KR101387875B1 - Apparatus for media recognition and method for media access direction estimating with the same - Google Patents

Abstract

The present invention relates to a media recognition apparatus for estimating the direction of introduction of media by differentiating a media image and a method of estimating the direction of media entry using the same. The present invention includes an image sensor for scanning a media image, and a differential average calculation unit for calculating a differential average for each divided region by equally dividing the scanned image into a predetermined number. In addition, an input direction estimation unit for comparing the calculated differential averages for each region in a predetermined order by a pre-programmed input direction estimation algorithm is provided and estimates the direction of entry of the medium according to the result. Accordingly, the present invention has the advantage that it is possible to quickly estimate the inlet direction of the medium based on a simple comparison result of the derivative mean.

Media Recognition, Direction Estimation, Differential, DSP

Description

A medium recognition device and a method for estimating a medium retraction direction using the same {APPARATUS FOR MEDIA RECOGNITION AND METHOD FOR MEDIA ACCESS DIRECTION ESTIMATING WITH THE SAME}

1 is a flow chart illustrating a method for estimating a bill insertion direction in a general bill recognition apparatus.

Figure 2 is a block diagram of a bill recognition device according to an embodiment of the present invention.

FIG. 3 is a flowchart illustrating a method of estimating a bill input direction using the bill recognition device shown in FIG. 2.

4 to 7 are bills images for explaining the bill injecting direction estimation method shown in Figure 3 and its derivative mean table.

DESCRIPTION OF THE REFERENCE NUMERALS to the main parts of the drawings "

10: LED 12: Image Sensor

14: transfer module 16: tilt correction

18: Banknote image extraction unit 20: Banknote size estimation unit

22: differential average calculation unit 24: inlet direction estimation unit

26 control unit 28 memory

The present invention relates to a media recognition device, and more particularly, to a media recognition device for estimating the direction of inflow of a medium using a differential mean for each region of a media image, and a method of estimating the direction of a medium inflow using the same.

The term " medium " used in this specification refers to, for example, banknotes, checks, tickets, certificates, etc., and may be various in thickness as compared with width or length.

The media recognition device is applied to various types of financial automatic devices and media handling devices such as banknote recognizers, vending machines, money exchangers, etc. to recognize magnetic, images, silver, fluorescent ink, various numbers and letters printed on the media, and to recognize the types and forgeries of media. It is determined whether it is normal or not. For this purpose, the medium recognition device must first estimate the direction in which the medium is drawn.

For example, FIG. 1 is a step-by-step method for estimating a bill insertion direction performed by a conventional bill recognition apparatus.

In the first step S1 of FIG. 1, the banknote recognizing apparatus scans an image of the banknote using an image sensor. At this time, the scanned image is generally formed larger than the bill size in order to prevent only a part of the bill image is scanned by the shaking or vibration that may occur in the transfer operation. Accordingly, the scanned image includes a bill image and a margin image of the surroundings.

In the second step S2, the banknote recognizing apparatus determines whether the banknotes are aligned, and if there is a tilt, corrects the tilt by the angle, and extracts only the banknote image excluding the peripheral margin in the third step S3.

Subsequently, the boundary line of the banknote image is extracted to extract boundary line feature values (S4). The boundary feature values refer to a boundary line direction, a disadvantage, a branch point, a line value, and the like, divided by a predetermined number. The boundary feature values thus extracted are compared with the values stored in the database (S6) to estimate the pulling direction of the bill (S7).

However, the bill recognition apparatus as described above scans an image to extract the feature values, converts the image into a filtered black-and-white image, and generates a binary code corresponding to the digital image by digital filtering. Determine the outline. In addition, the feature values are extracted through complex processing such as quantization and binary conversion of vector tablets and coordinates.

As described above, the conventional banknote recognizing apparatus has a problem that the algorithm for extracting feature values necessary for estimating the bill retracting direction is complicated and the processing steps are large and the execution speed is slow.

In addition, since the boundary line feature of the image is used, the vibration of the banknote transfer process, the noise of the circuit elements, the output change of the light emitting diode (Light Emitting Diode) that irradiates light on the banknote, and the sensitivity change of the image sensor are used. weak.

