the side edge of the tape away from the light source is detected to detect whether the tape is attached to the paper sheet, and it is necessary to devise the position of the light source, which complicates the structure of the device. .

  The present invention provides a sheet discriminating apparatus and a sheet discriminating method capable of discriminating whether or not a tape is pasted even if a thin tape is pasted in view of the above situation. The purpose is to

  In order to achieve the above object, a sheet discriminating apparatus according to the present invention is a sheet discriminating apparatus for discriminating whether a tape is attached to a sheet based on a captured image of the sheet, and is valid. Storage unit for storing a reference edge point which is an edge point of the density gradient direction range along the longitudinal direction of the valid paper sheet included in a captured image of a thin paper sheet, and imaging of the paper sheet to be discriminated An edge point extraction unit that extracts an edge point of a density gradient direction range along a longitudinal direction of a sheet to be discriminated from an image, and an edge point extracted by the edge point extraction unit to the reference edge point A remaining edge point extraction unit which removes a remaining edge point by removing a corresponding edge point and an edge point estimated to be generated due to a bright area of a pixel not included in the captured image of the valid paper sheet; Based on the remaining edge point A configuration that includes an analysis unit for analyzing whether the tape is stuck to the paper sheet serving as the determination target Te (first configuration).

  According to such a configuration, since the tape attachment is optically detected, it is possible to determine whether the tape is attached to the paper even if the thin tape is attached. In addition, since bright reflections and shadows of the side edge of the tape generated by irradiating light to the paper sheet from obliquely above are not used, there is no need to devise the position of the light source. Therefore, the structure of the device can be simplified.

  Further, by extracting the remaining edge points as described above, it is possible to remove the edge points generated due to the design of the paper sheet from the edge points which may be generated due to the tape side edge. In addition, since the pixels become dark due to the influence of the tape in the area where the tape is attached, the extraction of the remaining edge point as described above may cause the tape side edge to be likely to be generated due to the tape side edge. It is possible to remove an edge point estimated to have not occurred due to an edge. Therefore, the detection accuracy of tape sticking can be made high.

  In the sheet discriminating apparatus having the first configuration, the reference edge point is a first reference edge point which is an edge point of a first density gradient direction range included in a captured image of the valid sheet, and The first density gradient direction range and the second reference edge point which is an edge point of the second density gradient direction range which is the reverse direction, and the edge point extraction unit determines from the captured image of the sheet to be discriminated An edge point of the first concentration gradient direction range and an edge point of the second concentration gradient direction range are extracted, and the remaining edge point is an edge point of the first concentration gradient direction range extracted by the edge point extraction unit A first remaining edge from which an edge point corresponding to the first reference edge point and an edge point estimated to be generated due to a bright region are not included in a pixel not included in the captured image of the valid paper sheet Point and the above From the edge points of the second density gradient direction range extracted by the edge point extraction unit, edge points corresponding to the second reference edge point and pixels not included in the captured image of the valid paper sheet are bright It is preferable that the configuration (second configuration) includes a second remaining edge point from which an edge point estimated to have occurred due to the region is removed.

  According to such a configuration, it is possible to detect each of the right side tape side end and the left side tape side end, and it is possible to further increase the detection accuracy of tape sticking.

  In the sheet discriminating apparatus having the first or second configuration, the analyzing unit determines whether a tape is attached to the sheet to be discriminated based on the density of the remaining edge point in a predetermined size area. It is preferable that it is the structure (3rd structure) to analyze.

  According to such a configuration, by changing the shape of the predetermined size area, it is possible to detect tapes having different attachment directions. For example, by making the predetermined size area longitudinally elongated, it is possible to detect a tape stuck longitudinally (in the short direction of paper sheets), and by making the predetermined size area obliquely inclined, it is inclined obliquely The attached tape can be detected.

  In the paper sheet discriminating apparatus according to any one of the first to third configurations, the remaining edge point extraction unit converts the picked up image of the paper sheet to be discriminated into the picked up image of the valid paper sheet. An edge point located around the thin line-like area is also the edge when it is estimated that a thin fine line-like area having a bright pixel exists along the short direction of the paper sheet to be discriminated which is not included It is preferable that it is the structure (4th structure) remove | eliminated from the edge point extracted by the point extraction part.

  According to such a configuration, it is possible to remove from the edge points that may be generated due to the side edge of the tape, the edge points that appear under the influence of the thin bright line generated by the strong fold. Therefore, the detection accuracy of the tape sticking can be further enhanced.

  In the sheet discriminating apparatus having any one of the first to fourth configurations, in the captured image of the sheet to be discriminated, from one end to the other end in the longitudinal direction of the sheet to be discriminated A first boundary that changes from the first area to a second area that is darker than the first area, and a second area that changes from the third area to a fourth area that is brighter than the third area. Based on the pixel value of a pixel located in a region between a detection unit that detects a boundary and the first boundary and the second boundary located closer to the other end than the first boundary, A determination unit configured to determine whether a tape is attached to an area between the first boundary portion and the second boundary portion positioned closer to the other end than the first boundary portion (fifth embodiment) It is preferable that

  According to such a configuration, it is possible to determine whether the tape is attached without extracting the edge point, and therefore, even if the edge generated by the side end of the attached tape is not clear, the sheet is a tape Can be detected.

  In the sheet discriminating apparatus having the fifth configuration, the judging unit judges that a distance between the first boundary and the second boundary located closer to the other end than the first boundary is determined. Only in the region between the first boundary and the second boundary located closer to the other end than the first boundary, but only when the width is the minimum width and the maximum width of the target tape. Based on the pixel value of the pixel located, it is determined whether the tape is stuck in the area between the first boundary and the second boundary located closer to the other end than the first boundary. It is preferable that it is a structure (6th structure).

  According to such a configuration, a combination to be determined is determined without determining all combinations of the first boundary and the second boundary positioned closer to the other end than the first boundary. Since the narrowing is performed in consideration of the width of the tape, the determination time can be shortened and the false detection can be prevented.

  In the sheet discriminating apparatus having the fifth or sixth configuration, the determination unit may be configured to interpose the first boundary portion and the second boundary portion positioned closer to the other end than the first boundary portion. A tape is attached to a region between the first boundary portion and the second boundary portion located on the other end side of the first boundary portion based on the dispersion of pixel values of pixels located in the region. It is preferable that it is the structure (7th structure) which determines whether it exists.

