CN111462392A - Method and device for identifying paper money based on multispectral image similarity algorithm - Google Patents
Method and device for identifying paper money based on multispectral image similarity algorithm Download PDFInfo
- Publication number
- CN111462392A CN111462392A CN202010270998.2A CN202010270998A CN111462392A CN 111462392 A CN111462392 A CN 111462392A CN 202010270998 A CN202010270998 A CN 202010270998A CN 111462392 A CN111462392 A CN 111462392A
- Authority
- CN
- China
- Prior art keywords
- image
- light source
- similarity
- imaging
- banknote
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07D—HANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
- G07D7/00—Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency
- G07D7/20—Testing patterns thereon
- G07D7/2008—Testing patterns thereon using pre-processing, e.g. de-blurring, averaging, normalisation or rotation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07D—HANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
- G07D7/00—Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency
- G07D7/20—Testing patterns thereon
- G07D7/2016—Testing patterns thereon using feature extraction, e.g. segmentation, edge detection or Hough-transformation
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07D—HANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
- G07D7/00—Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency
- G07D7/20—Testing patterns thereon
- G07D7/202—Testing patterns thereon using pattern matching
Abstract
A method and a device for identifying paper money based on a multispectral image similarity algorithm are disclosed, wherein the method comprises the following steps: collecting a multispectral imaging image of the paper money to be identified, selecting two spectral bands of light sources, wherein one spectral band of light source is used as a reference light source, the other spectral band of light source is used as a comparison light source, and respectively obtaining imaging images of the reference light source and the comparison light source of the paper money to be identified; selecting similar areas and/or dissimilar areas from images imaged by the reference light source and the comparison light source as identification characteristic areas of the paper money; and performing image similarity calculation on the identification characteristic areas of the images imaged by the two light sources to obtain a similarity matching result, and judging the authenticity of the paper money according to the similarity matching result. The similarity calculation of multispectral different channels is adopted to identify true and false without depending on counterfeit money samples; the similarity characteristic among multispectral images is adopted, the method does not depend on a true currency template, and the execution efficiency and the storage space utilization have great advantages.
Description
Technical Field
The invention relates to the field of banknote authenticity image identification of financial equipment, in particular to a method and a device for identifying banknotes based on a multispectral image similarity algorithm.
Background
The financial machines and tools need to realize the real-time counting and identifying functions of paper money; where image authentication is an important authentication method. Due to the manufacturing process characteristics of the paper money, whether the paper money is RMB or foreign currency, true currency of each denomination version has the specific multispectral image characteristics; the counterfeit currency is difficult to realize the matching of multispectral full characteristics.
The banknote image has different image characteristics under different spectrums (fig. 1. A-1. B are imaging images of different light source spectrum sections of the banknote, and 1.C is a spectrum distribution diagram); an algorithm of multispectral image similarity is provided by analyzing the imaging difference of the banknote image under different spectrums; the algorithm can identify the authenticity characteristics of the paper money, thereby realizing the anti-counterfeiting target of the financial machine.
To realize counterfeit money discrimination, there are several methods:
one method is to adopt a false authentication method for authentication, see patent CN 201110139804-a banknote multispectral image analysis method, select a region pair to construct a feature vector (e.g., a mean difference of the region pair or other features (variance and gray level co-occurrence matrix, etc.) to calculate the feature vector); and calculating a correlation coefficient.
The method of false authentication requires a large number of counterfeit money samples; a large amount of counterfeit money samples are collected with large workload and cannot be collected comprehensively; the method is dependent on counterfeit money samples and is not suitable for quick adaptation of newly added paper money identification.
One method is to adopt a true-to-true discrimination method for discrimination, see patent CN 201710030776-a method and device for discriminating counterfeit paper money, select a multispectral characteristic area, perform characteristic matching with a preset true money sample information base template, if the matching is true money, if the mismatching is false money; the method of true discrimination is adopted for discrimination.
According to the scheme of true authentication, the characteristics of the true currency sample information base template are required to be matched, and the preset sample information base still needs to be collected and stored.