In addition, a large storage capacity is consumed because a database for storing the feature value of the bill is required.

Accordingly, an object of the present invention is to provide a medium recognizing apparatus for minimizing a time required for estimating a retracting direction of a medium and a method for estimating a retraction direction using the same.

Another object of the present invention is to provide a medium recognizing apparatus for improving accuracy in estimating a medium retracting direction and a method for estimating a medium retracting direction using the same.

According to a feature of the present invention for achieving the object as described above, the present invention comprises an image sensor unit for scanning an image for the medium to be introduced from the outside; A differential average calculation unit for dividing the image scanned by the image sensor unit into a predetermined number of regions, and calculating a differential average of the divided region-specific images; And a drawing direction estimating unit for estimating a drawing direction of the medium with reference to the calculated differential mean.

According to an embodiment of the present invention, a tilt correction unit for selectively correcting an incoming medium has a tilt, a media image extractor extracting only a media image in a state where the tilt of the medium is aligned, and a media size from the extracted media image. It is preferable that the medium size further comprises an estimation unit.

The medium size estimating unit may determine the type of medium by estimating the vertical length of the extracted medium image.

Preferably, the medium size estimating unit determines a new priesthood and an old priesthood according to the type of the medium.

The inflow direction estimating unit preferably uses an inflow direction estimation algorithm for estimating the inflow direction according to the front / rear and upright / inverted states of the medium by comparing the differential averages of the images for each region.

The estimation algorithm continuously compares and estimates only the differential average of two image images corresponding to each of the derivative averages of the image for each region in order to determine the front / rear and upright / inverted states of the medium. It is preferable.

The differential average calculation unit preferably divides the region where the differential averages are differently provided in two or more.

The present invention may further include a memory for storing a reference differential average for each area according to the retraction direction for each medium, and the retraction direction estimating unit compares the differential average for each area of the retracted medium with the reference differential average. It is desirable to estimate the entry direction of the medium.

According to another aspect of the present invention, the present invention provides a method for scanning an image of an incoming medium, dividing an image of the scanned medium into a predetermined number of regions, and calculating a differential average of the divided regions of the image. And estimating a retraction direction of the medium with reference to the calculated differential mean.

According to the present invention, if the incoming medium has an inclination, selectively correcting it, extracting only the medium image with the inclination of the medium aligned, and estimating the size of the medium from the extracted medium image. It is preferred to further comprise a step.

In the size estimating step, it is preferable to determine the type of the medium by estimating the vertical length of the extracted medium image.

In the size estimating step, it is preferable to determine the priesthood and old priesthood according to the type of the medium.

In the pulling direction estimating step, it is preferable to use a pulling direction estimation algorithm for estimating the pulling direction according to the front / back side and the upright / vertical state of the medium by comparing the differential averages of the images for each region.

In order to determine the front / rear and upright / inverted states of the medium, the retraction direction estimation algorithm continuously compares the derivative averages of two image images corresponding to each other among the derivative averages of the image for each region. It is desirable to estimate.

In the region dividing step, it is preferable to divide the region into two or more with respect to the region where the differential mean is provided differently.

In the pulling direction estimating step, the reference differential mean for each region according to the entry direction for each medium is stored, and the differential direction for each region of the inserted medium is compared with the reference differential mean to estimate the inlet direction of the medium. desirable.

The present invention having the configuration as described above simply scans the media image and easily estimates the retracting direction of the banknote using a pre-programmed retrieval direction algorithm according to the differential mean characteristic for each area divided into a predetermined number. Accordingly, the present invention improves the processing speed delayed by a complicated algorithm, and minimizes the influence of noise, LED output change and image sensor sensitivity change of circuit elements.

Hereinafter, with reference to the accompanying drawings will be described in detail a banknote recognition device and a method of estimating the medium direction using the same according to an embodiment of the present invention.

Although the present embodiment has been described using 10,000 won bills, it can be applied to various media recognition devices that recognize media printed with specific images, such as checks, gift certificates, and tickets, as well as other types of bills such as 1,000 won bills and 5,000 won bills. It should be noted.