  In the area where the tape is attached, the unevenness of density is small, so the dispersion of pixel values becomes small. Therefore, according to such a configuration, the detection accuracy of the tape sticking can be enhanced.

  In the sheet discriminating apparatus according to any one of the fifth to seventh configurations, the determination unit determines the first boundary portion and the first boundary portion based on a pixel value of the first region and a pixel value of the fourth region. It is preferable that it is a structure (the 8th structure) which determines whether the area | region with the said 2nd boundary part located in the said other end side rather than a 1st boundary part is stuck.

  According to such a configuration, the conditions for detecting the first boundary portion and the second boundary portion can be set loosely, so that tape sticking is detected even when the background density of the sheet to be discriminated is not uniform. can do.

In order to achieve the above object, a sheet discrimination method according to the present invention is applied to a sheet discrimination apparatus that determines whether a tape is attached to a sheet based on a captured image of the sheet. A sheet discriminating method, wherein the sheet discriminating apparatus extracts edge points of a density gradient direction range along a longitudinal direction of a sheet to be discriminated from a captured image of a sheet to be discriminated an edge point extraction step of the paper sheet determination apparatus, the edge points extracted by the edge point extraction process, along a longitudinal direction of the legitimate paper sheet included in the captured image of legitimate sheet An edge point corresponding to a reference edge point which is an edge point of the density gradient direction range and an edge point estimated to be generated due to a bright region are removed from pixels not included in the image of the valid paper sheet To extract the remaining edge points Di-point extracting process, the paper sheet determination apparatus is configured to include an analysis step of analyzing whether the tape is stuck to the paper sheet serving as the determination target based on the remaining edge points.

  According to such a configuration, since the tape attachment is optically detected, it is possible to determine whether the tape is attached to the paper even if the thin tape is attached. In addition, since bright reflections and shadows of the side edge of the tape generated by irradiating light to the paper sheet from obliquely above are not used, there is no need to devise the position of the light source. Therefore, implementation of the above-described paper sheet discrimination method structure becomes easy.

  Further, by extracting the remaining edge points as described above, it is possible to remove the edge points generated due to the design of the paper sheet from the edge points which may be generated due to the tape side edge. In addition, since the pixels become dark due to the influence of the tape in the area where the tape is attached, the extraction of the remaining edge point as described above may cause the tape side edge to be likely to be generated due to the tape side edge. It is possible to remove an edge point estimated to have not occurred due to an edge. Therefore, the detection accuracy of tape sticking can be made high.

  According to the present invention, even when a thin tape is attached, it can be determined whether the tape is attached to a sheet.

A diagram showing an outline of a sheet discrimination method according to the present invention Block diagram showing the configuration of a sheet discriminating apparatus according to the first embodiment of the present invention Diagram showing an example of a smoothing filter Figure showing an example of Laplacian filter Figure showing an example of a horizontal direction Prewitt filter Figure showing an example of a vertical direction Prewitt filter Flow chart showing the first estimation process Pixel value histogram of the captured image of the bill to be discriminated Flow chart showing the second estimation process A diagram showing average pixel values in the lateral direction of a captured image of a bill to be discriminated A diagram showing an example of a density detection filter Diagram showing another example of density detection filter Block diagram showing the configuration of a sheet discriminating apparatus according to the second embodiment of the present invention Flow chart showing the boundary detection process Diagram showing the arrangement of the point of interest, the left area, and the right area Flow chart showing determination processing Diagram schematically showing the pixel value distribution when the tape is attached to a place where the background density is not uniform A diagram schematically showing the pixel value distribution when there is ink bleeding at a place where the background density is not uniform.

  Hereinafter, preferred embodiments of the paper sheet discrimination method according to the present invention will be described in detail with reference to the attached drawings. In the following, after the outline of the paper sheet discrimination method according to the present invention is described, an embodiment of a bill discrimination device to which the paper sheet discrimination method according to the present invention is applied will be described.

<Outline of the paper sheet discrimination method according to the present invention>
FIG. 1 is a view showing an outline of a sheet discrimination method according to the present invention. In the paper sheet discrimination method according to the present invention, first, the paper sheet to be discriminated is irradiated with light, and the paper sheet to be discriminated based on the transmitted light or the reflected light from the paper sheet to be discriminated The captured image of is generated (see (1) in FIG. 1).

  Next, edge points in the density gradient direction range along the longitudinal direction of the sheet to be discriminated are extracted from the captured image of the sheet to be discriminated (see (2) in FIG. 1). Here, the edge refers to a portion where the contrast of the image is significantly changed. In the following description, the direction from the bright side to the dark side of the edge density is taken as the density gradient direction. Further, an edge point is a pixel corresponding to an edge in an image.

  For example, as shown in (2) of FIG. 1, the longitudinal direction of the sheet to be discriminated has a width of ± 12.5 degrees with respect to the direction from the left end to the right end of the sheet to be discriminated The edge point of the density gradient direction range (indicated as A direction range in (2) in Fig. 1) along with the width of ± 12.5 degrees with respect to the direction from the right edge to the left edge of the paper sheet The edge points of the density gradient direction range (indicated as the B direction range in (2) of FIG. 1) along the longitudinal direction of the sheet to be discriminated are extracted.

  In the case of paper sheets such as banknotes, it is assumed that the tape is stuck along the short direction, so the extraction of the specific edge point as described above may result in an edge point that may be generated due to the tape side edge. Can be extracted.

  Next, an edge point corresponding to a reference edge point that is an edge point of a density gradient direction range along the longitudinal direction of a valid paper sheet included in a captured image of a valid paper sheet from the extracted specific edge point And the remaining edge point is extracted by removing the edge point estimated to be generated due to the region where the pixels not included in the captured image of the valid paper sheet are bright, that is, the region having a large pixel value in this embodiment See FIG. 1 (3)). Here, the term "authorized sheet" means a sheet which is not soiled such as dirt, wrinkles or creases and is not attached with a tape.