Disclosure of Invention
In view of the technical defects and technical drawbacks in the prior art, embodiments of the present invention provide a method and an apparatus for performing banknote authentication based on a multispectral image similarity algorithm, which overcome the above problems or at least partially solve the above problems, and the specific scheme is as follows:
as a first aspect of the present invention, there is provided a method of banknote authentication based on a multispectral image similarity algorithm, the method comprising the steps of:
step 1, acquiring a multispectral imaging image of a paper currency to be identified, selecting two spectral bands of light sources, wherein one spectral band of the light source is used as a reference light source, the other spectral band of the light source is used as a comparison light source, and respectively acquiring imaging images of the paper currency to be identified by the reference light source and the comparison light source, namely a reference light source imaging image and a comparison light source imaging image;
step 2, selecting similar areas and/or dissimilar areas from images imaged by two light sources, namely a reference light source imaging image and a comparison light source imaging image, as identification characteristic areas of the two light source imaging images;
and 3, performing image similarity calculation on the reference light source imaging image and the identification characteristic area of the comparison light source imaging image to obtain a similarity matching result, and judging the authenticity of the paper money according to the similarity matching result.
Further, in step 3, the image similarity calculation of the identification feature region of the reference light source imaging image and the comparison light source imaging image is specifically as follows:
because the light source of each spectrum section has background interference or noise interference, in order to eliminate the influence of the background or noise interference, a similarity characteristic threshold value threshold _ Ref _ light _ source of an identification characteristic region selected from an image imaged by a reference light source is firstly calculated, and similarity calculation is carried out on the identification characteristic region selected from the image imaged by the light source based on the guidance of the image similarity characteristic threshold value threshold _ Ref _ light _ source, wherein the specific formula is as follows:
wherein, the threshold _ Ref _ light _ source is a similarity feature threshold of an authentication feature region selected from an image imaged according to a reference light source, the Com _ light _ source is an image feature of the authentication feature region selected from the image imaged by a contrast light source, the threshold _ true _ money is an image experience value of a region corresponding to a genuine banknote image, and sim is an image similarity matching result.
Further, in step 3, the step of judging the authenticity of the banknote according to the similarity matching result specifically comprises:
if the identification characteristic area is a similar area in the images imaged by the two selected light sources and the similarity matching result is similar, judging the paper currency to be a true currency, otherwise, judging the paper currency to be a false currency;
if the identification characteristic area is a dissimilar area in the images imaged by the two selected light sources and the similarity matching result is dissimilar, the paper currency is judged to be a true currency, otherwise, the paper currency is judged to be a false currency.
Further, the method further comprises:
after collecting a multispectral imaging image of a paper currency to be identified, positioning the corner point position of the paper currency in the acquired multispectral imaging image to obtain the four corner point positions of the paper currency in the imaging image;
and performing affine transformation on the image through the acquired positions of the four corners, transforming the random imaging image into a complete banknote image which has a target pixel size and a regular shape, and taking the complete banknote image as an imaging image.
Further, positioning corner positions of the acquired banknote positions in the multispectral imaging image, and acquiring four corner positions of the banknote positions in the imaging image specifically are as follows:
finding four boundaries of the paper money in the image according to the gray difference between the paper money in the image and the background;
after finding the four boundaries of the paper money, fitting boundary points of the four boundary areas to obtain boundary straight lines of the four boundaries;
and calculating intersection points of the boundary straight lines intersected every two to obtain four intersection points, namely the positions of four corner points of the paper money position.
As a second aspect of the present invention, there is provided an apparatus for banknote authentication based on a multispectral image similarity algorithm, the apparatus comprising: the device comprises an image acquisition module, a reference light source imaging image and comparison light source imaging image acquisition module, an identification characteristic region selection module and an image similarity identification module;
the image acquisition module is used for acquiring a multispectral imaging image of the paper money to be identified;
the reference light source imaging image and comparison light source imaging image acquisition module is used for selecting light sources of two spectral bands, wherein the light source of one spectral band is used as a reference light source, the light source of the other spectral band is used as a comparison light source, and imaging images of the reference light source and the comparison light source, namely a reference light source imaging image and a comparison light source imaging image, are respectively acquired from a multi-spectral imaging image;
the identification characteristic region selection module is used for selecting similar regions and/or dissimilar regions from images formed by two light sources, namely a reference light source imaging image and a comparison light source imaging image, as identification characteristic regions of the two light source imaging images;
the image similarity identification module is used for carrying out image similarity calculation on identification characteristic areas of the images imaged by the two light sources to obtain a similarity matching result, and judging the authenticity of the paper money according to the similarity matching result.