2 is a block diagram of a banknote recognition device according to an embodiment of the present invention.

In FIG. 2, the banknote recognition device according to the present invention includes a light emitting diode (Light Emitting Diode, hereinafter referred to as “LED”) that irradiates light onto the banknotes transferred in a predetermined direction (for example, a transverse direction) by the transfer module 14 (10). And an image sensor 12 for scanning an image of the banknote reflected by the irradiated light. In this case, the image sensor 12 scans an area larger than the bill specification in case the bill is not drawn in a straight state due to the vibration of the transfer module 14. Accordingly, the scanned image includes an image of the bill and a margin image. The image sensor 12 preferably uses a charge-coupled device or a contact image sensor.

When the banknotes scanned by the image sensor 12 are not aligned horizontally and there is a skew, a tilt correction unit 16 is provided to correct the tilt. This is because the tilt may occur when the banknote to be transferred by the vibration of the transfer module 14 occurs. When the banknote is transferred in a horizontal state and no inclination occurs, it is obvious that the inclination correcting unit 16 does not perform the inclination correction operation.

A banknote image extracting unit 18 is provided to extract only the banknote image by removing the margin image from the scanned image of the banknote which is tilted or transferred in a horizontal state.

Then, the banknote size estimation unit 20 for estimating the size of the banknote from the extracted banknote image is provided. The banknote size estimation unit 20 is for judging the winding type of the current banknote according to the estimated banknote size. In the present embodiment, a description of the 10,000 won bill is used. The banknote size estimation unit 20 is used to determine the new bill and the old bill of which the recent standard has been changed.

The banknote image is divided equally into a predetermined number (for example, 12 equal parts) to estimate the retracting direction of the banknote, and six regions (a, b, c, d, e, f) located in the left and right regions of the divided regions. A derivative mean calculation unit 22 for calculating a derivative mean for the is provided. Since the number of the divided regions and the derivative average calculating region is a number determined according to the differential average characteristic of the thousand won bill, it is obvious that the number of divided regions and the differential average calculation region is changed according to the characteristics of the medium.

In addition, a drawing direction estimating unit 24 for comparing the calculated differential mean for each region in order by a drawing direction estimation algorithm and estimating the drawing direction of the bill according to the result is provided. The retraction direction estimation algorithm is pre-programmed so that the retrieval direction of the paper money can be estimated according to the comparison result by setting the comparison order in advance by using the characteristics of the image differential averages according to the paper types of the paper money. Stored in memory 28.

In addition, the LED 10, the image sensor 12, the transfer module 14, the tilt correction unit 16, the banknote image extraction unit 18, the banknote size estimating unit 20 to estimate the pulling direction of the banknote ), A derivative average calculating section 22, a drawing direction estimating section 24, and a control section 26 for controlling the memory 28 are provided. The control unit 26 preferably controls the whole bill recognition apparatus so as to perform the recognition operation of the bill, judging the kind of paper money, whether the forgery is normal, whether the banknote is normal, and processing accordingly. To this end, the control unit 26 may be configured by changing and applying various types of control units such as a microprocessor, a central control unit, or a digital signal processor according to operating characteristics. Can be.

The memory 28 stores the banknote size data, various data for each banknote including a differential mean for each region divided into a predetermined number, and the retraction direction estimation algorithm. The memory 28 may use a main memory device for storing various programs and data for driving the automated teller machine including the banknote recognition device or other media handling device.

Next, with reference to Figures 3 to 7 will be described a bill entry direction estimation method using a bill recognition apparatus according to a preferred embodiment of the present invention.

3 is a flowchart illustrating a step-by-step method for estimating a bill insertion direction using the bill recognition apparatus shown in FIG. 2. 4A to 7A illustrate a drawing direction of a banknote for explaining a drawing direction estimating method shown in FIG. 3, and a differential average table for each region according to the drawing direction is illustrated in FIGS. 4B to 7B. When the inlet direction of Figs. 4A to 7A is inverted, the order of the derivative mean tables of Figs. 4B to 7B is the same as the reverse order, and thus the derivative mean table in the inverted direction is omitted.