  "The density gradient direction range along the longitudinal direction of the sheet to be discriminated" used when extracting a specific edge point, and "a captured image of a valid sheet" used when obtaining a reference edge point Although it does not have to be completely coincident with “the density gradient direction range along the longitudinal direction of the legitimate paper sheet to be included”, it is desirable that they are completely coincident.

  By extracting the remaining edge points as described above, it is possible to remove the edge points generated due to the design of the paper sheet from the edge points which may be generated due to the tape side edge. Further, in the area where the tape is attached, the pixel becomes dark due to the influence of the tape, that is, the pixel value becomes small in the present embodiment, so the extraction of the remaining edge point as described above causes the tape side edge. From the possible edge points, edge points estimated to have not occurred due to the tape side edge can be removed.

  Finally, based on the extracted remaining edge point, it is analyzed whether the tape is stuck to the sheet to be discriminated.

  As described above, in the paper sheet discrimination method according to the present invention, tape sticking is optically detected. Therefore, even when a thin tape is attached, it can be determined whether the tape is attached to a sheet. In addition, since bright reflections and shadows of the side edge of the tape generated by irradiating light to the paper sheet from obliquely above are not used, there is no need to devise the position of the light source. Furthermore, in the paper sheet discrimination method according to the present invention, due to the extraction of the remaining edge point as described above, the edge point that may be generated due to the tape side edge is attributed to the pattern of the paper sheet. The generated edge points can be removed. Also, since the pixel value is small due to the influence of the tape in the area where the tape is attached, the extraction of the remaining edge points as described above may result in the edge point that may have arisen due to the tape side edge. It is possible to remove an edge point estimated to have not occurred due to the edge. Therefore, the detection accuracy of tape sticking can be made high.

  Hereinafter, an embodiment of a bill discriminating apparatus to which the bill discriminating method according to the present invention described with reference to FIG. 1 is applied will be described in detail. However, the bill discriminating method according to the present invention is not only bill The present invention can also be applied to any other sheet such as a check, a bill and a gift certificate.

First Embodiment
FIG. 2 is a block diagram showing the configuration of the banknote discrimination device 10 according to the first embodiment of the present invention. In addition, in FIG. 2, only the component required in order to demonstrate the characteristic of the banknote discrimination | determination apparatus 10 is shown, and the description about a general component is abbreviate | omitted.

  The banknote discrimination device 10 includes an image line sensor unit 1, a control unit 2, and a storage unit 3. The control unit 2 further includes an image data acquisition unit 2a, a denomination identification unit 2b, an edge point extraction unit 2c, a remaining edge point extraction unit 2d, and an analysis unit 2e. In addition, the storage unit 3 stores a denomination identification template 3a, various filters 3b, a reference edge point 3c, and analysis reference information 3d.

  The image line sensor unit 1 is a sensor unit that receives transmitted light or reflected light from a bill transported by a transport mechanism (not shown), and is configured by arranging a plurality of light receiving sensors in a straight line. The light to be applied to the bill is preferably light of an infrared wavelength which does not easily react to stains, but may be visible light such as white, red, green or blue. In addition, the image line sensor unit 1 also performs a process of outputting the light reception result to the image data acquisition unit 2 b of the control unit 2.

  The control unit 2 is a processing unit that generates image data of a bill based on the output from the image line sensor unit 1 and analyzes the generated image data to determine whether a tape is attached to the bill.

  The image data acquisition unit 2a is a processing unit that develops output data from the image line sensor unit 1 in a memory, combines the data for each bill, and generates image data for the entire bill. Further, the image data acquisition unit 2a performs a process of outputting the generated image data to the denomination identification unit 2b and the edge point extraction unit 2c, respectively.

  The denomination identification unit 2b determines the denomination of the bill by comparing the feature pattern of the image data generated by the image data acquisition unit 2a with the denomination identification template 3a, which is a feature pattern for each denomination of the bill. Do.

  The edge point extraction unit 2c smoothes the image data output from the image data acquisition unit 2b using various filters 3b, and extracts edge points in the density gradient direction range along the longitudinal direction of the bill from the smoothed image data Is a processing unit that performs processing to The edge point extraction unit 2c also performs a process of outputting an edge point image indicating the extracted edge point to the remaining edge point extraction unit 2d.

  In the present embodiment, the edge point extraction unit 2c has a width of ± 12.5 degrees with respect to the direction from the left end to the right end of the banknote, and the concentration gradient direction range along the longitudinal direction of the banknote (hereinafter referred to as A range) Of the density gradient direction range (hereinafter also referred to as the B direction range) along the longitudinal direction of the bill with an edge point of) and a width of ± 12.5 degrees with respect to the direction from the right end to the left end of the bill Extract edge points. The reason for having a width of ± 12.5 degrees here is to take into consideration that the tape is stuck obliquely. However, +12.5 degrees is an example, and any value between 0 degrees and 45 degrees may be used. Similarly, -12.5 degrees is also an example, and any value between -45 degrees and 0 degrees may be used. Further, each concentration gradient direction range may not be line symmetrical with respect to the longitudinal direction of the bill, and both concentration gradient direction ranges may not be line symmetrical with respect to the short direction of the bill. Further, although either one of the right side tape side end and the left side tape side end can not be detected with certainty, only one of the former edge point and the latter edge point may be extracted.

  In the filter process of image data, the center of the filter and the pixel of interest are made to overlap, and the coefficient of the filter and the pixel value of the pixel are multiplied at each overlapping position, and the sum of the respective multiplication values is made the pixel value of the pixel of interest. .

  In smoothing, for example, a smoothing filter in which coefficients of 3 × 3 pixels shown in FIG. 3 are all 1 can be used. The smoothing can remove random noise superimposed on the image. Even when the smoothing is not performed, it is possible to detect the edge point and the density gradient direction.

  In addition, in detection of edge points, for example, an 8 direction Laplacian filter of 3 × 3 pixels shown in FIG. 4 can be used. The zero crossing point detected by the filter processing using the Laplacian filter is the edge portion of the original image.

  In addition, in detection of the density gradient direction, for example, a 3 × 3 pixel horizontal direction Prewitt filter shown in FIG. 5 and a 3 × 3 pixel vertical direction Prewitt filter shown in FIG. 6 can be used. Density gradient according to the ratio of the positive / negative combination and the absolute value between the pixel value of the target pixel obtained by the filtering process using the horizontal Prewitt filter and the pixel value of the target pixel obtained by the filtering process using the vertical Prewitt filter The direction can be detected.

