JP2009046276A - Paper sheet double feed detection method and paper sheet processing apparatus - Google Patents

Abstract

A multi-feed detection method and a paper sheet capable of accurately determining multi-feed based on a thickness detection value of a check even when the maximum thickness of a check to be conveyed is more than twice the minimum thickness. Propose a similar processing device.
A multi-feed detection system includes a thickness detector for detecting a thickness of a check that is transported on a transport path of a check reading apparatus, and the thickness of the check that passes is determined to be thick. Sampling from the detector 3 is performed at a constant period. The sampled detected thickness change amount Δt is sequentially calculated, and the calculated detected thickness change amount Δt is compared with a predetermined threshold Th2. When the change amount Δt is equal to or greater than the threshold Th2, the check 4 is It is determined that multiple feeds have been made. In order to prevent a decrease in determination accuracy due to the error component of the detection value, a plurality of detection values are sequentially extracted in the order of sampling, and an average value thereof is calculated. Judgment is made.
[Selection] Figure 3

Description

  The present invention relates to a paper sheet multi-feed detection method and a paper sheet processing apparatus for detecting whether or not paper sheets such as checks transported along a transport path are transported in an overlapping state.

  In a paper processing apparatus, for example, a check processing apparatus, a check is separated one by one from a check storage unit in which checks are stored in a bundle, sent out to a transport path, and transported along the transport path. Processing such as reading magnetic character information from a check. In a processing apparatus that performs various processes while transporting paper sheets to be processed, it is necessary to transport paper sheets one by one. For this reason, when the paper sheets are conveyed in an overlapped state, the process is surely detected and the processing is stopped or interrupted.

As a method for detecting double feeding of paper sheets, a thickness detector is arranged on the conveyance path, and the thickness of the paper sheets passing through the detection position of the thickness detector is detected and detected. When the thickness exceeds the thickness of one sheet, a method is generally adopted in which it is determined that double feeding is performed. Patent Documents 1 to 5 disclose such a double feed detection method.
Japanese Patent Laying-Open No. 2005-225661 JP 2005-132599 A JP-A-2005-306491 JP 2005-225659 A JP 2005-225581 A

Here, in the double feed determination based on whether or not the detected thickness exceeds the thickness of one sheet, the maximum thickness Tmax of the sheet to be conveyed is the sheet to be conveyed. It is assumed that the thickness is less than twice the minimum thickness Tmin.
Tmin × 2> Tmax

  In a check processing apparatus that processes checks, the maximum thickness of the checks is 0.14 mm, and the minimum thickness is 0.10 mm. Therefore, the double feed determination is accurately performed by the method disclosed in the above patent document. Is possible.

  However, when handling various types of paper sheets, it is assumed that a sheet having a thickness outside the above conditions is conveyed. For example, when the thickness of a thin paper sheet is 0.075 mm and the thickness of a thick paper sheet is 0.16 mm, the thickness when the thin paper sheets are conveyed in a state of being stacked two sheets. The detected value is 0.15 mm, which is thinner than the thick paper sheet thickness of 0.16 mm. Therefore, in this case, it is not possible to distinguish whether two thin paper sheets are sent in a stack or one thick paper sheet is sent.

  An object of the present invention is to provide a double feed detection method capable of accurately performing double feed determination based on the thickness of a paper sheet even when the maximum thickness of the paper sheet is twice or more the minimum thickness. It is to propose.

In order to solve the above problem, the double feed detection method of the present invention is:
A thickness detector for detecting the thickness of the paper sheet transported along the transport path is disposed at a predetermined position in the transport path,
When the paper sheet passes through the detection position by the thickness detector, the detection value of the thickness detector is sampled at a predetermined cycle,
The amount of change in the sampled detection value is calculated sequentially,
Compare each calculated amount of change with a predetermined threshold,
When the amount of change is equal to or greater than the threshold value, it is determined that the paper sheet is being fed multiple times.

  Here, in order to prevent a decrease in determination accuracy due to an error component of the detection value, the detection values are sequentially extracted in a sampling order, an average value thereof is calculated, and the amount of change in the detection value is calculated. Instead, it is desirable to calculate the amount of change of each calculated average value and perform the multifeed determination based on the amount of change.

  In the double feed detection method of the present invention, the amount of change in thickness when the paper sheet passes the detection position is detected. When the paper sheets are transported in an overlapping manner, generally, the two paper sheets are transported in a state of being shifted forward and backward in the transport direction. Accordingly, the thickness of the paper sheet passing through the detection position in this state is initially one sheet thickness, but changes from the middle to the thickness of two sheets. In the method of the present invention, paying attention to the thickness fluctuation (change amount) caused by such a step when paper sheets are double-fed, it is determined that a double-fed state occurs when the thickness fluctuation occurs. I am doing so. Therefore, even when the precondition of the conventional double feed detection method that the maximum thickness of the paper sheet is less than twice the minimum thickness is not satisfied, the double feed determination can be performed with high accuracy.

  Here, the detection thickness fluctuation point due to the level difference caused by the overlap of the paper sheets is distinguished from the detection thickness fluctuation time point that occurs when the leading edge and the trailing edge of the paper sheet pass the detection position. There is a need.

  The point in time when the leading edge of the paper sheet passes through the detection position of the thickness detector can be detected as follows. First, at a time before the paper sheet reaches the detection position, a zero detection value representing zero thickness is sampled at least once from the thickness detector. The detection value of the thickness detector is sampled at a predetermined period from the detection point of the zero detection value, and the difference between the sampled detection value and the zero detection value is sequentially calculated. It is determined that the time when the difference from the zero detection value is equal to or greater than a predetermined value is the time when the leading edge of the paper sheet in the transport direction has passed the detection position. Then, this tip passage time or a point after a predetermined number of sampling cycles from the tip passage time is defined as a judgment start time for multifeed judgment, and each detection value or each sampled after this judgment start time The double-feed determination of the paper sheets may be performed based on the change amount calculated from the average value.

  In this case, in order to increase the detection accuracy of the tip passage time, the zero detection value of the thickness detector is sampled a plurality of times, and the tip passage time is determined using an average value of these zero detection values. desirable.

  Next, when the trailing edge of the paper sheet passes through the detection position of the thickness detector, it can be detected as follows. First, the detection value of the thickness detector is sampled over a predetermined sampling period from the sampling point of the zero detection value. The final detection value of the thickness detector sampled at the end of the sampling period is compared with the zero detection value. If the difference between the final detection value and the zero detection value is less than or equal to a predetermined value, it is determined that the trailing edge of the paper sheet in the transport direction has passed through the detection position at the end of the sampling period. .

  In this case, it is possible to know the time when the trailing edge of the paper sheet has passed by detecting the time when the detected thickness returns to zero. That is, when it is determined that the trailing edge of the paper sheet has passed, the difference between the final detection value and each detection value sampled before the sampling time of the final detection value is sequentially determined. The time point when the difference from the final detection value exceeds a predetermined value is determined to be the time point when the rear end of the paper sheet has passed through the detection position. Then, the rear end passage time, or a time point that is backed by a plurality of sampling cycles from the rear end passage time is determined as the determination end time of the multifeed determination, and from the determination start time to the determination end time. It is possible to determine whether or not the paper sheet is to be fed based on the amount of change in each detected value or each average value sampled.