When the banknote is inserted and the transfer by the transfer module 14 is started, in accordance with the control of the control unit 26 in the tenth step (S10) of Figure 3 LED 10 is irradiated with light to the incoming banknote, The image sensor 12 scans an image of the banknote reflected by the irradiated light. In this case, the scanned image includes a bill image and a margin image.

In the twelfth step S12, the tilt correction unit 16 corrects the tilt when a banknote is inclined while the banknote is drawn in an oblique state according to a transport state such as vibration of the transport module 14. If the slope does not occur, the operation proceeds to the next operation.

Accordingly, the banknote image extracting unit 18 extracts only the banknote image excluding the margin image from the image of the tilted state (S14).

Subsequently, in a sixteenth step S16, the banknote size estimating unit 20 estimates the size of the extracted banknote image and compares the estimated length of the banknote with the size data of the banknote stored in the memory 28. For example, the size of a new voucher issued recently (Jan. 2007) is 148 * 68mm in width * and 161 * 76mm in the size of the license. Accordingly, in the present embodiment, the priesthood and old books are distinguished by comparing whether the length of the extracted banknote is smaller than 72 mm.

As a result of the inspection, if the vertical length of the banknote is less than 72mm, the banknote size estimation unit 20 determines the banknote as a new priesthood bill (S20), and if the vertical length of the banknote is larger than 72mm, the banknote size estimation unit ( 20) it is determined that the bill is krwon 10,000 won (S30).

When the bill notes are determined as described above, the billing direction of the bill is subsequently estimated. For this purpose, the differential average calculation unit 42 selects a predetermined number of bill paper images, for example, width * length as shown in FIGS. 4A to 7A. The 4 * 3 is divided equally into 12 regions, and the image of each region is differentiated to calculate the differential averages (S21 and S31).

Subsequently, the pulling direction estimating unit 24 compares the differential means for each region according to the pulling direction estimation algorithm.

First, the retraction direction estimator 24 compares the derivative mean of the area b with the derivative average of the area e in step 22. When the differential mean of the b area is small, the inflow direction estimator 24 compares the differential averages of the a area and the c area (S23 and S33).

In the case where the banknote is a full-won priesthood, the differential mean for each region is calculated as shown in FIG. 4B, that is, the differential mean of region '6' is smaller than the derivative mean of region e of '36', and the derivative mean of region a. When the derivative average of '15' is smaller than '51' in the area c, the retracting direction estimating unit 24 estimates the retracting direction of the banknote as the front face of the priesthood of FIG. 4A (S24).

On the other hand, when the derivative mean '31' of the region a by the reverse order of FIG. 5B is larger than the derivative mean '4' of the region c according to the reverse order of FIG. 5B, the back priesthood bill as shown in FIG. 5A is inserted in the inverting direction. It is recognized as a case (S25).

In operation 22, when the derivative mean of the region b is greater than the derivative average of the region e, the retraction direction estimator 24 compares the derivative average of the region d and the derivative average of the region f (S26).

When the derivative mean is calculated as shown in FIG. 5B, that is, when the derivative mean '4' of the area d is smaller than the derivative mean '31' of the area f, the retracting direction estimator 24 determines that the back bill of the priesthood is drawn in the upright direction. It is assumed that (S27).

On the other hand, if the derivative mean '51' of the region d in the reverse order of FIG. 4B is larger than the derivative mean '15' of the region f, it is recognized as a case in which the priesthood banknote of FIG. 4A is drawn in the inverted direction (S28).

Similarly, in the case where the banknote is the 10,000 won bill, the differential mean '5' of the region b in the reverse order of FIG. 7B is smaller than the derivative mean '23' of the region e (S32), and the derivative average '20' of the region a. If the derivative mean '37' of the area c is smaller than S37 (S33), it is assumed that the bill-back direction of the bill is drawn in the inverted direction (S34).

