KR20040021412A - Apparatus for recognizing and counting currency of a plurality of countries and method for the same - Google Patents

Abstract

PURPOSE: A device and a method for recognizing securities of many nations are provided to recognize a kind of securities and count the securities by comparing the pattern information of the selected nation with an image pattern of the inputted securities. CONSTITUTION: A hopper(100) transfers the securities into the device if the securities are detected by a hopper sensor. An automatic feeder(102) automatically separates/feeds the securities in the hopper one by one. A transferor(104) transfers the fed securities to an outlet through a contact image sensor(106) and a magnetic sensor. A selector(114) selects the proper outlet depending on a processing result and discharges the fed securities. The contact image sensor reads an image of the securities by a line unit and stores the image data in an internal memory. A control panel(118) controls the device, shows the processing result, and is connected to a PC or an external device such as a display device for a customer.

Description

Apparatus for recognizing and counting currency of a plurality of countries and method for the same}

The present invention relates to a securities recognition counting device and method of a plurality of countries, and if the user selects a currency of a desired country by storing pattern information about the securities of a plurality of countries in advance, the pattern information of the corresponding country's securities is inputted. The present invention relates to an apparatus and a method for recognizing and counting rolls by comparing image patterns of securities.

Conventional securities counters were only able to process the recognition of securities in one country fixed in one device. In addition, there was no outlet for separating securities that were abnormally processed, ie unrecognizable or ambiguous, among a plurality of securities.

Therefore, when the type of securities changes, the securities recognition system must be changed to suit the characteristics of the securities. Therefore, the inconvenience and cost of installing the securities recognition device of each country is increased and recognition is increased. Improper or ambiguous securities are recognized and have a disadvantage of low reliability.

Accordingly, the present invention has been made in order to solve the above problems, and if the user selects the currency of the desired country by storing the pattern information on the securities of a plurality of countries in advance, the pattern information of the corresponding country securities and the input The purpose of the present invention is to provide a securities recognition counting device and a method of securities in a number of countries that can compare and classify securities image patterns.

Figure 1 shows the internal configuration of the securities recognition counting apparatus according to an embodiment of the present invention.

2 is a block diagram illustrating the recognition operation of the securities recognition counting device according to an embodiment of the present invention.

Figure 3 is a block diagram showing an actual implementation of the securities recognition counting apparatus according to an embodiment of the present invention.

4 is a flowchart illustrating the operation of the main controller of the securities recognition counting device according to an embodiment of the present invention.

5 is a flowchart illustrating an operation of a signal processing unit of a security recognition counting device according to an embodiment of the present invention.

* Description of the symbols for the main parts of the drawings *

100: hopper 102: automatic paper feeder

104: transfer unit 106: contact type image sensor

108: drive unit 110: external connection

112: stacker 114: selector

116: reject pocket 118: operation panel

According to the present invention, when the user selects a currency of a desired country by pressing a currency selection key, the first step of loading pattern information related to the corresponding currency stored in the memory, and detecting the securities in the hopper by the hopper sensor to detect the motor A second step of feeding the securities into the apparatus by driving the second securities; and a third step of detecting the state of the loaded securities to determine whether the conveyance, chains, or jams are detected, and detecting the securities and scanning the image to extract the image. A fourth step of recognizing the denomination by storing and comparing the pattern information with respect to the currency, and a fifth step of counting the number of purchases and the amount of the securities while discharging the securities in which the denomination is recognized in a plurality of discharge pockets according to the recognition result It includes.

The securities recognition counting apparatus of the present invention includes an input unit for currency selection, an automatic feeding unit for separating and feeding a plurality of securities mounted on a hopper into sheets, a driving unit for driving a motor by detecting the securities by a hopper sensor; The control unit detects the status of the loaded securities and determines whether it is in transit, chain, or jam, the memory unit which stores pattern information about the securities of various countries necessary for class recognition, and the securities by detecting the securities. The sensor unit scans the entire image, detects the edge of the security from the image of the security stored in the memory, calculates geometric information such as the position and tilt of the security, and inputs the size of the security and the target currency known in advance. Compare the criteria ranges and establish multiple areas of interest that best represent the pattern of securities in the country. It extracts the feature points, compares the pattern information for each kind and the extracted feature points, and recognizes the final kind as the closest kind and discharges it to the stacker while counting the number of securities purchased and the amount of paper. It has a conveying part for conveying up to.