  When the sampling period is short, the trailing edge of the paper sheet may not be able to finish passing the detection position during that time. For example, when the amount of overlap between two sheets is small and the sheets are sent in a long state, it is assumed that the trailing edge cannot pass through the detection position during the sampling period. In this case, since the detected thickness does not return to zero, it can be determined that the trailing edge of the paper sheet has not passed through the detected position based on the detected thickness.

  In this case, the end point of the sampling period in which the final detection value is sampled, or the sampling point that goes back by a predetermined sampling period from the end point is determined as the determination end point of the multifeed determination, Based on the amount of change of each detected value or each average value sampled from the determination start time to the determination end time, the multi-feed determination of the paper sheets can be performed.

  Next, when the multi-feed is performed with the leading edge of the paper sheet slightly deviated back and forth in the transport direction, it is accurately determined that the rear end of the paper sheet is slightly deviated back and forth. It is necessary to detect this, and it is desirable to frequently perform multi-feed determination on the front end side and the rear end side of the paper sheet at short intervals.

In order to accurately detect the case where the leading edge side of the paper sheet is slightly fed back and forth, in the leading edge side multi-feed determination based on the predetermined number k of detection values obtained sequentially from the determination start time. Is
A set of 2 (m + n) detection values (m and n are integers) sequentially obtained from the determination start time is incremented by “1” from a predetermined initial value, and 2 (m + n) = Extract several sets until k
For each set, calculate the average value of (m + n) detection values in the first half and the average value of (m + n) detection values in the second half, and calculate the difference F between these average values,
When the difference F is equal to or greater than the threshold value Tha, it is determined that the paper sheet has been double fed.

For example, when k = 50, m = 5, and the initial value of n = 0,
5 by 10 by 6 by 12 by 7 by 14
・
・
A set of 21 detection values of 50 by 25 is extracted, a difference F of detection values in each set is calculated, and it is determined whether or not the difference F is greater than or equal to a threshold value. If there is a difference in the level of the paper sheet between or near the extracted detection value groups, the difference F is equal to or greater than the threshold value, and thus the difference is detected. In other words, it can be seen that the paper sheet is in a double feed state.

  Here, if the leading edge of the paper sheet is chipped, torn, or has a hole, the thickness detection value fluctuates when passing through the detection position, as in the case where the step portion passes. To do. When there is a chip, tear, hole or the like, the thickness detection value fluctuates to the zero thickness side. Therefore, it is possible to detect such passage of chips, breaks, holes, and the like as follows and prevent these passages from being mistakenly recognized as having passed through the stepped portion.

  That is, for the k detection values extracted in the front end side multi-feed determination, the 2 (m + n) detection values of each set are divided into m pieces in the order opposite to the sampling order, and the fractional number is less than m. The detection value of the minute is ignored, and each average value am of the m detection values is calculated. When the difference between each average value am and the zero detection value is equal to or smaller than a predetermined value, that is, when the thickness after fluctuation indicates zero thickness, it can be determined that a chip, a tear, a hole, or the like has passed. Therefore, the tip-side double feed determination based on the 2 (m + n) detection values is invalidated.

  In addition, when the leading edge of the paper sheet is bent and doubled, the calculated difference F changes in the direction of decreasing thickness. If the change direction of the difference F is a direction in which the thickness becomes thin so that the detection value fluctuation due to bending or the like is not erroneously detected as double feeding, the double feeding determination by the difference F is invalidated. do it.

Next, in order to accurately detect the case where the rear end side of the paper sheet is slightly shifted back and forth, the predetermined number k of detection values obtained sequentially from the end of the determination. In the rear end side double feed judgment based on
A set of 2 (m + n) detection values (m and n are integers) obtained retroactively from the determination end time is increased by “1” from a predetermined initial value, and 2 (m + n) ) = K is extracted until it becomes k,
For each set, calculate the average value of (m + n) detection values in the first half and the average value of (m + n) detection values in the second half, and calculate the difference F between these average values,
When the difference F is equal to or greater than the threshold value, it is determined that the paper sheets are double fed.

  Even in this case, in order not to misidentify the passage of a chip, tear, hole or the like attached to the rear end portion of the paper sheet as the step portion has passed, the rear end side double feed determination is performed. 2 (m + n) detection values in each set are divided into m pieces in the sampling order, and the detection values corresponding to fractions less than m are ignored, and each m detection values are detected. When each average value am of the values is calculated and the trailing edge of the paper sheet has passed the detection position, the difference between each average value am and the final detection value is equal to or less than a predetermined value. May invalidate the result of the rear end side double feed determination based on the 2 (m + n) detection values.

  Here, when the trailing edge of the paper sheet has not passed through the detection position, if the difference between each average value am and the zero detection value is equal to or less than a predetermined value, 2 (m + n) The result of the rear end side double feed determination based on the detected values may be invalidated.

  Further, when the rear end portion of the paper sheet is bent and doubled, the calculated difference F changes in the direction of increasing the thickness. When the change direction of the difference F is a direction in which the thickness is increased so that the detection value fluctuation due to bending or the like is not erroneously detected as double feed, the double feed determination based on the difference F is invalidated. do it.

  Next, in the case where there is a step at the intermediate portion in the conveyance direction of the paper sheet, detection values having a constant thickness are continuously obtained before and after the step. Therefore, even if steps are detected at wide intervals, steps can be detected with high accuracy. In addition, there are many cases where large folds, creases, etc. are attached to the middle part of the front and back of the paper sheets. If a step is detected at a narrow interval, the large creases and creases are mistaken as steps, and the detection accuracy decreases. There is also a possibility to do. Therefore, it is desirable to identify the step in the middle part of the paper as follows.

  That is, the detection value group included from the determination start time point to the determination end time point is extracted every m pieces in succession according to the sampling order, and detection values for fractions not reaching k are ignored. To do. The k detected values are divided into the first half k / 2 and the second half k / 2, the respective average values are calculated, and the difference F between the average values is calculated. When the difference F is equal to or greater than the threshold value, it can be determined that the step has passed, and therefore, it can be determined that the paper sheet has been double-fed.

  Even in this case, if the paper sheet is torn and has a hole or the like, the thickness detection value greatly fluctuates when they pass through the detection position, so it may be mistaken for a step to pass. There is. Therefore, when the difference between each detection value included in each of the k detection value groups and the zero detection value is equal to or smaller than a predetermined value, that is, when the detection thickness indicates zero, It is desirable to determine that a torn hole or the like has passed and invalidate the double feed determination based on the difference F between the average values of the k detection value groups.

