CN102592351B - Authentication method of papery medium - Google Patents

Abstract

The invention provides an authentication device and an authentication method of papery medium, the authentication device comprises an image data obtaining part, a standard template data storing part, a bad line detecting and processing part, an image cutting part and a judging part, wherein the bad line detecting and processing part is used for detecting whether the image data obtained by the image data obtaining part have a bad line and recovering the image according to the characteristics of the bad line. The image is processed by bad line detection and bad line, and the judging part matches the recovered image with a standard template to obtain an authentication result, so that the phenomena of the possible fault of a certain light emitting element of a CIS (contact image sensor) or the dirt of a CIS light transmitting panel, which can be encountered by the papery medium authentication device, have no influence on the authentication, thereby improving the acceptance rate of the papery medium authentication device.

Description

The discrimination method of paper sheet medium

Technical field

The present invention relates to a kind of authentication technique of paper sheet medium, especially in paper sheet medium identification device, can effectively process certain light-emitting component fault of imageing sensor (CIS) or CIS light transmission piece and have external dirty technology.

Background technology

The identification device of differentiating paper sheet medium (bank note) is installed on the ATM (Automatic Teller Machine) of financial institution.Storage part at paper sheet medium identification device has the paper sheet of discriminating medium template data.When having dropped into the paper sheet medium, the view data obtaining section of paper sheet medium identification device reads the paper sheet medium with optical mode, obtain the gray level image data of paper sheet medium to be identified, deep or light according to the pattern of obtained paper sheet medium gray level image to be identified and gray-scale value, between template and testing result, mate, judge thus kind (Currency Type) and the true and false of described paper sheet medium.

Long service time when paper sheet medium identification device, certain light-emitting component fault of CIS or the dirty phenomenon of CIS light transmission piece may appear.Because CIS is line sweep, when certain light-emitting component fault of CIS or the dirty phenomenon of CIS light transmission piece, on the reflectogram that imageing sensor collects or transmission plot, there will be a gray average close to 0 black line (following by this black line referred to as " bad line ", this phenomenon is called to " the bad line phenomenon of CIS ").Because there are randomness in certain light-emitting component fault of CIS or the dirty position of CIS light transmission piece, therefore bad line position in the image of paper sheet medium to be measured is also stochastic distribution.In this case, if still follow existing identification process, can directly affect the receptance of paper sheet medium identification device, the performance of whole paper sheet medium identification device is brought to tremendous influence, affect customer experience, be unfavorable for marketing and the use of device.

Therefore, provide a kind of technology that can effectively process the bad line phenomenon of CIS in paper sheet medium identification device very necessary.

Summary of the invention

One of purpose of the present invention is to provide a kind of paper sheet medium identification device, and this paper sheet medium identification device can effectively be processed the bad line phenomenon of CIS, and its receptance improves greatly.

The present invention also provides above-mentioned paper sheet medium identification device to differentiate the method for paper sheet medium.

This paper sheet medium identification device comprises: transfer passage, in order to the paper sheet medium is carried one by one and is passed through the view data obtaining section; The view data obtaining section, in order to obtain the paper sheet medium by the time raw image data; The standard form data store, differentiate the needed standard form data message of this paper sheet medium in order to storage; Bad line Check processing section, in order to detect this obtained raw image data of this view data obtaining section, whether there is bad line, and in the situation that the existence of bad line detected, obtain the position of this bad line in whole image, and carry out image repair according to the feature of this bad line; The image cutting part, in order to cut the recognition image information of effective identification paper sheet medium; And detection unit, in order to the recognition image information to cutting, implement template matches identification, judge that the similarity of this recognition image information and this standard form data message is to obtain identification result.

Preferably, this image cutting part carries out effective identified region cutting in order to the image information to the bad line Check processing of process section and forms characteristic image information.

Further, this characteristic image information complete image information that is this paper sheet medium or be at least a portion image information of this paper sheet medium.

The discrimination method of this paper sheet medium, it comprises: step 1, drop into the paper sheet medium, this paper sheet medium enters transfer passage; Step 2, the image when imageing sensor scanning paper sheet medium passes through transfer passage forms raw image data; Step 3, the raw image data that bad line Check processing section obtains the view data obtaining section is analyzed, and detects and whether has bad line; If detect, finish not find to enter step 4 by bad line; If there is bad line, confirm simultaneously the position of bad line in image, according to bad line gray feature, implement bad line reparation, then enter step 4; Step 4, image cutting part are obtained paper sheet dielectric image zone; Namely survey paper sheet dielectric image zone boundary, and by paper sheet dielectric image zone from original image, cutting out, obtain for the effective recognition image information of identification; And step 5, detection unit calls the standard form data in storage part, carries out template matches identification, judges that namely the similarity of this recognition image information and these standard form data is to obtain identification result.