The securities recognition counting device sets the information related to the currency if the user selects the corresponding national currency by using a currency selection key or similar method before counting the securities. Here, the information related to the call is pattern information of the call, a song, a chain, a jam variable value, a concentration variable for adjusting brightness, a plurality of regions of interest, a feature point, and the like.

In addition, the securities recognition counting device has two central processing units, that is, a main control unit and a signal processing unit. Each of them simultaneously scans the image of the securities, chains and jams, and the entered securities to recognize the paper type.

When the main control unit selects a currency of a desired country by a currency selection key or a similar method, the main control unit sets a variable value for determining whether a song, chain, or jam is suitable for the currency. When the securities are fed, the sensor compares the thickness, size, time passed by each sensor, and the value of the set retransmission, chain, and jam variables to determine the state of the securities.

The signal processor stores in advance a pattern information on the position, size, and thickness of the securities of a number of countries that are required for the securities recognition. When the user selects the currency of the desired country by the currency selection key, the pattern information of the country's security is loaded from the memory, and the security information is recognized by comparing the pattern information of the country's security with the image of the input security.

Hereinafter, an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

Figure 1 shows the internal configuration of the securities recognition counting apparatus according to an embodiment of the present invention.

Hopper (100) is a place to put a plurality of securities, and when the securities in the hopper is detected by the Hopper Sensor (Hopper Sensor) the motor is driven and the securities are introduced into the device. The automatic paper feeder 102 automatically separates and feeds a plurality of securities placed on the hopper. The transfer unit 104 transfers the securities to the discharge port via the contact image sensor 106 and the magnetic sensor. The selector 114 selects and discharges an appropriate outlet according to the processing result of the loaded securities.

As shown in FIG. 1, the securities recognition counting device has two outlets, a stacker 112 and a reject pocket 116. Unrecognized or ambiguous securities are discharged separately into reject pocket 116.

When the hopper sensor recognizes the securities, the motor driving unit 108 drives the motor to feed the securities into the device, and the close-type image sensor 106 reads the images of the securities line by line and stores them as image data in the internal memory. do. The operation panel 118 operates the securities recognition counting device and shows the processing result, and the external connection unit 110 is located at the back of the securities recognition counting device and connects a printer, an external output device such as a customer display device, or a PC. Can be.

2 is a block diagram illustrating the recognition operation of the securities recognition counting device according to an embodiment of the present invention.

In order to recognize the securities, pattern information about the securities of multiple countries and variable values set for the currency of the corresponding countries are stored in memory, and the user selects the currency of the desired country by using a currency selection key or a similar method. Pattern information and variable values of the country's securities are loaded from memory.

When the securities loaded on the hopper of the securities recognition counting device are detected by the hopper 200 sensor, the main controller drives the motor and operates the clutch 202. When the securities are fed, four sensors, up to the outlet, are Reading Point 1 (204), Reading Point 2 (210), Jam Point (216), and Selector Point. 220, the sensor passes. Each sensor should be configured with at least two left / right sides so that it can be recognized regardless of the width or position of the security. The intervals between the sensors are arranged not to exceed the width of the security, and each sensor is positioned between the transfer rollers 208 and serves to monitor the transport status of the security.

The leading point 1 sensor 204 is located in front of the close-type image sensor for scanning the image of the entered security and monitors the inflow of the security. The output of the reading point 1 sensor 204 is connected to the main controller 330 and the signal processor 350 together. When the security is detected in the reading point 1 sensor 204, the close-type image sensor 206 is driven to start an image scan, and when the security passes the reading point 1 sensor 204, the operation stops after a slack time. The scanned analog image signal is converted into a digital signal in preprocessing. At this time, by setting the parameters of the preprocessor according to the securities selected by the user, it is possible to obtain image data having an optimal brightness concentration. Further, by setting an optimal reference value for the securities set by the currency selection key, it is appropriately determined whether there is an abnormality in the feeding state of the securities. In other words, depending on the amount of permeation according to the thickness and the print density, the mid-feed is sensed and the allowable limit of the chain, jam and skew is set according to the size.

The leading point 2 sensor 210 detects the securities like the leading point 1 sensor 204 and monitors the security transfer status.

The jam point sensor 216 monitors the transfer status of the security and the jam in which the security is jammed inside the machine.