  Next, in the paper sheet double feed detection method of the present invention, the thickness detector employs a detection value having a linear relationship with respect to the detected thickness, and prior to double feed detection, A detection value when the thickness is zero in the thickness detector is read as an initial detection value Vo, and a detection value when a reference gauge having a known thickness t is passed through the detection position of the thickness detector is as follows: It is read as the reference thickness detection value Vt, and the initial sensitivity of the thickness detector is set to (Vo−Vt) / t. In this way, the sensitivity is measured for each individual, and in actual double-feed detection of the paper sheet, based on the minimum thickness of the paper sheet to be transported, a thickness serving as a reference for multi-feed determination A variation threshold t2 may be set, and the threshold may be set to {(Vo−Vt) / t} × t2. In this way, it is possible to correct variations in sensitivity due to the usage environment, changes in the thickness detector over time, etc., and to detect the thickness with high accuracy. Sending judgment can be performed.

  Here, as the initial detection value Vo and the reference thickness detection value Vt, it is desirable to employ an average value of a plurality of detection values.

  Further, a displacement member that is displaced according to the thickness of the paper sheet to be transported is disposed at the detection position of the transport path, and the displacement amount of the displacement member is detected by the thickness detector. Also good.

  In this case, as the displacement member, a displacement part that is displaced by a paper sheet that passes through the detection position is formed at one end, and a detection part that is displaced by an amplified displacement amount is formed at the other end. It is preferable to employ a moving lever and detect the amount of displacement of the detection unit by the thickness detector because the detection sensitivity can be increased.

  The thickness detector may be an optical sensor such as a semiconductor position detector.

Moreover, the paper sheet processing apparatus of the present invention uses the above-mentioned paper sheet multi-feed detection method,
A paper sheet insertion portion for inserting paper sheets, a transport path for transporting the paper sheets inserted into the paper sheet insertion section, and collecting the paper sheets that have passed through the transport path It has a collection part, a conveyance mechanism which conveys the paper sheets along the conveyance path, and the thickness detector arranged on the conveyance path.

  With such a paper sheet processing apparatus, even when the maximum thickness of the paper sheet is twice or more the minimum thickness, it is possible to accurately determine the double feed based on the thickness of the paper sheet.

  In the paper sheet double feed detection method of the present invention, a level difference that occurs when paper sheets are double fed based on the detection value of the thickness detector that detects the thickness of the paper sheet passing through the conveyance path. Is detected, and if a step is detected, it is determined that the sheets are being fed in a multiple manner. Therefore, even if the precondition for the multi-feed determination based on the conventional thickness that the maximum thickness of the paper sheet must be less than twice the minimum thickness is not satisfied, the multi-feed of the paper sheet The state can be detected with high accuracy.

  Hereinafter, an embodiment of a paper sheet double feed detection method to which the present invention is applied will be described with reference to the drawings.

  FIG. 1A is a schematic configuration diagram showing the main part of a check double feed detection system according to the present embodiment, and FIG. 1B is a schematic configuration diagram showing its thickness detector. The double feed detection system 1 includes, for example, a length detector 2 (hereinafter referred to as an LD detector 2) disposed in the conveyance path 101 of the check reading apparatus 100, and a downstream of the LD detector 2 in the conveyance path 101. A thickness detector 3 disposed at a position on the side, and a multifeed determination circuit 5 that performs multifeed determination of the check 4 based on the outputs of the LD detector 2 and the thickness detector 3. .

  The double feed determination circuit 5 supplies a drive pulse signal having a predetermined duty ratio and cycle to the thickness detector 3 via the light emission pulse control circuit 6 to drive the thickness detector 3. The detection signal of the thickness detector 3 is converted into, for example, a 10-bit digital value via a signal processing circuit 8 including an A / D converter 7 and then supplied to the double feed determination circuit 5. The double feed determination circuit 5 is a microprocessor unit mainly composed of a CPU, a ROM, and a RAM, reads the detection value of the thickness detector 3 at a constant sampling period, and based on the read detection value, the conveyance path 101. The double feed determination is made as to whether or not the check 4 that passes through is double fed.

  As shown in FIG. 1B, the thickness detector 3 includes an LED 11, a one-dimensional semiconductor position detection element 12, an LED drive circuit 13, and a signal processing circuit 14. The detection light from the LED 11 irradiates the detection surface 16 via the light projection lens 15. The reflected light from the detection surface 16 converges on the light receiving surface 12a of the one-dimensional semiconductor position detecting element 12 via the light receiving lens 17. The one-dimensional semiconductor position detecting element 12 outputs a voltage internally divided according to the light receiving position of the reflected light from both ends, and detects the light receiving position of the reflected light based on the difference between the potentials at both ends.

  In this example, the detection surface 16 is the rear end surface of the pressing lever 103 for pressing the check 4 against the magnetic sensitive surface 102a of the magnetic head for reading magnetic ink characters (hereinafter referred to as MICR head) 102 in the check reading device 100. Is formed. The holding lever 103 rotates about the rotation center shaft 105 according to the thickness of the check 4 passing between the tip surface 103 a and the MICR head 102. Since the distance from the rotation center axis 105 to the detection surface 16 is longer than the distance from the rotation center axis 105 to the tip surface 103a, the displacement of the tip surface 103a is amplified and transmitted to the detection surface 16. The detection surface 16 is displaced toward and away from the thickness detector 3, and the irradiation position of the reflected light on the light receiving surface 12a of the one-dimensional semiconductor position detection element 12 is moved in accordance with the amount of displacement. Therefore, based on the detection signal from the one-dimensional semiconductor position detecting element 12, the thickness of the check 4 passing through the MICR head 102, which is the detection position of the thickness detector 3, can be known.

  FIG. 2 is an explanatory diagram showing a sampling (reading) start point T (1) and a sampling (reading) cycle of the detection value of the thickness detector 3 by the double feed determination circuit 5. In the double feed determination circuit 5, a time T (1) after the check is conveyed by a specified amount from the time T (0) when the tip 4a of the check 4 conveyed on the conveyance path 101 is detected by the LD detector 2. ) Sampling of the detection value of the thickness detector 3 is started. In the check reading apparatus 100 of this example, a stepping motor (not shown) is used as a check conveyance motor, and the check is conveyed by a check conveyance stepping motor (not shown) by a specified number of steps (N1 step). Sampling of the detection value of the thickness detector 3 is started from time T (1) after the conveyance.

  From this sampling start point T (1), the LED 11 is caused to emit light step by step in a cycle of 4 steps, and the detection value is sampled in the light emission interval. That is, the LED 11 is caused to emit light for a time Δ corresponding to one step feeding, and the detection value of the thickness detector 3 is read during this time. The reading is performed a plurality of times during a certain period ΔT from the time when the LED light emission start is delayed by an unstable region (reading Delay), and the read values are averaged, and this is used as a detection value for one sampling point. take in. Such a sampling operation is repeated at a cycle TS of 4 steps. For example, one step is a length corresponding to 0.282 mm (90 dpi).