Wherein, the method that in step 3, whether bad line of detection exists is: the image of paper sheet medium is placed in to the xy coordinate system, establishes:

1. the gradation of image value matrix is x ₀Row, y ₀OK;

2. g (x, y) is the gray-scale value of the capable x column position of y of image;

3.

Gray average for image x row;

According to constraint condition

$\left\{\begin{array}{c}g(x,y)<{\mathrm{\&epsiv;}}_1,0<y<y_0;\\ \stackrel{\&OverBar;}{g}\left(x\right)<{\mathrm{\&epsiv;}}_2,0<y<y_0;\\ |\stackrel{\&OverBar;}{g}(x-1)-\stackrel{\&OverBar;}{g}\left(x\right)|>{\mathrm{\&epsiv;}}_3;\\ |\stackrel{\&OverBar;}{g}(x+1)-\stackrel{\&OverBar;}{g}\left(x\right)|>{\mathrm{\&epsiv;}}_4;\end{array}\right.$ ε wherein ₁, ε ₂, ε ₃, ε ₄All be greater than 0, be setting threshold,

The gradation of image value matrix that imageing sensor is obtained carries out the automatic search detection, if there is a certain row x=n in the gradation of image matrix ₀, i.e. the n of gradation of image matrix ₀Row meet above-mentioned constraint condition, on the process decision chart picture, have bad line, otherwise do not have bad line.

When having bad line, bad line is taked to repair as drag:

g(n ₀,y)=ω ₁g(n ₀-2,y)+ω ₂g(n ₀-1,y)+ω ₃g(n ₀+1,y)+ω ₄g(n ₀+2,y)，

Wherein, y ∈ (0, y ₀), ω ₁, ω ₂, ω ₃, ω ₄For bad line is repaired the linear weighting coefficient of gray scale match, ω ₁+ ω ₂+ ω ₃+ ω ₄=1, namely on bad line, the gray-scale value of any point is used with this arbitrfary point and is substituted with the first two point of delegation and the linear weighted function value of the gray-scale value of latter two point.

Because having carried out bad line, the paper sheet dielectric image to be identified detects and bad line reparation, make certain light-emitting component fault of CIS or the dirty phenomenon of CIS light transmission piece that paper sheet medium identification device may occur affect differentiating nothing, thereby improve the receptance of paper sheet medium identification device, be conducive to marketing and the use of device.

The accompanying drawing explanation

Fig. 1 is a kind of paper sheet medium identification device construction module figure that the preferred embodiment of the present invention provides;

Fig. 2 is the paper sheet medium discrimination method process flow diagram that the preferred embodiment of the present invention provides;

When Fig. 3 is the bad line of CIS appearance, the image schematic diagram that imageing sensor obtains;

When Fig. 4 repairs bad line, the image that imageing sensor obtains and standard form coupling schematic diagram;

Fig. 5 is that the image in the preferred embodiment of the present invention, imageing sensor obtained is placed in the schematic diagram that the xOy coordinate system waits to do the reparation of bad line;

Fig. 6 is the image schematic diagram after the bad line in the image that imageing sensor is obtained is repaired; And

Fig. 7 is by the schematic diagram of the image after the reparation of bad line and standard form coupling.

Embodiment

For further setting forth the present invention, introduce specific embodiments of the invention below in conjunction with diagram.

Fig. 1 be a preferred embodiment of the present invention provide paper sheet medium identification device 100 system module structural drawing.This paper sheet medium identification device 100 is embedded in the ATM (automatic teller machine) such as ATM() etc. in the paper sheet media processing apparatus, by transfer passage, paper sheet medium to be identified (for example bank note) carried one by one and passed through the view data obtaining section.Concrete, this papery class medium identification device 100 comprises view data obtaining section 10, standard form data store 20, bad line Check processing section 30, image cutting part 40 and detection unit 50, wherein, this view data obtaining section 10 in order to obtain the paper sheet medium by the time raw image data; This standard form data store 20 is differentiated the needed standard form data message of this paper sheet medium in order to storage; Whether this bad line Check processing section 30 exists bad line in order to detect this obtained raw image data of this view data obtaining section, and in the situation that the existence of bad line detected, obtain the position of this bad line in whole image, and carry out image repair according to the feature of this bad line; This image cutting part 40 is in order to cut the recognition image information of effective identification paper sheet medium; This detection unit 50 is implemented template matches identification in order to this recognition image information to cutting, judges that the similarity of this recognition image information and this standard form data message is to obtain identification result.Wherein, this image cutting part carries out effective identified region cutting in order to the image information to the bad line Check processing of process section and forms characteristic image information.The complete image information that this characteristic image information is this paper sheet medium or be at least a portion image information of this paper sheet medium.