  In the above description, an example is shown in which an edge point and a density gradient direction are detected using an 8-way Laplacian filter, a horizontal direction Prewitt filter, and a vertical direction Prewitt filter, but other filters such as a Sobel filter may be used. It is also possible to combine various filters as appropriate. Also, the sizes and coefficients of the various filters are not particularly limited.

  Referring back to FIG. 2, the remaining edge point extraction unit 2d will be described.

  The remaining edge point extraction unit 2d is included in an edge point corresponding to the reference edge point of the denomination identified by the denomination identification unit 2b from the edge point extracted by the edge point extraction unit 2c and in the captured image of the valid bill. The edge point estimated to have arisen due to the large area of the pixel value which is not detected is removed.

  In addition, the remaining edge point extraction unit 2d is a thin line-shaped area having a large pixel value along the short direction of the bill to be discriminated which is not included in the captured image of the valid bill in the image data of the bill to be discriminatively symmetrical When it is estimated that there exists, the edge point located around the thin line region is also removed from the edge points extracted by the edge point extraction unit 2c. The order of removing the above three types of edge points is not particularly limited. Also, one type may be sequentially removed, or a plurality of types of edge points may be simultaneously removed.

  In the present embodiment, the remaining edge point extraction unit 2d generates an edge point image indicating an edge point of the A direction range extracted by the edge point extraction unit 2c, an image indicating a reference edge point of the A direction range described later, and The result of each estimation process is used to generate an edge point image indicating the remaining edge points in the A direction range. Similarly, the remaining edge point extraction unit 2d generates an edge point image indicating an edge point of the B direction range extracted by the edge point extraction unit 2c, an image indicating a reference edge point of the B direction range described later, and estimations described later. The result of the processing is used to generate an edge point image indicating the remaining edge points in the B direction range. Also, the remaining edge point extraction unit 2d performs a process of outputting an edge point image indicating a remaining edge point in the A direction range and an edge point image indicating a remaining edge point in the B direction range to the analysis unit 2e.

  Here, an example of the creation procedure of the reference edge point will be described. The reference edge point is an edge point appearing in the image data of a valid bill, and is used as a mask pattern of edge points to be removed in order to extract an edge point of the tape.

  First, for the image data of a plurality of (for example, 200) valid banknotes, edge points in the A direction range are detected, and if there are edge points at the same position with the number of judgment threshold (for example 5) or more, the A direction Set as the effective edge point of the range. Then, expansion processing is performed to expand the effective edge point to the peripheral pixels, with the main purpose of absorbing the positional deviation between the bill to be discriminated and the valid bill. The peripheral pixels may be, for example, eight neighboring pixels adjacent to each other, or one neighboring pixel adjacent in a direction from the left end to the right end of the bill. The effective edge point of the A-direction range obtained in this manner is taken as the reference edge point of the A-direction range. The storage unit 3 stores reference edge points in the A direction range in the form of an image indicating reference edge points in the A direction range.

  Similarly, for the image data of a plurality of (for example, 200) valid banknotes, edge points in the B direction range are detected, and if there are edge points at the same position with the number of determination threshold (for example, 5) or more, B It is the effective edge point of the direction range. Then, expansion processing is performed to expand the effective edge point to the peripheral pixels, with the main purpose of absorbing the positional deviation between the bill to be discriminated and the valid bill. The peripheral pixels may be, for example, eight neighboring pixels adjacent to each other, or one neighboring pixel adjacent in a direction from the left end to the right end of the bill. The effective edge point of the B-direction range obtained in this manner is taken as the reference edge point of the B-direction range. The storage unit 3 stores reference edge points in the B-direction range in the form of an image showing reference edge points in the B-direction range.

  Since the reference edge point in the A direction range and the reference edge point in the B direction range described above differ depending on the denomination, they are stored for each denomination. In addition, the creation of the reference edge point described above is performed before the product shipment of the banknote discrimination device 10, and the banknote discrimination device 10 is shipped with the reference edge point 3c stored in advance in the storage unit 3. Creation of a reference edge point may be implemented using apparatuses other than the banknote discrimination device 10, and may be implemented using the banknote discrimination device 10 itself. Further, an interface for enabling rewriting of the storage content of the storage unit 3 may be provided in the bill discriminating device 10 so that addition or correction of the reference edge point 3c can be easily performed.

  Here, a first estimation is performed to estimate whether the target edge point is an edge point generated due to a region having a large pixel value such as, for example, a wrinkle or a fold not included in a captured image of a valid bill. The process will be described with reference to the flowchart of FIG. The remaining edge point extraction unit 2d executes the first estimation process. The target edge point processed by the first estimation process has the largest edge point extracted by the edge point extraction unit 2c. For example, before execution of the first estimation process, the edge point corresponding to the reference edge point of the gold type identified by the denomination identification unit 2b is removed from the edge point extracted by the edge point extraction unit 2c. In this case, the remaining edge points become the notable edge points to be processed by the first estimation process.

  The remaining edge point extraction unit 2d obtains the most frequently appearing pixel value of the captured image of the bill to be discriminated (step S1). FIG. 8 is a diagram showing an example of a pixel value histogram related to a captured image of a bill to be discriminated. In the example shown in FIG. 8, the density of the captured image of the bill to be discriminated is 256 tones, the pixel of pixel value 0 is the darkest pixel, and the pixel of pixel value 256 is the brightest pixel, and the pixel value 188 Is the most frequent pixel value. In the present embodiment, the “most frequently appearing pixel value” is used in step S2 and subsequent steps as a pixel value representing the background of the captured image of the bill to be determined. The processes after step S2 are performed on each notable edge point.

  Next, the remaining edge point extraction unit 2d sets each of the pixel values of the three pixels adjacent to the left side of the target edge point (the left end side of the bill) and continuously arranged in the direction from the right end to the left end of the bill. Pixel value of three pixels smaller than the value of “most frequently appearing pixel value” minus 5 or the pixel value of three pixels continuously arranged in the direction from the left end to the right end of the bill adjacent to the right side (right end side of the bill) It is determined whether the first condition that each is smaller than the above-mentioned "most frequently appearing pixel value" minus 5 is satisfied (step S2).