The selector point sensor 220 detects a reference time point for driving the swing selector 222 in addition to the basic transfer state monitoring function. According to the recognition result of the securities or the state of the transfer of the securities, a high speed swing selector 222 is driven to select and discharge the securities to the stacker 226 and the reject pocket 224. Change the direction of securities emissions. In this case, if the selector point sensor 220 is not accurately driven, it is difficult to smoothly discharge the securities and damage of the jam or the securities occurs. The length measurement of the security is measured by the pulse signal of the encoder 218, and controls the motor so that the pulse output of the encoder 218 is constant for constant speed operation.

Figure 3 is a block diagram showing an actual implementation of the securities recognition counting apparatus according to an embodiment of the present invention.

As shown, the securities recognition counting device includes a user interface 310, a main control unit 330, a signal processor 350, and a power supply unit 370. In the first central processing unit 331 of the main control unit 330, when a user selects a currency of a desired country by using a currency selection key, a variable value for determining whether the securities are transported, chained or jammed is set. When the securities are fed, they check the thickness, size, and time passing by each sensor to determine the state of the securities by comparing them with the values of the convoy, chain, and jam variables.

Although not shown in the figure, the EEPROM existing in the first central processing unit 331 has an operating speed, an operation level of a hopper, various states that should be preserved in the event of an error, progress and stop of the apparatus, system options and power off. Is saving.

The volatile memory SRAM 333 adjusts the operation state of each sensor, the motor operation state, information on the tilt of the scanned securities image, cumulative number and amount information of the counted volume, reconstruction, chain, jam parameters, and brightness. It stores temporary data necessary for device operation such as concentration variable and variable or processing result.

The second central processing unit 351 of the signal processing unit 350 scans an image of the securities when the start signal is input from the main control unit 330 and the securities are detected by the leading point 1 sensor, and is pre-memorized by the currency selection key. The securities are recognized by comparing the pattern information of the securities loaded in the country with the scanned securities image.

The image signal preprocessor (DIPP) 354 receives the analog output signal of the close-type image sensor, performs a sampling process to compensate for the distortion, converts the signal into a digital signal, and passes the image data via the FIFO 355 to prevent data loss. To the memory.

The Flash ROM 352 stores program code and system options that must be preserved even when the power is turned off, system status, pattern information of securities in multiple countries for securities recognition, multiple areas of interest for each currency, Information such as variable values for determining feature points, mid-songs, chains, jams, data for distortion compensation of image data, and the like are stored.

When the device is powered on or restarted, the program code stored in the flash ROM 352 is transferred to the SDRAM 353 to execute the program. The scanned image data is preprocessed in the image signal preprocessing 354 and then stored in the SDRAM 353 via the FIFO 355.

The PLD 356 has a function of transmitting data between the main controller 330 and the signal processor 350, a resolution conversion function according to a speed difference between the close-type image sensor and the image signal preprocessor 354, and a second central processing apparatus. The data bus width conversion function according to the interface and the data rate difference between the 351 and the image signal preprocessor 354 is implemented.

In addition, the signal processor 350 may view or store additional information such as monitoring the sensor state of the device, downloading the scanned image, upgrading the security recognition program, and managing the device history through the PC interface.

The user interface 310 transmits and receives a user's key input and LCD output data from the main control unit 330, and displays a keypad that the user can input and displays the number of securities purchased and counted amount, which is the processing result of the device. Include.

The input of the power supply unit 370 uses an AC voltage of 85V ~ 265V and the output is a DC voltage of 3.3V, 5V, -5V, 12V, 24V.

In the main control unit 330, if the security is not detected by the reading point 1 sensor within a predetermined time after the security enters the apparatus, the main control unit 330 regards it as a mis-feeding occurrence and processes an error. When the securities are detected by the reading point 1 sensor, the securities are determined to be retransmitted, chained, or jamd by comparing with the preset reference value according to the selected currency.

Whether or not the transmission is determined by comparing the value of the transmission parameters for the selected call preloaded in the memory and the amount of light transmitted through the reading point 1 sensor. For example, when the dollar is fed, if the value of the value of the transmission variable set by the currency selection key is 70 and the amount of light input by the sensor is 50, it is determined as a transmission error.