Fig. 2 is the method flow diagram that adopts 100 pairs of paper sheet media of paper sheet medium identification device to differentiate, comprises the following steps: step 1, drop into the paper sheet medium, and this paper sheet medium enters transfer passage, and flow process starts; Step 2, the image when imageing sensor scanning paper sheet medium passes through transfer passage forms raw image data; Step 3, the raw image data that bad line Check processing section obtains the view data obtaining section is analyzed, and detects and whether has bad line; If detect, finish not find to enter step 4 by bad line; If there is bad line, confirm simultaneously the position of bad line in image, according to bad line gray feature, implement bad line reparation, then enter step 4; Step 4, image cutting part are obtained paper sheet dielectric image zone; Namely survey paper sheet dielectric image zone boundary, and by paper sheet dielectric image zone from original image, cutting out, obtain for the effective recognition image information of identification; And step 5, detection unit calls the standard form data in storage part, carries out template matches identification, judges that namely the similarity of this recognition image information and these standard form data is to obtain identification result.

This view data obtaining section 10 forms by detecting through the projection sensor of image and/or the reflective sensor of detection of reflected image, when described paper sheet medium transmits through transfer path, view data obtaining section 10 is obtained fluoroscopy images and/or the reflected image of described paper sheet medium, and the fluoroscopy images that obtains and/or reflected image are reached to bad line Check processing section 30.

In bad line Check processing section 30 original image that at first the detected image sensor obtains, whether there is bad line, if there is bad line in the original image of the paper sheet medium dropped into, and get accurately the positional information of bad line in whole image.Because the CIS imaging is line sweep, if there is bad line, on the original image of paper sheet medium to be measured, will occur that a gray-scale value is close to 0 black line.Because the position existence of bad line of CIS is uncertain, therefore also there is uncertainty in the position of bad line in the original image of paper sheet medium to be measured.Suppose when bad line (a bad line) appears in CIS the image that imageing sensor obtains such as Fig. 3.

If any measure is not taked in bad line zone, directly implement image cutting and template matches identification, when bad line is in certain pending template matches identification regional just, obviously coupling can not be successful, as Fig. 4, the identification of paper sheet medium to be identified ends in failure, and finally causes normal paper sheet medium to be rejected by financial self-service equipment (as ATM).

In the present embodiment, utilize the reflected image of the input paper sheet medium that reflective sensor obtains to detect whether there is bad line.The reflected image that drops into the paper sheet medium is placed in to coordinate system as described in Figure 5,

If:

1. reflected image gray-scale value matrix is x ₀Row, y ₀OK;

2. g (x, y) is the gray-scale value of the capable x column position of y of reflected image;

3.

Gray average for reflected image x row.

According to constraint condition

$\left\{\begin{array}{c}g(x,y)<{\mathrm{\&epsiv;}}_1,0<y<y_0;\\ \stackrel{\&OverBar;}{g}\left(x\right)<{\mathrm{\&epsiv;}}_2,0<y<y_0;\\ |\stackrel{\&OverBar;}{g}(x-1)-\stackrel{\&OverBar;}{g}\left(x\right)|>{\mathrm{\&epsiv;}}_3;\\ |\stackrel{\&OverBar;}{g}(x+1)-\stackrel{\&OverBar;}{g}\left(x\right)|>{\mathrm{\&epsiv;}}_4;\end{array}\right.$ Wherein, ε ₁, ε ₂, ε ₃, ε ₄All be greater than 0, be setting threshold, (1.1)

For example can establish ε ₁=20, ε ₂=20, ε ₃=80, ε ₄The reflected image gray-scale value matrix that=80 pairs of imageing sensors obtain carries out the automatic search detection.

If traversal reflected image gray-scale value matrix does not all meet the row x of constraint condition (1.1), on the reflected image that the process decision chart image-position sensor obtains, there do not is bad line, and then infer that there is not the bad line phenomenon of CIS in this paper sheet medium identification device, directly enter identification paper sheet dielectric image zone cutting link, finally template matches identification is implemented in the paper sheet dielectric image zone cut out, obtain final recognition result.

If there are a certain row in the reflected image gray matrix

x=n ₀

Be the n of gradation of image matrix ₀Row meet above-mentioned constraint condition (1.1), judge on reflected image and have bad line, and namely there is the bad line phenomenon of CIS in this sheet discriminating device, and bad line position is arranged in the n of the gray level image of imageing sensor acquisition ₀Row, enter bad line and repair link.