Further, the image similarity identification module performs image similarity calculation on the identification characteristic regions of the reference light source imaging image and the comparison light source imaging image specifically as follows:
because the light source of each spectrum section has background interference or noise interference, in order to eliminate the influence of the background or noise interference, a similarity characteristic threshold value threshold _ Ref _ light _ source of an identification characteristic region selected from an image imaged by a reference light source is firstly calculated, and similarity calculation is carried out on the identification characteristic region selected from the image imaged by the light source based on the guidance of the image similarity characteristic threshold value threshold _ Ref _ light _ source, wherein the specific formula is as follows:
wherein, the threshold _ Ref _ light _ source is a similarity feature threshold of an authentication feature region selected from an image imaged according to a reference light source, the Com _ light _ source is an image feature of the authentication feature region selected from the image imaged by a contrast light source, the threshold _ true _ money is an image experience value of a region corresponding to a genuine banknote image, and sim is an image similarity matching result.
Further, the image similarity identification module judges the authenticity of the paper currency according to the similarity matching result, and specifically comprises the following steps:
if the identification characteristic area is a similar area in the images imaged by the two selected light sources and the similarity matching result is similar, judging the paper currency to be a true currency, otherwise, judging the paper currency to be a false currency;
if the identification characteristic area is a dissimilar area in the images imaged by the two selected light sources and the similarity matching result is dissimilar, the paper currency is judged to be a true currency, otherwise, the paper currency is judged to be a false currency.
Further, the device also comprises an image corner point positioning module and an image affine transformation module;
the image corner positioning module is used for positioning the corner positions of the paper currency in the acquired multispectral imaging image after acquiring the multispectral imaging image of the paper currency to be identified, so as to obtain the positions of four corner points of the paper currency in the imaging image;
the image affine transformation module is used for carrying out affine transformation on the image through the acquired positions of the four corners, transforming the random imaging image into a complete banknote image which is in a target pixel size and is in a regular shape, and taking the complete banknote image as an imaging image.
Further, the image corner positioning module performs corner position positioning on the acquired banknote position in the multispectral imaging image, and the acquiring of the four corner positions of the banknote position in the imaging image specifically comprises:
finding four boundaries of the paper money in the image according to the gray difference between the paper money in the image and the background;
after finding the four boundaries of the paper money, fitting boundary points of the four boundary areas to obtain boundary straight lines of the four boundaries;
and calculating intersection points of the boundary straight lines intersected every two to obtain four intersection points, namely the positions of four corner points of the paper money position.
The invention has the following beneficial effects:
1. the invention adopts the similarity calculation of multispectral different channels to identify true and false without depending on counterfeit money samples.
2. The invention adopts the similarity characteristic between multispectral images, does not depend on a true currency template, and has great advantages in execution efficiency and storage space utilization.
3. The similarity calculation method effectively solves the problems of image background and noise influence.
4. The invention can be applied to various currency denomination versions such as RMB currency, foreign currency and the like; each particular currency denomination version has its own multispectral image features, and multiple pairs of discriminating feature regions can be selected for analysis on that particular banknote according to the method.
Drawings
Fig. 1A to 1B are schematic diagrams illustrating that a banknote image provided by an embodiment of the present invention has different image characteristics under different spectra;
FIG. 1C is a graph of a spectrum profile provided by an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a method for banknote identification based on a multispectral image similarity algorithm according to an embodiment of the present invention;
fig. 3A and 3B are schematic diagrams of corner point positioning performed on a banknote region of an original imaging image according to an embodiment of the present invention;
FIG. 4A is an image of green reflectance spectrum G provided by an embodiment of the present invention;
fig. 4B is an imaging image of the infrared reflection spectrum IR provided by the embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Taking a 50 euro banknote as an example, as shown in fig. 1A and 1B, the banknote image has different image characteristics under different spectra, and the spectral distribution diagram is shown in fig. 1C.