Whether the chain is determined by comparing the value of the chain variable for the selected currency preloaded in the memory with the size of the security passing through the leading point 1 sensor. For example, if the dollar is fed, if the chain variable value set by the currency selection key is 13 cm and the length of the securities passing through the sensor is 15 cm, it is determined as a chain error.

The jam is determined by comparing the jam variable value for the selected call pre-loaded in memory, the time passing through each sensor such as the leading point 1 sensor, the leading point 2 sensor, the jam point, and the selector point sensor and the passage time between the sensors. do. For example, if the dollar is fed and the reference time through the Reading Point 1 sensor is 5 seconds by the currency selection key and the reference time between the Reading Point 1 sensor and the jam point sensor is set to 3 seconds, the dollar is fed and leading If the time exceeding the point 1 sensor is measured as 7 seconds, it is determined as a jam error.

If all the results are processed normally, the selector 114 is controlled to be discharged to the stacker 112, and if the abnormal paper or securities recognition processing result is processed as a reject, it is discharged to the reject pocket 116. In addition, the main control unit 330 may control the entire securities recognition counting device and serve as a simple counting function.

When a signal is received from the main controller 330, the signal processor 350 scans the entire image of the securities having the optimal image quality and stores the image in the memory. Here, the optimal image quality refers to the optimal density. Since the brightness and color of the securities in each country are different, the image is scanned by setting the density-related variables differently. When a call is selected by a call selection key or a similar method, the signal processor 350 sets a variable related to density corresponding to the call so as to produce an optimum picture quality.

The signal processor 350 detects an edge boundary from the scanned image of the security and calculates geometric information such as the position, the tilt and the size of the security using the edge boundary. It is possible to determine whether or not the primary target currency by using the size information of the corresponding country securities set by the currency selection key.

As such, if the size of the scanned security is within the reference value, it is converted into normalized image data by compensating for tilt and geometric distortion.

The geometric information of the security is used to establish a number of areas of interest that best represent the pattern characteristics of the country's security. These areas of interest are preset by training during the development process. In addition, the predetermined feature point is extracted according to the securities of the corresponding country in the region of interest. For example, if the amount number area of the security is set to multiple areas of interest, the characteristic point of the security is the amount number of the security. In addition, different areas may be set according to the characteristics of the securities of the corresponding country, and the entire area of the security and a plurality of areas of interest may be extracted as feature points.

If the characteristic points of the securities are extracted, the final category is recognized as the closest class by comparing the pattern information of the class of the currency and the characteristics of the extracted securities. Here, the pattern information of the corresponding currency is selected by the currency selection key and the entire image of the security is used, so that various currencies can be recognized.

The integrated judgment is made by integrating the conveyed state of the loaded securities and the result of the paper type recognition, and then separated into the stacker and reject pocket according to the operating conditions. Normally processed securities are discharged to the stacker, counting the number of securities purchased and the amount of paper used, and abnormally processed securities are discharged to the reject pocket.

4 is a flowchart illustrating the operation of the main controller of the securities recognition counting device according to an embodiment of the present invention.

First, when a user selects a currency of a desired country by pressing a currency selection key, a variable for determining whether the securities are retransmitted, chained or measured by the currency selection key is set (S400). When a plurality of securities are placed on a hopper and the hopper sensor detects the securities (S402), the securities are fed into the apparatus by driving a motor (S404). After the motor is driven, if the securities are not detected by the reading point sensor 1 within a predetermined time, an error processing occurs due to an incorrect paper feed (S406). Based on the above results, it is determined whether the securities entered, the chain, or the jam are entered.

Whether or not the transmission is determined by comparing the value of the transmission parameters for the call set by the call selection key and the amount of light transmitted through the reading point 1 sensor (S410). For example, when the dollar is fed, if the value of the feed variable set by the currency selection key is 70 and the amount of light input by the sensor is 50, it is determined as a feed error.

After determining whether or not to send the chain, whether the chain is determined by comparing the chain variable value for the currency set by the currency selection key and the size of the security passing through the leading point 1 sensor (S412). For example, if the dollar is fed, the chain variable value set by the currency selection key is 13 cm, and the size of the direction of the securities passing through the sensor is 15 cm.