When treating that imageing sensor obtains reflected image x=n ₀When row are judged as bad line, utilize image processing techniques to treat recognition image and carry out bad line reparation.Because bad line both sides are normal gray-scale value, and gray level image any gray-scale value and this vertex neighborhood in the point gray-scale value have certain correlativity, the present embodiment is taked to repair as drag to bad line:

g(n ₀,y)=ω ₁g(n ₀-2,y)+ω ₂g(n ₀-1,y)+ω ₃g(n ₀+1,y)+ω ₄g(n ₀+2,y) （1.2）

Y ∈ (0, y wherein ₀), ω ₁+ ω ₂+ ω ₃+ ω ₄=1, for example get

Namely on bad line the gray-scale value of any point use with this with the first two of delegation and latter two gray-scale value and average substitute.Image after bad line is repaired as shown in Figure 6.

Then, the image that 40 pairs of this image cutting parts have carried out bad line reparation carries out paper sheet dielectric image regionally detecting and cutting, effective identified region, from the whole image that imageing sensor obtains, cutting out, is obtained to the image for template matches identification, as shown in Figure 7.

Detection unit 50 utilizes the standard form of these standard form data store 20 storages, and paper sheet dielectric image area image is carried out to template matches identification, obtains final recognition result.

Because having carried out bad line, the image to the paper sheet medium detects and bad line reparation, make certain light-emitting component fault of CIS or the dirty phenomenon of CIS light transmission piece that paper sheet medium identification device may occur affect differentiating nothing, thereby improve the receptance of paper sheet medium identification device, be conducive to marketing and the use of device.

Be below only the preferred embodiment of the present invention, it should be pointed out that above-mentioned preferred implementation should not be considered as limitation of the present invention, protection scope of the present invention should be as the criterion with the claim limited range.For those skilled in the art, without departing from the spirit and scope of the present invention, can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (2)

1. the discrimination method of a paper sheet medium, it comprises:

Step 1, drop into the paper sheet medium, and this paper sheet medium enters transfer passage;

Step 2, the image when imageing sensor scanning paper sheet medium passes through transfer passage forms raw image data;

Step 3, the raw image data that bad line Check processing section obtains the view data obtaining section is analyzed, and detects and whether has bad line; If detect, finish not find to enter step 4 by bad line; If there is bad line, confirm simultaneously the position of bad line in image, according to bad line gray feature, implement bad line reparation, then enter step 4;

Step 4, image cutting part are obtained paper sheet dielectric image zone; Namely survey paper sheet dielectric image zone boundary, and by paper sheet dielectric image zone from original image, cutting out, obtain for the effective recognition image information of identification;

Step 5, detection unit are called the standard form data in storage part, carry out template matches identification, judge that namely the similarity of this recognition image information and these standard form data is to obtain identification result;

It is characterized in that, the method that in this step 3, whether bad line of detection exists is:

The image of paper sheet medium is placed in to the xy coordinate system,

If:

1. the gradation of image value matrix is x ₀Row, y ₀OK;

2. g (x, y) is the gray-scale value of the capable x column position of y of image;

3.

Gray average for image x row;

According to constraint condition

$\left\{\begin{array}{c}g(x,y)<{\mathrm{\&epsiv;}}_1,0<y<y_0;\\ \stackrel{\&OverBar;}{g}\left(x\right)<{\mathrm{\&epsiv;}}_2,0<y<y_0;\\ |\stackrel{\&OverBar;}{g}(x-1)-\stackrel{\&OverBar;}{g}\left(x\right)|>{\mathrm{\&epsiv;}}_3;\\ |\stackrel{\&OverBar;}{g}(x+1)-\stackrel{\&OverBar;}{g}\left(x\right)|>{\mathrm{\&epsiv;}}_4;\end{array}\right.$ ε wherein ₁, ε ₂, ε ₃, ε ₄All be greater than 0, be setting threshold,

The gradation of image value matrix that imageing sensor is obtained carries out the automatic search detection, if there is a certain row x=n in the gradation of image matrix ₀, i.e. the n of gradation of image matrix ₀Row meet above-mentioned constraint condition, on the process decision chart picture, have bad line, otherwise do not have bad line.

2. the discrimination method of paper sheet medium as claimed in claim 1, is characterized in that, in this step 3 to bad

Line is taked to repair as drag:

g(n ₀,y)=ω ₁g(n ₀-2,y)+ω ₂g(n ₀-1,y)+ω ₃g(n ₀+1,y)+ω ₄g(n ₀+2,y)

Wherein, y ∈ (0, y ₀), ω ₁, ω ₂, ω ₃, ω ₄For bad line is repaired the linear weighting coefficient of gray scale match, ω ₁+ ω ₂+ ω ₃+ ω ₄=1, namely on bad line, the gray-scale value of any point is used with this arbitrfary point and is substituted with the first two point of delegation and the linear weighted function value of the gray-scale value of latter two point.

Images