The invention provides a method for identifying paper money based on a multispectral image similarity algorithm, which comprises the following steps as shown in figure 2:
step 1, acquiring a multispectral imaging image of a paper currency to be identified, selecting two spectral bands of light sources, wherein one spectral band of the light source is used as a reference light source, the other spectral band of the light source is used as a comparison light source, and respectively acquiring imaging images of the reference light source and the comparison light source of the paper currency to be identified from the multispectral imaging image, namely a reference light source imaging image and a comparison light source imaging image;
step 2, selecting similar areas and/or dissimilar areas from images imaged by two light sources, namely a reference light source imaging image and a comparison light source imaging image, as identification characteristic areas of the two light source imaging images;
and 3, performing image similarity calculation on the reference light source imaging image and the identification characteristic area of the comparison light source imaging image to obtain a similarity matching result, and judging the authenticity of the paper money according to the similarity matching result.
The image similarity calculation of the identification characteristic regions of the images imaged by the two light sources specifically comprises the following steps:
because the light source of each spectrum section has background interference or noise interference, in order to eliminate the influence of the background or noise interference, a similarity characteristic threshold value threshold _ Ref _ light _ source of an identification characteristic region selected from an image imaged by a reference light source is firstly calculated, and similarity calculation is carried out on the identification characteristic region selected from the image imaged by the light source based on the guidance of the image similarity characteristic threshold value threshold _ Ref _ light _ source, wherein the specific formula is as follows:
wherein, threshold _ Ref _ light _ source is a similarity feature threshold of an identification feature area selected from an image imaged by a reference light source, and the feature threshold reflects a foreground feature of the reference light source image; the Com _ light _ source is the image characteristics of an identification characteristic area selected from the image imaged by the contrast light source according to the foreground characteristics of the image of the reference light source; the foreground characteristics of the light sources can be calculated and compared; threshold _ true _ money is an image experience value of a region corresponding to the true coin image, comparison is carried out according to the foreground characteristic of the comparison light source and the foreground experience value of the true coin comparison light source, the similarity of the test sample and the true coin can be compared and evaluated, and the feature of the test sample is evaluated according to the feature of the true coin; sim is the image similarity matching result.
The method for judging the authenticity of the paper money through the similarity matching result specifically comprises the following steps:
if the identification characteristic area is a similar area in the images imaged by the two selected light sources and the similarity matching result is similar, judging the paper currency to be a true currency, otherwise, judging the paper currency to be a false currency;
if the identification characteristic area is a dissimilar area in the images imaged by the two selected light sources and the similarity matching result is dissimilar, the paper currency is judged to be a true currency, otherwise, the paper currency is judged to be a false currency.
As shown in fig. 4A and 4B, in the embodiment of the present invention, two bands, namely, a green reflection spectrum G and an infrared reflection spectrum IR, are selected for analysis, the green reflection spectrum G is a reference light source, the infrared reflection spectrum IR is a comparison light source, and fig. 4A is a front image of an infrared channel of a 50 euro banknote; fig. 4B is a front image of a banknote G channel of 50 euro, and a similar region or a dissimilar region is selected from the images imaged by the two light sources as an identification feature region of the banknote, as shown in the figure, for example, corresponding 01 region and 21 region in the two images have a similar feature, that is, similar regions, and corresponding 02 region and 11 region in the two images have a dissimilar feature, that is, dissimilar regions.
The manufacturing process of the paper currency finally determines that the paper currency has the characteristics of different spectral imaging differences; however, such differences have certain laws which mainly have two performance characteristics, similar or dissimilar.
1. The similarity is that image similarity characteristics exist between imaging of a green reflection spectrum G (G L light Source) and imaging of an infrared reflection spectrum IR (IR L light Source), the similarity characteristics are particularly obvious in image structure characteristics, so that a calculation method of image similarity is adopted to evaluate similarity matching, the evaluation results are similar to True coins (True Money) and are not similar to false coins (Counterfeit Money), the 01 areas and the 21 areas have similarity characteristics, if the identification characteristic areas are the 01 areas and the 21 areas in the images imaged by the two selected light sources, the similarity matching results are similar, namely the paper Money is the True Money, and otherwise the paper Money is the false Money.
2. The dissimilarities are: the green reflection spectrum G imaging and the infrared reflection spectrum IR imaging have the characteristic of image dissimilarity; the similarity matching evaluation can be carried out by using the same image similarity calculation method; the evaluation results are not similar to be true coins, and the similar results are false coins; the 02 area and the 11 area have dissimilar characteristics, if the identification characteristic area is the 02 area and the 11 area in the images imaged by the two selected light sources, the similarity matching result is dissimilar, namely the paper currency is a true currency, otherwise, the paper currency is a false currency.