After determining whether the chain is determined, the jam is determined by comparing the jam variable value for the call set by the call selection key, the time passing through each sensor such as the leading point 1 sensor, the jam point, the selector point sensor, and the passage time between the sensors. (S414). For example, if the dollar is fed and the reference time through the Reading Point 1 sensor is 5 seconds by the currency selection key and the reference time between the Reading Point 1 sensor and the jam point sensor is set to 3 seconds, the dollar is fed and leading If the time exceeding the Point 1 sensor is measured as 7 seconds, a jam error is determined. If a wrong feed or jam of the securities occurs, the motor is stopped and the brake is operated to stop the driving, and when the heavy feed and the chain are generated, the discharge pocket is discharged (S408).

If the securities of the securities transport, chain, jam is processed as normal, select the discharge direction according to the result of the securities recognition (S416) and controls the swing selector. If the securities are recognized as normal, the securities are discharged to the stacker (S418), and the number and amount of the securities fed are counted and displayed on the display device (S420). If the securities are recognized as abnormal, the securities are discharged to the reject pocket (S422) and the number of rejected securities is counted (S424).

5 is a flowchart illustrating an operation of a signal processor of a security recognition counting device according to an exemplary embodiment of the present invention.

First, when a user selects a currency of a desired country by pressing a currency selection key, geometrical variable values such as pattern information and size of the corresponding currency are loaded from the pre-stored flash memory into the RAM (S500). When the user puts a plurality of securities in the hopper, the hopper sensor detects the securities. Then, the main control unit sends a start signal to the second central processing unit of the signal processing unit and drives the motor to feed the security into the device. The plurality of securities sensed by the hopper sensor are separated and fed into sheets by the automatic feeder. When the securities fed to the device are detected by the reading point 1 sensor (S502), the securities are read in line units by the close-type image sensor and stored as internal memory image data (S504). Using the conventional edge detection method, the edge boundary of the security is detected from the image data stored in the internal memory (S506), and geometric information such as the position and the tilt of the security is calculated (S508). Here, the size of the securities introduced using the geometric information of the securities can be known, and the first judgment is made by comparing with the size reference range of the target currency known in advance (S510). When the size of the securities put in this manner is within the reference range, it is converted into image data normalized to a predetermined size and compensated for tilting or geometric distortion (S512).

The geometric information of the securities is used to set a plurality of areas of interest that best represent the pattern characteristics of the securities of the country (S514). These areas of interest are preset during development and stored in inactive memory. Then, a predetermined feature point is extracted according to the securities of the corresponding country from the set region of interest (S516). Different areas can be set according to the characteristics of the country's securities, and the entire area of the securities and a number of areas of interest can be extracted as feature points. According to the above results, the final category is recognized as the closest class by comparing the pattern information of the class of the currency previously created in the development process and the feature points of the securities extracted (S518). Here, the currency selection key selects and uses the pattern information of the corresponding currency and uses the entire image of the securities, thereby recognizing various currencies.

After performing the recognition for each type, the process proceeds to the stacker discharge process (S520). The integrated judgment is made by integrating the conveyed state of the loaded securities and the result of the paper type recognition, and then separated into the stacker and reject pocket according to the operating conditions. Normally processed securities are discharged to the stacker (S522) to count the number of securities purchased and the type of securities (S524), and abnormally processed securities are discharged to the reject pocket (S526) and the rejected securities are discharged. It counts (S528). After the securities are discharged to the stacker and the reject pocket, it is checked whether the work is finished. If the work is not finished, the reading point 1 sensor returns to the step of detecting the securities (S530).

In the present invention, the state of the securities fed by the detection of the conveyance, chain, jam or not, and scanned the image of the securities, but the image of the securities loaded and the state of the securities by detecting the state of the securities of the transport, chain, jam You can also determine whether or not.

In the above description, the main control unit and the signal processing unit have been described separately, but it is also possible to implement them as one processor.

As described above, the present invention is not limited to the specific embodiments, and those skilled in the art will understand that various changes and modifications can be made without departing from the technical spirit of the present invention.

As described above, according to the present invention, securities of multiple countries can be recognized by one device and can be discharged into a stacker or reject pocket depending on the result. Therefore, the inconvenience of having to separately install the securities recognition device of each country has the effect of reducing the cost.