  The double feed determination circuit 5 repeats reading of the detection value of the thickness detector 3 in the sampling cycle of 4 steps as described above, and performs double feed detection of the check 4 passing through the transport path 101. The double feed determination circuit 5 of this example has three types of modes as double feed detection control.

  FIG. 3 is an explanatory diagram showing three types of double feed detection control executed in the double feed determination circuit 5. As shown in FIG. 3A, the first mode is a mode in which a double feed is determined when the detected thickness t of the check 4 exceeds a predetermined threshold value Th1. In this mode, it is a precondition for detection that the maximum thickness of the check 4 to be conveyed is less than twice the minimum thickness. In this case, detection is difficult or impossible when the amount of overlap of the check 4 is small.

  In the second mode, as shown in FIG. 3B, when the length L of the check 4 detected by the LD detector 2 exceeds a predetermined threshold value L (Th), double feeding is performed. This is a mode for determining that there is. The threshold value L (Th) is set to a value exceeding the maximum length of the check to be conveyed.

  The third mode is according to the present invention. As shown in FIG. 3C, the thickness variation Δt is detected from the detected thickness t of the check 4, and the thickness variation Δt exceeds a certain threshold Th2. In other words, this is a mode in which it is determined that a double feed is made when a step 4b of the check 4 conveyed in an overlapping manner is detected. Also in this case, detection is difficult or impossible when the amount of overlap of the check 4 is small.

  In the double feed determination circuit 5 of this example, the double feed detection of the check 4 is performed by a mode switching operation, for example, by a combination of the first and second modes or a combination of the third and second modes. Yes. That is, when it is known that the precondition that the maximum thickness of the check 4 to be read in the check reading apparatus 100 is less than twice the minimum thickness is satisfied, double feed detection is performed by combining the first and second modes. Can be done. When such a thickness precondition is not guaranteed as in the case of reading various checks, double feed detection may be performed by a combination of the third and second modes.

(Set initial sensitivity of thickness detector)
The detected value of the thickness detector 3 of the double feed detection system 1 has a linear relationship with the thickness of the check 4. In this example, in order to correct variations in sensitivity due to individual differences, assembly errors, etc., the output of the thickness detector 3 after the thickness detector 3 is assembled to the transport path 101, for example, before shipping. The sensitivity is measured, and the measured output sensitivity is set in the built-in memory of the double feed determination circuit 5 or the like.

  In the measurement of output sensitivity, first, the detection value when the thickness is zero in the thickness detector 3 is read as the initial detection value Vo. Next, a detection value when a reference gauge having a known thickness t is passed through the detection position of the thickness detector 3 is read as a reference thickness detection value Vt. From the read value, (Vo−Vt) / t which is the initial sensitivity of the thickness detector 3 is calculated.

  In this example, in order to remove measurement errors and the like, the detection value of the thickness detector 3 is read a plurality of times, for example, for 5 sampling points without a check, and the average value of these detection values is Vo. . Even when the reference gauge is passed, the detected values are read a plurality of times, for example, for 5 sampling points, the average value of these detected values is set to Vt, and the initial sensitivity is calculated and set in the built-in memory or the like. Yes. For example, the thickness t of the reference gauge is 0.20 mm.

(Setting of threshold values Th1, Th2, and L (Th) for double feed detection)
The thresholds Th1, Th2, and L (Th) for detecting double feed by the double feed detection system 1 can be set as follows.

  First, the threshold value Th1 in the first mode (FIG. 3A) is obtained by the following equation using the initial sensitivity of the thickness detector 3.

    Th1 = {(Vo−Vt) / t} × t1

  The thickness t1 is a paper thickness threshold value. For example, when the maximum thickness of the check 4 to be transported is 0.14 mm and the minimum thickness is 0.10 mm, a value between them is 0.17 mm.

  Similarly, the thickness variation threshold Th2 in the third mode (FIG. 3C) is set as follows. First, based on the minimum thickness of the check 4 to be conveyed, a thickness fluctuation value t2 is set as a reference for double feed determination, and the output value of the thickness detector 3 corresponding to this thickness fluctuation value t2 It is set as a threshold Th2 for sending determination. Using the initial sensitivity of the thickness detector 3, this threshold value Th2 is obtained by the following equation.

    Th2 = {(Vo−Vt) / t} × t2

  For example, when the minimum thickness of the check 4 to be conveyed is 0.075 mm, the thickness fluctuation value t2 is set to, for example, 0.05 mm in consideration of measurement errors and the like.

  Note that the length threshold L (Th) in the second mode (FIG. 3B) is set to a value that exceeds the maximum length of the check 4 to be conveyed.

(Double feed detection control: 3rd mode)
Next, FIG. 4 is a schematic flowchart showing a double feed detection operation in the third mode (FIG. 3C) of the double feed detection system 1. FIG. 5 is an explanatory diagram showing an example of an output waveform obtained from the thickness detector 3 when the check 4 (1) and the check 4 (2) overlap each other and passes through the detection position. On the other hand, the check 4 (2) is in a double feed state in which the check 4 (2) is shifted backward by ΔL1 on the leading end side and is shifted backward by ΔL2 on the rear end side.

  Referring to these drawings, when the checks 4 (1) and 4 (2) are conveyed along the conveyance path 101 in an overlapped state, the tip 4 a of the check 4 (1) is moved to the LD detector 2. (Step ST1 in FIG. 4). Sampling of the detection value (AD conversion value) a of the thickness detector 3 is started from a point after the checks 4 (1) and 4 (2) are conveyed by a predetermined amount (the number of N1 steps) (FIG. 4). Steps ST2 and 3, time T (1) in FIG. This sampling start point T (1) is, for example, a point in time when the tip 4a of the check 4 (1) has reached a predetermined distance from the detection position of the thickness detector 3 (reading position of the MICR head 102).

  Sampling is performed every time the check is conveyed for four steps, as described with reference to FIG. When the feed amount for one step is 0.282 mm (90 dpi), the detection value a is sampled every four steps (1.128 mm feed). The end point of the sampling period is a time point after the trailing end of the check having the maximum length to be conveyed has passed the detection position, and at this time point, the sampling of the detection value is ended (step ST4 in FIG. 4, FIG. 4). 5 time T (r)).

  Thereafter, the sampled detection values a (1) to a (r) (r: positive integer) are processed (step ST5), and the sampling points T (1) to T (T) included in the sampling period are processed. From r), a determination start point and a determination end point for the multifeed determination are determined (steps ST6 and ST7 in FIG. 4, time points T (p) and T (q) in FIG. 5).

  Thereafter, each detection value a (p) to a (q) (p, q: positive integer) obtained from the determination start point T (p) to the determination end point T (q) is extracted, The extracted detection value is subjected to calculation processing, leading edge processing (step ST8 in FIG. 4, section T (p) to T (p + 49) in FIG. 5), middle of the check 4 on the leading edge side of the check 4 Intermediate processing (step ST9 in FIG. 4, section T (p) to T (q) in FIG. 5) and rear end processing for multifeed determination on the rear end side of the check 4 (step ST9 in FIG. 4, section T (p) to T (q)) Step ST10 in FIG. 4 and sections T (q-49) to T (q)) in FIG. 5 are executed to determine whether or not the check 4 is in a double feed state (step ST11 in FIG. 4).