Finally, aiming at the identification area, comparing the similarity identification result of the green reflection spectrum G imaging and the infrared reflection spectrum IR imaging with the similarity experience result of the genuine currency; thereby judging whether the identified paper money is a counterfeit money or not, and achieving the purpose of true identification and false identification; if the counterfeit money is detected, the money counter performs voice alarm of the counterfeit money, the motor stops, and the interface flicking frame alarm prompt operation is performed.
Because the size of the bill opening of the bill counter is larger than that of the paper money, the position of the paper money entering the bill counter equipment has certain randomness, and the imaging angle of the paper money imaging image through the CIS also has randomness; the banknote identification algorithm needs to be established in the confirmed banknote image area, so that the positioning of the corner point position and the image affine transformation need to be carried out on the banknote position in the imaging image, specifically:
after acquiring a multispectral imaging image of the paper currency to be identified, positioning the corner point position of the paper currency in the acquired imaging image to obtain the four corner point positions of the paper currency in the imaging image;
obtaining affine transformation parameters through the four corner positions, carrying out affine transformation on the images, and transforming random imaging images into complete paper money images which are in a target pixel size and are in a regular shape, taking the complete paper money images as imaging images, and obtaining imaging images of paper money to be identified by a reference light source and a comparison light source from the images as reference light source imaging images and comparison light source imaging images.
The method comprises the following steps of carrying out corner position positioning on the position of a paper currency in an acquired imaging image, and specifically obtaining the positions of four corner points of the position of the paper currency in the imaging image:
finding four boundaries of the paper money in the image according to the gray difference between the paper money in the image and the background;
after finding the four boundaries of the paper money, fitting boundary points of the four boundary areas to obtain boundary straight lines of the four boundaries;
and calculating intersection points of the boundary straight lines intersected every two to obtain four intersection points, namely the positions of four corner points of the paper money position.
The affine transformation formula is as follows:
the function of the affine transformation is a linear transformation from two-dimensional coordinates to two-dimensional coordinates, and the 3 x 3 transformation matrix in the middle of the above equation is defined as matrix a, the last action of which is (0, 0, 1). The transformation matrix transforms the original coordinates (x, y) into new coordinates (x ', y').
As shown in fig. 3A, 3B, and 4B, a specific process of positioning corner positions and affine transformation of an image is shown.
Fig. 3A is a diagram illustrating positioning of an original imaging image acquired by the CIS image acquisition module, and fig. 3B is a diagram illustrating region calibration for performing corner point positioning on a banknote region of the original imaging image.
FIG. 3B is marked as four corner points of a two-dimensional graph of the paper currency, coordinates of three corner points, namely, an upper left corner point, an upper right corner point and a lower right corner point, are taken as three sets of parameters of original coordinates, FIG. 4B is a complete paper currency image of an affine image (the size of a 50-Euro standard rectangle is defined as 600 pixels of a long side and 240 pixels of a short side), and the coordinates of the three corner points, namely, the upper left corner point, the upper right corner point and the lower right corner point, are also taken as three sets of parameters of new.
Any point on the paper money can obtain a new coordinate value through affine transformation and a matrix A; thereby realizing the conversion of the acquired non-standard image to the standard rectangle size.
As a second embodiment of the present invention, there is provided an apparatus for banknote authentication based on a multispectral image similarity algorithm, the apparatus including: the device comprises an image acquisition module, a reference light source imaging image and comparison light source imaging image acquisition module, an identification characteristic region selection module and an image similarity identification module;
the image acquisition module is used for acquiring a multispectral imaging image of the paper money to be identified; acquiring multispectral images of an image sensor based on an embedded circuit processing platform of financial equipment; the financial machine and tool equipment can adopt relevant electromechanical platforms such as a sorter or a cash counter; the Image Sensor can adopt a Charge Coupled Device (CCD) or a Contact Image Sensor (CIS) and other related Image acquisition modules;
the reference light source imaging image and comparison light source imaging image acquisition module is used for selecting light sources of two spectral bands, wherein the light source of one spectral band is used as a reference light source, the light source of the other spectral band is used as a comparison light source, and imaging images of the reference light source and the comparison light source, namely a reference light source imaging image and a comparison light source imaging image, are respectively acquired from a multi-spectral imaging image;
the identification characteristic region selection module is used for selecting similar regions and/or dissimilar regions from images imaged by two light sources, namely a reference light source and a comparison light source, as identification characteristic regions of the paper money;
the image similarity identification module is used for carrying out image similarity calculation on identification characteristic areas of the images imaged by the two light sources to obtain a similarity matching result, and judging the authenticity of the paper money according to the similarity matching result.