  In the area where the tape is attached, the pixel value becomes smaller due to the influence of the tape. Therefore, when the first condition is satisfied, it can be estimated that the target edge point is generated due to the tape side end. Therefore, when it is determined that the first condition is satisfied (YES in step S2), the remaining edge point extraction unit 2d is caused due to an area having a large pixel value in which the target edge point is not included in the captured image of the valid bill. It is estimated that the edge point is not generated (step S3).

  On the other hand, when the first condition is not satisfied, there is a possibility that the target edge point is not generated due to the tape side end. Therefore, when it is not determined that the first condition is satisfied (NO in step S2), the remaining edge point extraction unit 2d is adjacent to the left side (the left end side of the bill) of the attention edge point The pixel values of three pixels arranged in a row in the direction toward and the pixels of three pixels arranged in a direction from the left end of the bill to the right end adjacent to the right side (right end side of the bill) It is determined whether the second condition that one side of each value is smaller than the above-mentioned “most frequent pixel value” and the other is smaller than the above “most frequent pixel value” plus 5 (step S4).

  In the area where the tape is attached, the pixel value becomes smaller due to the influence of the tape. Therefore, when the second condition is satisfied, it can be estimated that the periphery of the target edge point is a relatively bright background, and the target edge point is generated due to the tape side edge. Therefore, when it is determined that the second condition is satisfied (YES in step S4), the remaining edge point extraction unit 2d is caused due to an area having a large pixel value in which the focused edge point is not included in the captured image of the valid bill. It is estimated that the edge point is not generated (step S3).

  On the other hand, when neither the first condition nor the second condition is satisfied, it is hard to think that the focused edge point is generated due to the tape side edge. Therefore, when it is not determined that the second condition is satisfied (NO in step S4), the remaining edge point extraction unit 2d determines that the region with a large pixel value in which the focused edge point is not included in the captured image of the valid bill. It is estimated that this is an edge point generated due to (step S5).

  Each determination condition used in the above-mentioned step S2 and step S4 is an example to the last, and the threshold etc. which were used in the determination condition are not limited in particular.

  Here, whether or not a thin line-like region having a large pixel value along the short direction of the bill to be discriminated which is not included in the captured image of the valid bill exists in the imaged image of the bill to be discriminated The second estimation process to be estimated will be described with reference to the flowchart of FIG. The second estimation process is performed by the remaining edge point extraction unit 2d.

  The remaining edge point extraction unit 2d obtains the most frequently appearing pixel value of the captured image of the bill to be discriminated (step S11). The process of step S11 is the same as the process of step S1 in the flowchart of FIG. 7 described above.

  Next, the remaining edge point extraction unit 2d sets the average value of the pixel values in the vertical direction (shorter direction of the banknote) to each pixel position in the horizontal direction (longitudinal direction of the banknote) for the captured image of the banknote to be discriminated. It calculates with (step S12). If an example of the calculation result obtained by the process of step S12 is represented by a graph, it will become like FIG.

  Next, the remaining edge point extraction unit 2d determines whether all of the conditions (I), (II) and (III) are satisfied, with the width of 3 pixels in the horizontal direction as the determination target (step S13). . The condition (I) is a condition that at least one of the average values of the three vertical pixel values is larger than the most frequently appearing pixel value in the three horizontal pixels to be determined. The condition (II) is a condition that the average value of the pixel values in the vertical direction which is the left end for three pixels in the horizontal direction to be determined is five or more larger than the average value of the pixel values in the vertical direction next to it. . The condition (III) is a condition that the average value of the vertical pixel values at the right end of the horizontal three pixels for determination is 5 or more larger than the average value of the vertical pixel values on the right thereof. .

  If all of the conditions (I), (II), and (III) are satisfied (YES in step S13), the remaining edge point extraction unit 2d determines the width 3 pixels in the horizontal direction to be determined in step S13. Is estimated to be a bright vertical thin line (step S14). In addition, a bright vertical thin line means a thin linear region having a large pixel value along the short direction of the bill to be discriminated which is not included in the captured image of the valid bill. This bright vertical thin line is generated by a strong crease.

  On the other hand, when at least one of the condition (I), the condition (II) and the condition (III) is not satisfied (NO in step S13), the remaining edge point extraction unit 2d detects the horizontal direction which is the determination target in step S13. It is estimated that the three pixels in width are not bright vertical thin lines (step S15).

  After that, while sliding three pixels in the horizontal direction, which is the determination target in step S13, in the horizontal direction one pixel at a time, the processing after step S13 described above is repeated.

  Then, in the present embodiment, an edge point extracted by the edge point extraction unit 2c is an edge point included in a bright vertical thin line and one pixel adjacent to the bright vertical thin line, that is, five horizontal pixels centered at the bright vertical thin line. It is removed from Thus, an edge point appearing by a thin line-like region having a large pixel value along the short direction of the bill to be discriminated which is not included in the captured image of the valid bill is erroneously regarded as an edge point by the tape side end It is possible to prevent detection.

  Note that each determination condition used in step S13 described above is merely an example, and the threshold value and the like used in the determination condition are not particularly limited.

  Also, the second estimation process described above can be performed independently regardless of the detection of the tape application. When the second estimation process described above is performed alone, it has a function of detecting a strong broken line (strong break).

  Returning to the description of FIG. 2, the analysis unit 2e will be described.

  The analysis unit 2e combines the edge point image indicating the remaining edge points in the A direction range sent from the remaining edge point extraction unit 2d and the edge point image indicating the remaining edge points in the B direction range with respect to the composite image. The density detection filter is applied to detect the density of the remaining edge point in a predetermined size area as a pixel value. When the density of the remaining edge point in the predetermined size area detected as the pixel value exceeds the threshold stored in the storage unit 3 as the analysis reference information 3 d, the analysis unit 2 e detects the tape side edge. In the present embodiment, both the right side tape side end and the left side tape side end are detected. For this reason, even if one of the right side tape side end and the left side tape side end overlaps with the thread portion of the bill, the other side end of the tape can be detected, and the tape sticking can be detected. it can.