  Below, the processing content in each said step is demonstrated.

(Step ST7: Determination of determination start point)
The point in time when the leading edge 4a of the double feed check 4 (1) passes through the detection position of the thickness detector 3 is detected as follows.

  Referring to FIG. 5, from the time point before the check 4 (1) arrives at the detection position, that is, from the sampled detection value a, 5 sampling points in the sampling order from the sampling start point T (1). Minute detection values (AD conversion values) a (1), a (2), a (3), a (4), and a (5) are extracted, and an average value thereof is calculated. This average value is determined to be a zero detection value D representing the case where the paper thickness is zero.

  The zero detection value D and the detection values a (6), a (7), a (8)... After 6 sampling points are sequentially compared, and the difference (variation width) between them is calculated. A sampling point at which this difference exceeds a predetermined value is determined to be a point in time when the tip 4a of the check 4 has passed.

  For example, as shown in FIG. 6A, when the difference between the detected value a (p-5) and the detected value a (p-4) exceeds a predetermined value, the detected value a (p-4) ) Is determined to be the time when the tip 4a passes. A sampling point after a predetermined number of times from the time when the tip passes, for example, a point T (p) (detected value a (p)) after 5 sampling points including the point is set as a determination start point. For the difference (variation range) serving as a determination criterion, for example, a detection value variation corresponding to a thickness of 0.03 mm is employed.

  Note that the tip passage time T (p-4) can be used as the determination start point as it is. In this example, since the tip 4a of the check 4 (1) presses the holding lever 103, the time after 5 sampling points, which is the time after which the chattering of the holding lever 103 converges and the detected value becomes stable, is determined. Judgment start point.

(Step ST7: Determination of determination end point)
The point in time when the rear end 4b of the check 4 (2) in the double feed state has passed the detection position of the thickness detector 3 is detected as follows.

  First, with reference to FIG. 5, among the sampled detection values a (1) to a (r), five sampling points are traced back from the sampling end point T (r) in reverse to the sampling order. Detection values a (r), a (r-1), a (r-2), a (r-3), and a (r-4) are extracted, and an average value of these is calculated. This average value is defined as a final detection value E.

  Thereafter, the calculated final detection value E is compared with the zero detection value D. When the rear end 4c of the check 4 has passed the detection position at the sampling end time T (r), the detection value of the thickness detector 3 is a value representing zero thickness. Therefore, when the difference between the final detection value E and the zero detection value D is equal to or smaller than a predetermined value considering the measurement error, the detection end point T (r) passes the detection position of the rear end 4c of the check 4 (2). It can be judged that it is finished.

  In this case, it is possible to know the time when the rear end 4c of the check 4 passes by detecting the time when the detected thickness returns to zero. That is, the final detection value E and the detection values a (r-5), a (r-6) at the sampling points T (r-5), T (r-6), T (r-7). a (r-7)... are sequentially retrospectively compared, and the time when the difference exceeds a predetermined value for the first time after the trailing end 4c of the check 4 (2) has passed the detection position. It can be determined that the end passage time T (q + 4). Also in the case of the rear end detection, for example, it can be determined that the rear end 4c has passed when a detection variation corresponding to a thickness of 0.03 mm occurs.

  In this example, as shown in FIG. 6B, a predetermined number of sampling points from the rear end passage time T (q + 4), for example, a sampling point T (q) that goes back by five sampling points including the rear end passage time. Is set as the judgment end point. This is because, similarly to the case where the determination start point T (p) is set, the time point when the pressure lever 103 returns to the static state before the displacement is determined is determined as the determination end point.

  On the other hand, when the amount of overlap between the two checks 4 (1) and 4 (2) is small and sent in a long state, the trailing edge 4c is detected at the sampling end point T (r). It is assumed that it cannot finish passing. In this case, the detected thickness does not return to zero. Therefore, when the final detection value E calculated as described above is greatly different from the zero detection value D, it can be determined that the rear end 4c of the check 4 (2) has not passed the detection position. .

  In this case, the sampling end point T (r) or a sampling point T (r-4) pointed back from the end point by a predetermined sampling point, for example, five sampling points, is determined for double feed determination. The end point can be determined.

(Step ST8: tip processing)
The double feed determination process in the tip side portion of the check 4 of this example is performed as follows. First, detection values a (p), a (p + 1), a (p + 2)... A (p + k) of a predetermined number k (here, k is an integer of 4 or more) in the order of sampling from the determination start point T (p). ) (P: positive integer). Next, a set of 2 (m + n) detection values (here, m is an integer equal to or greater than 2 and n is an integer equal to or greater than 0) is set to n in advance from the extracted k detection values. Multiple sets are extracted from the initial value by “1” until 2 (m + n) = k. For each set, the average value of the first half (m + n) detection values and the second half (m + n) are extracted. ) The average value is calculated, and the difference F between these average values is calculated.

In this example, k = 50, m = 5, and the initial value of n = 0. Therefore, as shown in FIG. 7, from the detected values a (p) to a (p + 49) at 50 sampling points T (p) to T (p + 49),
5 pairs each: a (p) to a (p + 4), a (p + 5) to a (p + 9)
6 pairs each: a (p) to a (p + 5), a (p + 6) to a (p + 11)
7 pairs each: a (p) to a (p + 6), a (p + 7) to a (p + 13)
・
・
・
25 pairs each: a (p) -a (p + 24), a (p + 25) -a (p + 49)
21 sets of are extracted. And the difference F (1) -F (21) of the detected value in each group is calculated, and it is discriminate | determined whether each difference is more than threshold value Th2. When the step 4b of the check 4 is formed between or in the vicinity of the extracted detection value groups, at least one of the differences F (1) to F (21) is equal to or greater than the threshold Th2. A step 4b is detected. In other words, it can be seen that the check 4 is in a double feed state.

  For example, in FIG. 7, the difference F (14) is obtained when the check 4 (1), 4 (2) has a set of 14 pieces before and after the step 4b of the checks 4 (1) and the detected value greatly change. 10) greatly exceeds the threshold value Th2, and it can be seen that it is a double feed state.

  Here, if the check 4 (1), 4 (2) is chipped, torn, or has a hole or the like, when the check 4 (1) and 4 (2) pass through the detection position, the same as when the step 4b passes. The thickness detection value a varies. When there is a chip, tear, hole, etc., the thickness detection value a varies to the zero thickness side. Therefore, it is possible to detect such passage of chips, breaks, holes, and the like as follows and prevent these passages from being mistakenly recognized as having passed through the stepped portion.

  That is, with respect to the extracted k detection values, 2 (m + n) detection values of each set are divided into m pieces, for example, in an order opposite to the sampling order, and detection values for fractions less than m pieces are obtained. Is ignored, and each average value am of the m detection values is calculated. When the difference between each average value am and the zero detection value D is equal to or less than a predetermined value in consideration of an error, that is, when the thickness after fluctuation indicates zero thickness, chipping, tearing, holes, etc. have passed It can be judged that.