Claims (15)

When the user selects a currency of a desired country by pressing a call selection key, loading pattern information related to the corresponding currency stored in a memory; A second step of sensing the securities in the hopper by a hopper sensor to drive a motor to feed the securities into the apparatus; A third step of detecting a state of the fed securities and determining whether the paper is conveyed, chained or jammed; Detecting a securities, scanning an image, extracting and storing the image, and comparing the pattern information with respect to the currency to recognize a paper type; The fifth step of counting the number of purchases and the amount of securities while discharging the securities recognized by the paper type into a plurality of discharge pockets according to the recognition result Securities recognition coefficient method comprising a. The method of claim 1, The first step further includes the step of loading the value of the transmission variable for the selected call stored in the memory, The securities recognition coefficient method of claim 3, characterized in that whether or not the heavy transport is determined by comparing the value of the transport parameters and the amount of light transmitted through the security. The method of claim 1, The first step further includes loading a chain variable value for the selected currency stored in the memory, In the third step, whether or not the chain is determined by comparing the value of the chain variable and the size of the securities sensed by the sensor. The method of claim 1, The first step further includes loading a jam variable value for the selected currency stored in the memory, In the third step, whether the jam is determined by comparing the jam variable value, the time that the security passes through each sensor, and the passing time between the sensors is determined. The method of claim 1, In the fourth step, when scanning the image of the security, the securities recognition coefficient method, characterized in that for setting the scan variable according to the selected currency. The method of claim 1, wherein the fourth step Step 4-1 in which the securities are scanned in units of lines by the sensor and stored as image data in the internal memory; Step 4-2 of detecting an edge boundary of the security from the image data stored in the internal memory and calculating geometric information such as the position and the tilt of the security; A step 4-3 comparing the size reference range of the target currency known in advance with the size of the securities injected using the geometric information; Steps 4-4 of setting a plurality of ROIs that best represent the pattern characteristics of the securities of the corresponding country by using the geometric information; Steps 4-5, extracting feature points preset according to the country's securities in the region of interest, and recognizing the final class as the closest class by comparing the pattern information for each class of the currency and the feature points of the extracted securities. Securities recognition coefficient method comprising a. The method of claim 6, wherein steps 4-3 If the size of the securities entered and the target currency in advance are within the standard range, they are converted to image data normalized to a certain size and compensated for tilt or geometric distortion. If the size is less than the reference range securities recognition coefficient method further comprising the step of separating and discharging to the reject pocket. The method of claim 1, The plurality of discharge pockets are a stacker and a reject pocket, and when the recognition of the security is successful, the discharge pocket is discharged to the stacker, and when the recognition fails, the discharge pocket is discharged to the reject pocket. . An input unit for call selection, An automatic paper feeder that separates and feeds a plurality of securities placed on a hopper, A driving unit for driving the motor by detecting the securities of the hopper sensor; A control unit which detects the state of the loaded securities and determines whether or not it is in transit, chains or jams; A memory unit for storing pattern information on securities of a number of countries required to recognize a class; A sensor unit that detects the security and scans the entire image of the security, Detects the edge boundary of the security from the image of the security stored in the memory, calculates geometric information including the location of the security, compares the amount of security entered and the reference range of the target currency in advance, The feature points are extracted by setting multiple areas of interest that best represent the pattern, and the pattern information for each category is compared with the pre-made pattern information and the extracted feature points. A signal processor for counting Transfer section for transferring fed securities to outlet Securities recognition counting device characterized in that it comprises a. The method of claim 9, And a selector unit for separating and discharging the securities into the stacker and the reject pocket according to the processing result of the securities. The method of claim 9, The control of the securities recognition coefficient of the control unit, characterized in that the control by comparing the value of the transmission value and the amount of light transmitted through the securities with respect to the currency set by the currency selection key. The method of claim 9, The control unit determines whether or not the chain by comparing the chain variable value for the currency set by the currency selection key and the size of the securities detected by the sensor. The method of claim 9, The whether or not the control is determined by comparing the jam variable value for the currency set by the currency selection key and the time that the securities pass through each sensor and the transit time between the sensors are determined. The method of claim 9, And the geometric information includes the inclination information of the securities, and corrects the inclination of the securities based on the inclination information before comparing the reference range. The method of claim 9, wherein the signal processing unit If the size of the securities entered and the target currency in advance are within the standard range, they are converted to image data normalized to a certain size and compensated for tilt or geometric distortion. If the size is less than the reference range, securities recognition counting device further comprising the step of separating and discharging to the reject pocket.