The image similarity identification module specifically calculates the image similarity of the identification characteristic regions of the images imaged by the two light sources by:
because the light source of each spectrum section has background interference or noise interference, in order to eliminate the influence of the background or noise interference, a similarity characteristic threshold value threshold _ Ref _ light _ source of an identification characteristic region selected from an image imaged by a reference light source is firstly calculated, and similarity calculation is carried out on the identification characteristic region selected from the image imaged by the light source based on the guidance of the image similarity characteristic threshold value threshold _ Ref _ light _ source, wherein the specific formula is as follows:
wherein, threshold _ Ref _ light _ source is a similarity feature threshold of an identification feature area selected from an image imaged by a reference light source, and the feature threshold reflects a foreground feature of the reference light source image; the Com _ light _ source is the image characteristics of an identification characteristic area selected from the image imaged by the contrast light source according to the foreground characteristics of the image of the reference light source; the foreground characteristics of the light sources can be calculated and compared; threshold _ true _ money is an image experience value of a region corresponding to the true coin image, comparison is carried out according to the foreground characteristic of the comparison light source and the foreground experience value of the true coin comparison light source, the similarity of the test sample and the true coin can be compared and evaluated, and the feature of the test sample is evaluated according to the feature of the true coin; sim is the image similarity matching result.
The image similarity identification module judges the authenticity of the paper money according to the similarity matching result and specifically comprises the following steps:
if the identification characteristic area is a similar area in the images imaged by the two selected light sources and the similarity matching result is similar, judging the paper currency to be a true currency, otherwise, judging the paper currency to be a false currency;
if the identification characteristic area is a dissimilar area in the images imaged by the two selected light sources and the similarity matching result is dissimilar, the paper currency is judged to be a true currency, otherwise, the paper currency is judged to be a false currency.
Because the size of the bill opening of the bill counter is larger than that of the paper money, the position of the paper money entering the bill counter equipment has certain randomness, and the imaging angle of the paper money imaging image through the CIS also has randomness; the identification algorithm of the paper currency needs to be established in the confirmed paper currency image area, so that the position of an angular point and the image affine transformation of the paper currency in the imaging image need to be positioned by an image angular point positioning module and an image affine transformation module;
the image corner positioning module is used for positioning the corner position of the paper currency in the acquired imaging image before selecting the identification characteristic region to obtain the positions of four corners of the paper currency in the imaging image;
the image affine transformation module is used for obtaining affine transformation parameters through the obtained four corner positions, carrying out affine transformation on the images, transforming random imaging images into complete paper money images which are in a target pixel size and are in a regular shape, taking the complete paper money images as imaging images, and obtaining imaging images of paper money to be identified by a reference light source and a comparison light source from the images as reference light source imaging images and comparison light source imaging images.
The image corner positioning module is used for positioning corner positions of the acquired banknote positions in the imaging image, and the method for acquiring the four corner positions of the banknote positions in the imaging image specifically comprises the following steps:
finding four boundaries of the paper money in the image according to the gray difference between the paper money in the image and the background;
after finding the four boundaries of the paper money, fitting boundary points of the four boundary areas to obtain boundary straight lines of the four boundaries;
and calculating intersection points of the boundary straight lines intersected every two to obtain four intersection points, namely the positions of four corner points of the paper money position.
The affine transformation formula is as follows:
wherein (Tx, T)y) Indicating the amount of translation, while the parameters a1-a4 reflect changes in image rotation, scaling, etc.