  Referring to FIG. 8, similarly to the case of FIG. 7, when k = 50 and m = 5, it is assumed that the hole 4d at the tip of the check 4 (1) is vacant. In this case, in the set of 14 detection values a (p) to a (p + 13) and a (p + 14) to a (p + 27), the difference F (10) exceeds the threshold Th2 by the hole 4d, and double feed is performed. It is judged that there is.

  Therefore, the 28 detection values are divided into five (= m) in five in the reverse order of the sampling order, and the three detection values a (p + 2) to a (p) which are fractions are ignored. For each of the five detection values, an average value am is calculated and compared with the zero detection value D. When there is substantially no difference, the average value indicates zero thickness, so that it can be determined that the detected value fluctuates due to the passage of a chip, a tear, a hole, or the like. The average value am of the detection values corresponding to the positions of the holes 4d is an average value indicating zero thickness. Therefore, even if a determination result indicating double feeding is output based on a set of 14 detection values, the determination result may be ignored (it may be invalidated). Thereby, the level difference 4b and the hole 4d can be distinguished.

  On the other hand, when the check 4 (1) is conveyed in a state where the leading end portion of the check 4 (1) is bent and doubled, the detection value greatly fluctuates when the bent portion passes and is erroneously determined to be double feeding. there is a possibility. Therefore, in this example, when the calculated change direction of each difference F is a direction in which the check thickness is reduced, the double feed determination is invalidated based on the difference.

  Referring to FIG. 9, as in FIG. 7, when k = 50 and m = 5, the tip portion of check 4 (1) is folded and overlapped, and the thickness is doubled. Shall. In this case, in each of the 14 detection value sets a (p) to a (p + 13) and a (p + 14) to a (p + 27), the change in the thickness between the overlapping portion 4e and the thin portion behind it. Therefore, the difference F (10) exceeds the threshold Th2 and is determined to be double feeding. In this case, the difference F (10) changes from the average value of the 14 detection values on the first half side in the direction in which the check value becomes thinner in the average value of the 14 detection values on the second half side. Even if the difference F (10) exceeds the threshold value Th2, the direction of change is the direction in which the check thickness decreases, so the double feed determination based on the difference F (10) is invalidated. As a result, it is possible to prevent erroneous determination caused by the overlapping portion 4e that is bent and doubled.

(Step ST10: rear end processing)
Next, in order to accurately detect the case where the trailing edge of the check 4 (2) is slightly fed back and forth, double feed judgment is performed in the same manner as shown in FIGS. Is called.

  Referring to FIG. 10, in this case, a predetermined number k (k is determined from the sampled detection values a (p) to a (q) sequentially from the determination end point T (q). The detected value of an integer of 4 or more is extracted.

  A set of 2 (m + n) detection values (m is an integer equal to or greater than 2 and n is an integer equal to or greater than 1) obtained from these k detection values sequentially from the determination end point is set to n in advance. Multiple sets are extracted from the initial value by “1” until 2 (m + n) = k. For each set, the average value of the first half (m + n) detection values and the second half (m + n) are extracted. ) The average value is calculated, and the difference F between these average values is calculated. If the difference F is equal to or greater than the threshold Th2, it is determined that the check has been double-fed. Specifically, it is the same as the tip processing described above.

In this example, k = 50, m = 5, and the initial value of n = 0. Therefore, as shown in FIG. 10, from the detected values a (q) to a (q-49) of 50 sampling points T (q) to T (q-49),
5 pairs each: a (q) to a (p-4), a (q-5) to a (q-9)
6 pairs each: a (q) to a (q-5), a (q-6) to a (q-11)
7 groups each: a (q) to a (q-6), a (q-7) to a (q-13)
・
・
・
25 pairs each: a (q) to a (q-24), a (q-25) to a (q-49)
21 sets of are extracted. And the difference F1 (1) -F1 (21) of the detected value in each group is calculated, and it is discriminate | determined whether each difference is more than threshold value Th2. When the step 4b of the check 4 is formed between or in the vicinity of the extracted detection value groups of each set, the difference is equal to or greater than the threshold Th2, and thus the step is detected. In other words, it can be seen that the check 4 is in a double feed state. For example, in FIG. 10, the difference F1 (10) is obtained when the check 4 (1), the step 4b1 of the check 4 (2) passes and the detection value largely changes and the 14 values are formed at the front and back. ) Greatly exceeds the threshold value Th2, indicating that it is in a double feed state.

  Here, when the check 4 (1), 4 (2) has a chipped portion, a torn portion, a hole, or the like at the rear end portion thereof, the same as when the step 4b1 passes when passing through the detection position. Further, the thickness detection value a varies. When there is a chip, tear, hole, etc., the thickness detection value a varies to the zero thickness side. Therefore, as in the case of the tip processing, it is only necessary to detect passages such as chippings, tears, holes, and the like and to prevent these passages from being mistakenly recognized as having passed through the stepped portion.

  That is, with respect to the extracted k detection values, 2 (m + n) detection values of each set are divided into, for example, m in order of sampling, and detection values for fractions less than m are ignored, and each m Each average value am of the detected values is calculated. When the difference between each average value am and the final detection value E is equal to or less than a predetermined value in consideration of an error, that is, when the thickness after fluctuation indicates zero thickness, a chip, tear, hole, etc. has passed It can be judged that.

  Here, when the trailing end 4c of the check 4 (2) has not passed the detection position, it is determined whether or not each average value am and the zero detection value D are substantially the same. In some cases, it may be determined that there is a chipped, torn, hole or the like in the overlapping check, and the result of the rear end side double feed determination based on the 2 (m + n) detection values may be invalidated.

  On the other hand, when the check 4 (2) is folded and double-folded, the detected value greatly fluctuates when the bent portion passes and is erroneously determined to be double feed. There is a possibility. Therefore, in this example, the same processing as when the tip portion is bent is performed, and when the calculated change direction of each difference F is a direction in which the check thickness increases, the difference is calculated based on the difference. The double feed judgment is disabled.

(Step ST9: Intermediate processing)
Next, when steps 4b and 4b1 are provided at the intermediate portion in the conveyance direction of checks 4 (1) and 4 (2), detection values having a constant thickness are continuously obtained before and after the steps. Therefore, even if steps are detected at wide intervals, the steps can be detected with high accuracy, and the calculation processing time can be reduced. In addition, there are many cases where a large crease, a crease or the like is attached to the middle part in the front-rear direction of the checks 4 (1), 4 (2). There is also a possibility that the detection accuracy may be lowered. Therefore, the steps in the middle part of the check are identified as follows.

  First, the detection values a (p) to a (q) included from the determination start point T (p) to the determination end point T (q) are sequentially extracted in every m pieces in the order of sampling according to the sampling order. However, the detection values for fractions that do not reach k are ignored. For example, 50 consecutive detection values (k = 50) are extracted every 5 (= m). That is, 50 detection values a (p) to a (p + 49), a (p + 54) to a (p + 104), a (p + 109) to a (p + 159). Any fraction less than 50 on the determination end point T (q) side is ignored.