  For example, the filter for density detection shown in FIG. 11 can be used, for example, when making a predetermined size region into a vertically long shape, and the filter for detecting density shown in FIG. Can be used. By making the predetermined size area longitudinally elongated, it is possible to detect the tape stuck longitudinally (in the direction of the short side of the sheet), and by making the predetermined size area obliquely inclined, the sheet is inclined and attached The detected tape can be detected.

  Unlike the present embodiment, the tape side edge is detected independently without combining the edge point image indicating the remaining edge point in the A direction range and the edge point image indicating the remaining edge point in the B direction range. You may do it. In this case, the detection of the right side tape side end and the detection of the left side tape side end can be performed separately. For example, even if one of the right side tape side end and the left side tape side end overlaps with the thread portion of the bill, the other side tape end can be detected, and the other side tape side end Since it is possible to grasp which is the right side tape side end or the left side tape side end, it is possible to detect the tape application and to estimate the tape application position.

Second Embodiment
The bill discriminating apparatus 10 according to the first embodiment of the present invention described above extracts edge points and discriminates whether or not a tape is stuck on a bill. However, it is also conceivable that the edges generated by the lateral ends of the applied tape are not sharp. For this reason, it is desirable to be able to detect that the tape is attached to the bill without extracting the edge point.

  Therefore, the bill discriminating device 20 according to the second embodiment of the present invention has a function capable of detecting that the tape is attached to the bill without extracting the edge point. The bill discriminating according to the first embodiment of the present invention The configuration is added to the device 10.

  FIG. 13 is a block diagram showing the configuration of the bill discriminating device 20. As shown in FIG. In FIG. 13, the same parts as those in FIG. 2 are indicated by the same reference numerals and the detailed description is omitted.

  Control unit 2 of banknote discrimination device 20 includes boundary detection unit 2f and determination unit in addition to image data acquisition unit 2a, denomination identification unit 2b, edge point extraction unit 2c, remaining edge point extraction unit 2d, and analysis unit 2e. It has 2g. In addition to the denomination identification template 3a, the various filters 3b, the reference edge point 3c, and the analysis reference information 3d, the storage unit 3 stores detection reference information 3e and judgment reference information 3f.

  The boundary detection unit 2 f uses the image data generated by the image data acquisition unit 2 a and is darker than the first area from the first area in the direction from one end to the other end in the longitudinal direction of the bill to be discriminated A first boundary portion which is a boundary which changes to the second region and a second boundary portion which is a boundary which changes to the fourth region which is brighter than the third region are detected.

  In the following description, one end of the bill to be discriminated in the longitudinal direction is taken as the left end of the bill, and the other end in the longitudinal direction of the bill to be discriminated is taken as the right end of the bill. You can get

  The boundary detection process performed by the boundary detection unit 2 f will be described using the flowchart in FIG. 14.

  First, the boundary detection unit 2f sets a point of interest 4 for two pixels in the horizontal direction in the captured image of the bill to be discriminated, and sets the left area 5 of 4 × 2 pixels adjacent to the left of the point of attention 4. It sets and the right area 6 of 4x2 pixels adjacent to the right side of an attention point is set (step S21). Accordingly, the positional relationship between the point of interest 4, the left side area 5 and the right side area 6 is as shown in FIG.

  Although the number of constituent pixels of each of the point of interest 4, the left side area 5 and the right side area 6 is not particularly limited, it is made to prevent the edge point generated by the symbol of the bill from entering each of the point of attention 4, left side area 5 and right side area 6 There is a need.

  In the present embodiment, the target point 4 is set such that the target point 4 slides in the lateral direction in the vicinity of the upper end where the symbol does not exist, with a bill whose symbol does not exist in the vicinity of the upper end. The vertical position of the point of interest 4 is changed according to the horizontal direction of the bill using the denomination identification template 3a etc., and the point of interest 4, the left area 5 and the right area 6 do not overlap the symbol of the bill. You may

  Next, the boundary detection unit 2f determines whether or not the focused point 4 corresponds to "a first boundary which is a boundary where the first region changes to a second region darker than the first region" (step S22). ). In the determination of step S22, if the left region 5 is a bright region and the right region 6 is a dark region, the focused point 4 corresponds to the first boundary portion. It is preferable that the storage unit 3 store the predetermined value, the determination formula, and the like used in the determination of step S22 as the detection reference information 3e.

  For example, when the average pixel value of the left area 5 is larger than the average pixel value of the right area 6 by a predetermined value or more, it can be determined that the focused point 4 corresponds to the first boundary. Further, for example, when the central pixel value of the left region 5 is larger than the central pixel value of the right region 6 by a predetermined value or more, it can be determined that the focused point 4 corresponds to the first boundary. Also, in consideration of the small unevenness in density and small dispersion of pixel values in the area where the tape is attached, the average pixel value of the left area 5 is larger than the average pixel value of the right area 6 by a predetermined value or more. Even if the center pixel value of the area 5 is larger than the center pixel value of the right side area 6 by a predetermined value or more, the target point 4 is the first boundary when the pixel values of the right side area 6 are dispersed more than the predetermined value. You may determine that it does not correspond.

  When it is determined that the point of interest 4 corresponds to “a first boundary portion that is a boundary that changes from the first region to a second region that is darker than the first region” (YES in step S22), the boundary detection unit 2f Let the point of interest 4 be the first boundary portion (step S23).

  On the other hand, if it is determined that the point of interest 4 does not fall under the “first boundary portion that is the boundary at which the first area changes to a second area darker than the first area” (NO in step S22), the point of interest 4 is It is determined whether it corresponds to "a second boundary portion which is a boundary which changes from the third region to the fourth region which is brighter than the third region" (step S24). If it is determined in step S24 that the left area 5 is a dark area and the right area 6 is a bright area, the focused point 4 corresponds to the second boundary. It is preferable that the storage unit 3 store the predetermined value, the determination formula, and the like used in the determination of step S24 as the detection reference information 3e.