Preferably, the invention also comprises an identification result processing module, wherein the identification result processing module is used for carrying out voice alarm, motor brake and interface pop-up frame alarm prompt operation on the counterfeit money when the counterfeit money is identified.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (10)
1.A method for identifying paper currency based on a multispectral image similarity algorithm is characterized by comprising the following steps:
step 1, acquiring a multispectral imaging image of a paper currency to be identified, selecting two spectral bands of light sources, wherein one spectral band of the light source is used as a reference light source, the other spectral band of the light source is used as a comparison light source, and respectively acquiring imaging images of the paper currency to be identified by the reference light source and the comparison light source, namely a reference light source imaging image and a comparison light source imaging image;
step 2, selecting similar areas and/or dissimilar areas from images imaged by two light sources, namely a reference light source imaging image and a comparison light source imaging image, as identification characteristic areas of the two light source imaging images;
and 3, performing image similarity calculation on the reference light source imaging image and the identification characteristic area of the comparison light source imaging image to obtain a similarity matching result, and judging the authenticity of the paper money according to the similarity matching result.
2. The method for banknote identification based on the multispectral image similarity algorithm according to claim 1, wherein in the step 3, the image similarity calculation for the identification feature areas of the reference light source imaging image and the comparison light source imaging image is specifically as follows:
because the light source of each spectrum section has background interference or noise interference, in order to eliminate the influence of the background or noise interference, a similarity characteristic threshold value threshold _ Ref _ light _ source of an identification characteristic region selected from an image imaged by a reference light source is firstly calculated, and similarity calculation is carried out on the identification characteristic region selected from the image imaged by the light source based on the guidance of the image similarity characteristic threshold value threshold _ Ref _ light _ source, wherein the specific formula is as follows:
wherein, the threshold _ Ref _ light _ source is a similarity feature threshold of an authentication feature region selected from an image imaged according to a reference light source, the Com _ light _ source is an image feature of the authentication feature region selected from the image imaged by a contrast light source, the threshold _ true _ money is an image experience value of a region corresponding to a genuine banknote image, and sim is an image similarity matching result.
3. The method for banknote authentication based on the multispectral image similarity algorithm as claimed in claim 1, wherein the step 3 of determining the authenticity of the banknote based on the similarity matching result specifically comprises:
if the identification characteristic area is a similar area in the images imaged by the two selected light sources and the similarity matching result is similar, judging the paper currency to be a true currency, otherwise, judging the paper currency to be a false currency;
if the identification characteristic area is a dissimilar area in the images imaged by the two selected light sources and the similarity matching result is dissimilar, the paper currency is judged to be a true currency, otherwise, the paper currency is judged to be a false currency.
4. The method for banknote authentication based on multispectral image similarity algorithm according to claim 1, further comprising:
after collecting a multispectral imaging image of a paper currency to be identified, positioning the corner point position of the paper currency in the acquired multispectral imaging image to obtain the four corner point positions of the paper currency in the imaging image;
affine transformation parameters are obtained through the four corner positions, affine transformation is carried out on the image, and therefore the random imaging image is transformed into a complete banknote image which is in a target pixel size and is in a regular shape, and the complete banknote image is used as an imaging image.
5. The method according to claim 4, wherein the positioning of the corner positions of the banknote positions in the multispectral imaging image to obtain the four corner positions of the banknote positions in the imaging image comprises:
finding four boundaries of the paper money in the image according to the gray difference between the paper money in the image and the background;
after finding the four boundaries of the paper money, fitting boundary points of the four boundary areas to obtain boundary straight lines of the four boundaries;
and calculating intersection points of the boundary straight lines intersected every two to obtain four intersection points, namely the positions of four corner points of the paper money position.
6. An apparatus for banknote authentication based on a multispectral image similarity algorithm, the apparatus comprising: the device comprises an image acquisition module, a reference light source imaging image and comparison light source imaging image acquisition module, an identification characteristic region selection module and an image similarity identification module;
the image acquisition module is used for acquiring a multispectral imaging image of the paper money to be identified;
the reference light source imaging image and comparison light source imaging image acquisition module is used for selecting light sources of two spectral bands, wherein the light source of one spectral band is used as a reference light source, the light source of the other spectral band is used as a comparison light source, and imaging images of the reference light source and the comparison light source, namely a reference light source imaging image and a comparison light source imaging image, are respectively acquired from a multi-spectral imaging image;
the identification characteristic region selection module is used for taking a similar region and/or a dissimilar region imaged by two light sources, namely a reference light source imaging image and a comparison light source imaging image, as identification characteristic regions of the two light source imaging images;
the image similarity identification module is used for carrying out image similarity calculation on identification characteristic areas of the images imaged by the two light sources to obtain a similarity matching result, and judging the authenticity of the paper money according to the similarity matching result.