  Then, 50 detection values are divided into the first 25 (k / 2) and the second 25 (k / 2), the average values am1 and am2 are calculated, and the difference F between the average values is calculated. . When the difference F is greater than or equal to the threshold Th2, it can be determined that the steps 4b and 4b1 have passed, and it is determined that the check 4 has been double-fed.

  Even in this case, if the check 4 is torn or has a hole or the like, the thickness detection value greatly fluctuates when they pass the detection position. is there. Therefore, if there is substantially no difference between each of the 50 detected values (= k) and the zero detected value D, that is, if the detected thickness indicates zero, it is broken. It is determined that a hole or the like has passed, and the multifeed determination based on the difference F between the average values of the 50 detected values is invalidated.

(Mode 2: Double feed detection method by length)
In the above-described double feed detection method, the sampling interval of the detection value is a four-step cycle as shown in FIG. 2, and is 1.128 mm every when converted into a conveyance distance. Further, the minimum value of the number of sampling points for detecting the double feed is 10. Accordingly, when the amount of overlap of the check 4 is less than about 12 mm (= 1.128 mm × 10 points), double feed detection cannot be performed.

  Therefore, in this example, the length of the check 4 being transported is detected based on the detection signal of the LD detector 2, and the detected length L exceeds the maximum length of the check 4 to be transported. Is determined to be in a double feed state. That is, when the detection length L exceeds the threshold value L (Th), it is determined that the double feed state is set.

(Mode 1: Double feed detection method by thickness)
The double feed detection method based on the thickness in this example will be briefly described.

  First, the detection value of the thickness detector 3 without the check 4 is read. That is, detection values a (1) to a (5) for five sampling points are extracted from the sampled detection values in the order of sampling from the sampling start point T (1), and an average value thereof is calculated. This average value is determined to be a zero detection value A representing the case where the paper thickness is zero. This is the same as the zero detection value D in the third mode.

Thereafter, the sampled detection value a (r) is sequentially compared with the zero detection value A in the order of sampling, and the difference C is calculated.
C = A−a (r)

  It is determined whether or not the calculated difference C exceeds a threshold Th1.

  When the state in which the difference C exceeds the threshold Th1 continues for a plurality of points, for example, 25 points, it is determined that the state is a double feed state. The reason for discriminating over a plurality of points is to avoid false detection when the check is folded or wrinkled, thereby improving detection accuracy.

  Further, the check double feed detection method described above can also be used in a paper sheet processing apparatus. This paper sheet processing apparatus communicates with a paper sheet insertion section that inserts and sets a check in a bundle, and communicates with the paper sheet insertion section. The communication unit includes a collection unit that collects checks and a conveyance mechanism that conveys paper sheets along a conveyance path.

  On the conveyance path, the above-described thickness detector 3 is arranged, and a magnetic head for reading magnetic character ink printed on the check and respective scanners for reading the images on the front and back sides of the check are arranged.

  This sheet processing apparatus has a control unit including a CPU, a memory, and the like, and is configured to realize the above-described check double feed detection method. If it is determined that the control unit is in the double feed state, the control unit stops the conveyance and performs error processing. An error is displayed on a display unit such as a display or an LED, and the connected host computer is notified of the error state.

It is a schematic block diagram of the principal part of the double feed detection system and thickness detector to which this invention is applied. It is explanatory drawing which shows the sampling period and sampling start point of a detected value. It is explanatory drawing which shows the 1st-3rd mode for double feed determination. It is a schematic flowchart which shows a double feed determination flow. It is explanatory drawing which shows double feed determination operation | movement. It is explanatory drawing which shows the determination method of a determination start and a determination end point. It is explanatory drawing which shows a front-end | tip process. It is explanatory drawing which shows the front-end | tip process in case there exists a check hole. It is explanatory drawing which shows the front-end | tip process in case a check is bent. It is explanatory drawing which shows a rear end process.

Explanation of symbols

1 Double feed detection system, 2 LD detector, 3 thickness detector, 4, 4 (1), 4 (2) check, 4a tip, 4b, 4b1 step, 4c rear end, 4d hole, 4e overlap due to bending Part 5, double feed determination circuit, 6 light emission pulse control circuit, 7 A / D converter, 8 signal processing circuit, 11 LED, 12 semiconductor position detection element, 13 LED drive circuit, 14 signal processing circuit, 15 light projection lens, 16 detection surface, 17 light receiving lens, Th2 threshold value, D zero detection value, E final detection value, a (1) to a (r) detection value, T (1) to T (r) sampling point, T (p) determination Start point, T (q) Judgment end point

Claims (21)