  For example, when the average pixel value of the right area 6 is larger than the average pixel value of the left area 5 by a predetermined value or more, it can be determined that the focused point 4 corresponds to the second boundary. For example, when the central pixel value of the right side area 6 is larger than the central pixel value of the left side area 5 by a predetermined value or more, it can be determined that the focused point 4 corresponds to the second boundary. Also, in consideration of the small unevenness in density and small dispersion of pixel values in the area where the tape is attached, the average pixel value of the right area 6 is larger than the average pixel value of the left area 5 by a predetermined value or more. Even if the center pixel value of the area 6 is larger than the center pixel value of the left area 5 by a predetermined value or more, the focused point 4 is the second boundary when the dispersion of the pixel values of the left area 5 is more than the predetermined value. You may determine that it does not correspond.

  When it is determined that the point of interest 4 corresponds to “the second boundary portion that is a boundary that changes from the third region to the fourth region brighter than the third region” (YES in step S24), the boundary detection unit 2f Let the point of interest 4 be the second boundary (step S25).

  On the other hand, if it is determined that the point of interest 4 does not fall under the “second boundary portion that is the boundary that changes from the third region to the fourth region brighter than the third region” (NO in step S24), the flow ends Do.

  Thereafter, the above-described boundary detection process is repeated while sliding the point of interest 4 in the horizontal direction by one pixel. When the boundary detection process is completed for all the points of interest 4, the boundary detection unit 2f outputs the information of the first boundary and the second boundary determined in step S23 and step S25 to the determination unit 2g.

  Returning to the description of FIG. 13, the determination unit 2 g will be described.

  The determination unit 2g uses the first boundary and the second boundary detected by the boundary detection unit 2f to determine the distance between the first boundary and the second boundary located to the right of the first boundary. Based on the pixel values of the pixels located in the area, it is determined whether the tape is attached to the area between the first boundary and the second boundary located on the right side of the first boundary.

  The determination process performed by the determination unit 2g will be described using the flowchart of FIG.

  First, the determination unit 2g determines, from among all combinations of the first boundary and the second boundary, a combination of the first boundary and the second boundary which can be a combination of the left side tape side end and the right side tape side end. It extracts (Step S31). In this embodiment, the combination of the first boundary and the second boundary located on the right side of the first boundary is extracted, and the first boundary and the second boundary located on the right of the first boundary are further extracted. Extraction is limited to combinations in which the distance between the parts is equal to or greater than the minimum width of the tape to be determined and is equal to or less than the maximum width. As described above, since the combination of the first boundary portion and the second boundary portion is narrowed in consideration of the width of the tape to be determined, it is possible to shorten the determination time and prevent erroneous detection.

  Next, the determination unit 2g focuses on a certain combination from the combinations extracted in step S31, and determines whether or not the portion between the focused first boundary portion and the second boundary portion corresponds to a tape sticking portion ( Step S32). The storage unit 3 may store the predetermined value, the determination formula, and the like used in the determination of step S32 as the determination reference information 3f.

  For example, when the average pixel value of the pixels located in the area between the focused first boundary and the second boundary is less than the predetermined value, the tape is attached between the focused first boundary and the second boundary. It can be determined that it corresponds to the department. Also, for example, when the central pixel value of the pixel located in the area between the focused first border and the second border is less than the predetermined value, the tape between the focused first border and the second border is a tape. It can be determined that it corresponds to the sticking section. Also, for example, when the dispersion of the pixel values of the region between the focused first boundary and the second boundary is less than the predetermined value, the portion between the focused first boundary and the second boundary is a tape attaching unit. It can be determined that it corresponds.

  Here, it is conceivable that the background pixel value of the bill to be discriminated is not uniform. Then, the pixel value distribution when the tape is attached to a place where the background density is not uniform is as shown in FIG. 17. Therefore, if the detection condition of the first boundary and the second boundary is tightened, the first boundary or There is a risk that the second boundary can not be detected. On the other hand, when the detection conditions of the first boundary and the second boundary are relaxed, the first boundary and the pixel value distribution as shown in FIG. The second boundary will be detected.

  Therefore, in this embodiment, in the determination of step S32, the average pixel value of the left region 5 shown in FIG. 15 corresponding to the focused first boundary and the right side shown in FIG. 15 corresponding to the focused second boundary. The sum of the average pixel values of the area 6 is obtained, and a value obtained by subtracting “the average pixel value of the pixels located in the area between the focused first boundary portion and the second boundary portion” is obtained from the sum. When the calculated value is less than the predetermined value, the average pixel value of the pixels located in the area between the focused first boundary portion and the second boundary portion is less than the predetermined value, or the focused first pixel is selected. Even if the central pixel value of the pixel located in the area between the border and the second border is less than the predetermined value, the portion between the focused first border and the second border corresponds to the tape sticking part It may be determined not to.

  When it is determined that the portion between the focused first boundary portion and the second boundary portion corresponds to the tape attaching portion (YES in step S32), the determination unit 2g determines the position between the focused first boundary portion and the second boundary portion. It is set as a tape sticking part (step S33).

  On the other hand, when it is determined that the portion between the focused first boundary portion and the second boundary portion does not correspond to the tape sticking portion (NO in step S32), the flow is ended as it is.

  Thereafter, the first boundary portion and the second boundary portion to which attention is paid are changed, and the determination in step S32 described above is repeated, and the determination in step S32 is performed on all the combinations extracted in step S31.

  Since it is possible to determine whether the tape is stuck without extracting the edge point by the above-mentioned border detection processing and judgment processing, even if the edge generated by the side end of the stuck tape is not clear, the paper It can be detected that the tape is attached to the leaves.

  Also, since the tape sticking portion has a small dispersion of pixel values and stains such as ink bleeding have a large dispersion of pixel values, the determination is performed in consideration of the dispersion of pixel values in any of steps S22, S24 and S32 described above. As a result, it is possible to prevent erroneous detection of contamination such as ink bleeding as sticking of a tape.

  The bill discriminating device 20 has a function of extracting an edge point and detecting that a tape is stuck on a bill, and a function of detecting that a tape is stuck on a bill without extracting an edge point. Although it was an installed device, the function that can detect that the tape is stuck on the bill without extracting the edge point explained in detail in this embodiment is that the edge point is extracted and the tape is stuck on the bill It is also possible to carry out independently regardless of the function that can detect what is being done.

  In addition, the function that can detect that the tape is attached to the bill without extracting the edge points described in detail in this embodiment and the mechanical thickness detection are performed to detect that the tape is attached. You may implement combining with the function which can be performed.