In addition, the inflow direction estimating unit 24 determines that the derivative mean '7' of the region b in FIG. 6B is smaller than the derivative mean '32' of the region e (S32), and the derivative mean '30' of the region a is the derivative of the region c. If the average is larger than '16' (S33), the billing direction of the bill is assumed to be drawn in the upright banknote in Fig. 6A (S35).

Further, in step 32 (S32), the derivative mean '32' of the region b is larger than the derivative average '7' of the region e, and the derivative average '16' of the region d is different from the region f in the reverse order of FIG. 6B. If the average is smaller than '30' (S36), the retracting direction estimator 24 estimates that the old-sided bill of FIG. 6A is retracted in the inverted direction (S37).

 On the other hand, when the derivative mean '37' of the region d in FIG. 7B is larger than the derivative mean '20' of the region f in FIG. 7B (S36), the inflow direction estimating unit 24 estimates the direction of the back of the old sphere.

As described above, according to the present invention, the inflow direction of the bills can be estimated only by a simple comparison operation by the inflow direction estimation algorithm using the differential mean for each area in which the image of each bill is equally divided into a predetermined number.

Meanwhile, the derivative average table of FIGS. 4B to 7B is stored in the memory 28, and the differential average calculating unit 22 calculates the differential average for each area of the banknote image, and then the control unit 26 compares the pulling direction. It may be changed to estimate.

In the above embodiment, the present invention has been described using a thousand won bill, but the present invention can be applied to a media recognition device and method for scanning and recognizing images of various media such as checks, tickets, certificates, as well as bills such as thousand won bills and five thousand won bills. have.

To this end, the order of inference estimation algorithm is applied by changing the comparison order and the processing result according to the characteristics of the medium to be recognized, that is, the derivative average characteristic of the image for each region divided into a predetermined number.

 While the present invention has been described with reference to exemplary embodiments, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, It will be apparent that various modifications, alterations, and other equivalent embodiments are possible. Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

As described above, the present invention estimates the pulling direction of the banknote according to the pulling direction estimation algorithm using the derivative mean of the banknote image. The present invention has the following effects.

First, since the present invention is performed by a simple operation of comparing the derivative average of each region of a bill image, the processing speed is faster than that of the conventional technique of comparing various feature points and feature values with a database.

In addition, the present invention is driven by the algorithm using the derivative average of the bill image, regardless of the change in the output of the LED, the sensitivity of the image sensor, the vibration of the transfer module and the noise generated in the other circuit elements, the accuracy and precision of bill recognition Is improved.

In addition, since the present invention stores only estimation algorithms excluding various feature points and feature values in the memory, the memory capacity consumed is reduced, thereby reducing the manufacturing cost.

Accordingly, the present invention improves the satisfaction by quickly and accurately estimating the incoming direction of the medium.

Claims (16)

An image sensor unit scanning an image of a medium drawn from the outside; A differential average calculation unit for dividing the image scanned by the image sensor unit into a predetermined number of regions, and calculating a differential average of the divided region-specific images; And And a retraction direction estimator for estimating a retraction direction of the medium by comparing the differential averages of the divided region-specific images. The method according to claim 1, A tilt correction unit for selectively correcting when the incoming medium has a tilt, A media image extraction unit for extracting only a media image in a state where the inclination of the media is aligned; and And a media size estimator for estimating a media size from the extracted media image. The apparatus of claim 2, wherein the medium size estimating unit And a media recognition device for estimating a media type by estimating a vertical length of the extracted media image. The apparatus of claim 2, wherein the medium size estimating unit Media recognition device for judging the priesthood and the old right according to the type of the medium. delete delete The method of claim 1, wherein the differential average calculation unit, And a media recognition device for dividing an area in which the differential mean is provided in two or more. delete Scanning the image for incoming media, Dividing an image of the scanned medium into a predetermined number of regions; Calculating a differential mean of images of each of the divided regions, and Estimating a retracting direction by sequentially comparing the derivative averages of the images of two regions among the derivative averages of the partial region images to determine the front / rear and upright / inverted states of the medium. A method of estimating the direction of media entry using media recognition device. delete delete delete delete delete delete delete

Images