A thickness detector for detecting the thickness of the paper sheet transported along the transport path is disposed at a predetermined position in the transport path,
When the paper sheet passes through the detection position by this thickness detector, the detection value of the thickness detector is sampled at a predetermined cycle,
The amount of change in the sampled detection value is calculated sequentially,
Compare each calculated amount of change with a predetermined threshold,
When the amount of change is equal to or greater than the threshold value, it is determined that the paper sheet is being double fed. In the paper sheet double feed detection method according to claim 1,
The detection values are sequentially extracted in order of sampling, and the average value thereof is calculated.
Instead of the change amount of the detected value, calculate the change amount of each calculated average value,
A double-feed detection method for paper sheets, wherein the multi-feed determination is performed based on a change amount of the average value. In the paper sheet multifeed detection method according to claim 1 or 2,
Sampling a zero detection value representing zero thickness from the thickness detector at least once before the paper sheet reaches the detection position;
The detection value of the thickness detector is sampled at a predetermined period from the detection point of this zero detection value,
The difference between the sampled detection value and the zero detection value is calculated sequentially,
Determining that the difference between the zero detection value and the predetermined value is a predetermined value or more, and that the leading edge of the paper sheet in the transport direction has passed the detection position;
This tip passage time, or a time after a predetermined number of sampling cycles from the tip passage time, is determined as the determination start time,
A double-feed detection method for paper sheets, wherein the multi-feed determination of the paper sheets is performed based on the change amount calculated from each detection value or each average value sampled after the determination start time. In the paper sheet multifeed detection method according to claim 3,
Sampling the zero detection value multiple times,
A method for detecting the double feed of paper sheets, wherein the leading edge passage time is determined using an average value of these zero detection values. In the paper sheet multifeed detection method according to claim 3 or 4,
Sampling of the detection value of the thickness detector over a predetermined sampling period from the sampling point of the zero detection value,
Comparing the final detection value of the thickness detector sampled at the end of the sampling period with the zero detection value;
If the difference between the final detection value and the zero detection value is less than or equal to a predetermined value, it is determined that the trailing edge of the sheet transport direction has passed through the detection position at the end of the sampling period. If not, it is determined that the trailing edge of the paper sheet has not passed through the detection position. In the paper sheet double feed detection method according to claim 5,
If it is determined that the trailing edge of the paper sheet has passed,
Calculate the difference between the final detection value and each detection value sampled before the sampling time of the final detection value in order,
Determining when the difference from the final detection value exceeds a predetermined value is the rear end passage time when the rear end of the paper sheet has passed the detection position,
The time when the rear end has passed or the time point that has been traced back by a plurality of sampling cycles from the time when the rear end has passed is determined as the determination end time,
A double-feed detection method for paper sheets, wherein multi-feed determination of the paper sheets is performed based on a change amount of each detection value or each average value sampled from the determination start time to the determination end time. In the paper sheet double feed detection method according to claim 5,
If it is determined that the trailing edge of the paper sheet has not passed,
The end time of the sampling period in which the final detection value is sampled, or a sampling time point that is a predetermined sampling period from the end time point is determined as the determination end time point,
A double-feed detection method for paper sheets, wherein multi-feed determination of the paper sheets is performed based on a change amount of each detection value or each average value sampled from the determination start time to the determination end time. In the paper sheet multi-feed detection method according to claim 6 or 7,
In the tip side double feed determination based on the predetermined number k of detection values obtained sequentially from the determination start time point,
A set of 2 (m + n) detection values (m and n are integers) sequentially obtained from the determination start time is incremented by “1” from a predetermined initial value, and 2 (m + n) = Extract several sets until k
For each set, calculate the average value of (m + n) detection values in the first half and the average value of (m + n) detection values in the second half, and calculate the difference between these average values.
A method for detecting double feeding of paper sheets, wherein when at least one of the calculated differences is equal to or greater than the threshold value, it is determined that the paper sheets are double fed. In the paper sheet multi-feed detection method according to claim 8,
About k detection values extracted in the front end side double feed determination,
Dividing 2 (m + n) detection values of each group into m pieces in the reverse order of the sampling order, ignoring the detection values for fractions less than m,
Each average value am of m detection values is calculated,
When the difference between each average value am and the zero detection value is equal to or less than a predetermined value, the leading edge double feed determination based on the 2 (m + n) detection values is invalidated. Sending detection method. In the paper sheet multi-feed detection method according to claim 8,
When the calculated difference is a difference in a direction in which the detected thickness decreases, the tip side double feed determination based on the 2 (m + n) detected values for which the difference is calculated is invalidated. A method for detecting the double feed of paper sheets. In the paper sheet multi-feed detection method according to claim 6 or 7,
In the rear end side double feed determination based on the predetermined number k of detection values obtained sequentially from the determination end time,
A set of 2 (m + n) detection values (m and n are integers) obtained retroactively from the determination end time is increased by “1” from a predetermined initial value, and 2 (m + n) ) = K is extracted until it becomes k,
For each set, calculate the average value of (m + n) detection values in the first half and the average value of (m + n) detection values in the second half, and calculate the difference between these average values.
A method for detecting double feeding of paper sheets, wherein when at least one of the calculated differences is equal to or greater than the threshold value, it is determined that the paper sheets are double fed. In the paper sheet double feed detection method according to claim 11,
About k detection values extracted in the rear end side double feed determination,
Divide 2 (m + n) detection values of each group into m pieces in the sampling order, and ignore detection values for fractions less than m,
Each average value am of m detection values is calculated,
When the trailing edge of the paper sheet has passed the detection position, if the difference between each average value am and the final detection value is equal to or less than a predetermined value, the 2 (m + n) detections Invalidate the result of the rear end side double feed judgment based on the value,
When the trailing edge of the paper sheet has not yet passed through the detection position, the difference between each average value am and the zero detection value is equal to or less than a predetermined value, and the 2 (m + n) detections. A method for detecting a double feed of paper sheets, wherein the result of the determination of the double feed on the rear end side based on the value is invalidated. In the paper sheet double feed detection method according to claim 11,
When the calculated difference is a difference in a direction in which the detected thickness increases, the tip-side double feed determination based on the 2 (m + n) detected values for which the difference is calculated is invalidated. A method for detecting the double feed of paper sheets. In the paper sheet multi-feed detection method according to claim 6 or 7,
The detection value group included from the determination start time point to the determination end time point is extracted in units of m in succession in accordance with the sampling order, and the detection values for fractions not reaching k are ignored,
Each k detection value is divided into the first half k / 2 and the second half k / 2, and the average value of each is calculated.
Calculate the difference between each average value,
A method for detecting double feeding of paper sheets, wherein when at least one of the calculated differences is equal to or greater than the threshold value, it is determined that the paper sheets are double fed. In the paper sheet double feed detection method according to claim 14,
When the difference between each detection value included in each of the k detection value groups and the zero detection value is equal to or less than a predetermined value, the difference is based on the difference between the average values of the k detection value groups. A method of detecting a double feed of paper sheets, wherein the double feed determination is invalidated. In the paper sheet double feed detection method according to claim 1,
As the thickness detector, a detector whose detection value has a linear relationship with respect to the detection thickness is adopted,
Prior to double feed detection, the detection value when the thickness is zero in the thickness detector is read as the initial detection value Vo,
A detection value when a reference gauge having a known thickness t is passed through the detection position of the thickness detector is read as a reference thickness detection value Vt,
The initial sensitivity of the thickness detector is set to (Vo−Vt) / t;
In the double feed detection of the paper sheet,
Based on the minimum thickness of the paper sheet to be transported, a thickness variation threshold t2 that is a reference for multi-feed determination is set,
The method for detecting double feeding of paper sheets, wherein the threshold value Th is set to {(Vo−Vt) / t} × t2. In the paper sheet double feed detection method according to claim 16,
The paper sheet multi-feed detection method, wherein an average value of a plurality of detection values is employed as the initial detection value Vo and the reference thickness detection value Vt, respectively. The paper sheet double feed detection method according to claim 17,
A displacement member that is displaced according to the thickness of the paper sheet to be transported is disposed at the detection position of the transport path,
A method for detecting a double feed of paper sheets, wherein the thickness detector detects a displacement amount of the displacement member. The paper sheet double feed detection method according to claim 18,
As the displacing member, a rotating lever in which a displacing portion that is displaced by a paper sheet passing through the detection position is formed at one end and a detecting portion that is displaced by an amplified displacement amount is formed at the other end is employed. ,
A method for detecting a double feed of paper sheets, wherein the thickness detector detects a displacement amount of the detection unit. In the paper sheet double feed detection method according to claim 19,
An optical sensor such as a semiconductor position detector is adopted as the thickness detector. A paper sheet processing apparatus using the method for detecting double feeding of paper sheets according to any one of claims 1 to 20,
A paper sheet insertion section for inserting paper sheets to be transported;
A transport path through which the paper sheets inserted into the paper sheet insertion portion are transported;
A collection unit for collecting the paper sheets that have passed through the conveyance path;
A transport mechanism for transporting the paper sheets along the transport path;
A sheet processing apparatus comprising the thickness detector disposed on the conveyance path.

Images