  As a drawback of a thickness sensor that can detect that a tape is attached to a sheet by performing mechanical thickness detection, the front end of the sheet is larger than the original thickness due to the impact when it enters the thickness sensor Since the thickness is detected, there is a problem that it can not be determined whether the tape is attached to the front end of the sheet even if it is not a thin tape.

  Therefore, it is determined whether or not the tape is attached to the front end of the sheet by the function of detecting that the tape is attached to the banknote without extracting the edge points described in detail in this embodiment. The mechanical thickness detection may be performed to determine whether the tape is attached to a portion other than the front end portion of the paper sheet by the function of detecting that the tape is attached.

Reference Signs List 1 image line sensor unit 2 control unit 2a denomination identification unit 2b image data acquisition unit 2c edge point extraction unit 2d residual edge point extraction unit 2e analysis unit 2f boundary detection unit 2g determination unit 3 storage unit 3a denomination identification template 3b filter Group 3c Reference edge point 3d Analysis standard information 3e Detection standard information 3g Judgment standard information 10, 20 Sheet discrimination device

Claims (9)

A sheet discrimination apparatus that determines whether a tape is attached to a sheet based on a captured image of the sheet, the sheet discrimination apparatus comprising:
A storage unit that stores a reference edge point that is an edge point of a density gradient direction range along the longitudinal direction of the valid paper sheet included in a captured image of the valid paper sheet;
An edge point extraction unit for extracting an edge point of a density gradient direction range along the longitudinal direction of the sheet to be discriminated from a captured image of the sheet to be discriminated;
From the edge point extracted by the edge point extraction unit, it is estimated that the edge point corresponding to the reference edge point and the pixel not included in the captured image of the valid paper sheet are generated due to the bright region A remaining edge point extraction unit that removes edge points and extracts remaining edge points;
A paper sheet discrimination apparatus comprising: an analysis unit that analyzes whether a tape is attached to the paper sheet to be discriminated based on the remaining edge point. The reference edge point is
A first reference edge point, which is an edge point of a first density gradient direction range included in a captured image of a valid paper sheet, and an edge of a second density gradient direction range, which is opposite to the first density gradient direction range And a second reference edge point which is a point
The edge point extraction unit
Extracting an edge point of the first density gradient direction range and an edge point of the second density gradient direction range from the captured image of the sheet to be discriminated;
The remaining edge point is
From the edge points of the first density gradient direction range extracted by the edge point extraction unit, edge points corresponding to the first reference edge point and pixels not included in the captured image of the valid paper sheet are bright From the first remaining edge point from which the edge point estimated to be generated due to the region is removed and the edge point of the second density gradient direction range extracted by the edge point extracting unit to the second reference edge point A corresponding edge point and a second remaining edge point from which an edge point estimated to be generated due to a bright region is removed including a corresponding edge point and a pixel not included in the captured image of the valid paper sheet. The sheet discriminating apparatus according to claim 1. The analysis unit
The paper sheet discrimination according to claim 1 or 2, wherein whether or not the tape is stuck to the paper sheet to be discriminated is analyzed based on the density of the remaining edge point in a predetermined size area. apparatus. The remaining edge point extraction unit
In the captured image of the sheet to be discriminated, there is a thin line-like region in which the pixels along the short direction of the sheet to be discriminated are not included in the imaged image of the valid sheet. The edge point located around the said thin wire | line-like area | region is also remove | eliminated from the edge point extracted by the said edge point extraction part, when it is estimated that it has done. Sheet discrimination device as described. The first area is changed from the first area to a second area darker than the first area in the direction from one end to the other end in the longitudinal direction of the sheet to be discriminated in the captured image of the sheet to be discriminated A detection unit that detects a first boundary portion that is a boundary to be moved and a second boundary portion that is a boundary that changes from a third region to a fourth region that is brighter than the third region;
The first boundary portion and the first boundary portion are based on pixel values of pixels located in a region between the first boundary portion and the second boundary portion positioned on the other end side with respect to the first boundary portion. The determination part which determines whether the tape is stuck in the area | region between the said 2nd boundary part located in the said other end side rather than a boundary part, It was provided with any one of the Claims 1-4 characterized by the above-mentioned. The sheet discriminating apparatus according to one aspect. The determination unit is
Only when the distance between the first boundary portion and the second boundary portion located closer to the other end side than the first boundary portion is equal to or greater than the minimum width of the tape to be determined and is equal to or less than the maximum width. ,
The first boundary portion and the first boundary portion are based on pixel values of pixels located in a region between the first boundary portion and the second boundary portion positioned on the other end side with respect to the first boundary portion. The sheet discrimination apparatus according to claim 5, wherein it is determined whether a tape is attached to a region between the boundary portion and the second boundary portion located on the other end side of the boundary portion. The determination unit is
The first boundary portion and the first boundary portion are based on dispersion of pixel values of pixels located in a region between the first boundary portion and the second boundary portion positioned closer to the other end side than the first boundary portion. The paper sheet according to claim 5 or 6, wherein it is determined whether a tape is attached in a region between the first boundary and the second boundary located closer to the other end than the first boundary. Classification device. The determination unit is
In the area between the first boundary and the second boundary located closer to the other end than the first boundary based on the pixel value of the first region and the pixel value of the fourth region The sheet discriminating apparatus according to any one of claims 5 to 7, wherein it is determined whether a tape is attached. A sheet discrimination method applied to a sheet discrimination apparatus for discriminating whether a tape is attached to a sheet based on a captured image of a sheet,
An edge point extracting step of extracting edge points of a density gradient direction range along the longitudinal direction of the sheet to be discriminated from the captured image of the sheet to be discriminated by the sheet discriminating apparatus ;
The edge of the density gradient direction along the longitudinal direction of the valid paper sheet included in the captured image of the valid paper sheet from the edge point extracted by the edge point extraction step by the paper sheet discrimination device An edge point corresponding to a reference edge point that is a point and an edge point estimated to be generated due to a bright region that is not included in the image of the valid paper sheet are extracted to extract a remaining edge point A remaining edge point extraction step;
A paper sheet discrimination method, comprising: an analysis step of analyzing whether or not a tape is attached to the paper sheet to be discriminated based on the remaining edge point.

Images