7. The device for banknote discrimination based on multispectral image similarity algorithm according to claim 6, wherein the image similarity discrimination module performs image similarity calculation on the discrimination feature areas of the reference light source imaging image and the comparison light source imaging image by specifically:
because the light source of each spectrum section has background interference or noise interference, in order to eliminate the influence of the background or noise interference, a similarity characteristic threshold value threshold _ Ref _ light _ source of an identification characteristic region selected from an image imaged by a reference light source is firstly calculated, and similarity calculation is carried out on the identification characteristic region selected from the image imaged by the light source based on the guidance of the image similarity characteristic threshold value threshold _ Ref _ light _ source, wherein the specific formula is as follows:
wherein, the threshold _ Ref _ light _ source is a similarity feature threshold of an authentication feature region selected from an image imaged according to a reference light source, the Com _ light _ source is an image feature of the authentication feature region selected from the image imaged by a contrast light source, the threshold _ true _ money is an image experience value of a region corresponding to a genuine banknote image, and sim is an image similarity matching result.
8. The device for banknote authentication based on multispectral image similarity algorithm as claimed in claim 6, wherein the image similarity authentication module determines the authenticity of the banknote based on the similarity matching result by:
if the identification characteristic area is a similar area in the images imaged by the two selected light sources and the similarity matching result is similar, judging the paper currency to be a true currency, otherwise, judging the paper currency to be a false currency;
if the identification characteristic area is a dissimilar area in the images imaged by the two selected light sources and the similarity matching result is dissimilar, the paper currency is judged to be a true currency, otherwise, the paper currency is judged to be a false currency.
9. The device for banknote authentication based on multispectral image similarity algorithm according to claim 6, further comprising an image corner point positioning module and an image affine transformation module;
the image corner positioning module is used for positioning the corner positions of the paper currency in the acquired multispectral imaging image after acquiring the multispectral imaging image of the paper currency to be identified, so as to obtain the positions of four corner points of the paper currency in the imaging image;
the image affine transformation module is used for obtaining affine transformation parameters through the obtained four corner positions, carrying out affine transformation on the image, transforming the random imaging image into a complete banknote image which is in a target pixel size and is in a regular shape, and taking the complete banknote image as an imaging image.
10. The device for banknote validation based on multispectral image similarity algorithm as claimed in claim 9, wherein the image corner positioning module performs corner position positioning on the banknote position in the multispectral imaging image, and the four corner positions of the banknote position in the imaging image are obtained by:
finding four boundaries of the paper money in the image according to the gray difference between the paper money in the image and the background;
after finding the four boundaries of the paper money, fitting boundary points of the four boundary areas to obtain boundary straight lines of the four boundaries;
and calculating intersection points of the boundary straight lines intersected every two to obtain four intersection points, namely the positions of four corner points of the paper money position.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010270998.2A CN111462392A (en) | 2020-04-08 | 2020-04-08 | Method and device for identifying paper money based on multispectral image similarity algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010270998.2A CN111462392A (en) | 2020-04-08 | 2020-04-08 | Method and device for identifying paper money based on multispectral image similarity algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111462392A true CN111462392A (en) | 2020-07-28 |
Family
ID=71681225
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010270998.2A Pending CN111462392A (en) | 2020-04-08 | 2020-04-08 | Method and device for identifying paper money based on multispectral image similarity algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111462392A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114821888A (en) * | 2022-05-06 | 2022-07-29 | 广州中智融通金融科技有限公司 | Counterfeit money management method, system, electronic device and readable medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101533536A (en) * | 2009-04-15 | 2009-09-16 | 谭一舟 | Duble light path imaging device used for discriminating spectrum difference and gravure characteristics of notes |
| AU2009238260A1 (en) * | 2009-11-13 | 2011-06-02 | Canon Kabushiki Kaisha | Forgery detection using finger print |
| CN107067533A (en) * | 2017-04-14 | 2017-08-18 | 深圳怡化电脑股份有限公司 | The method and device that a kind of bank note differentiates |
| CN108320373A (en) * | 2017-01-17 | 2018-07-24 | 深圳怡化电脑股份有限公司 | A kind of method and device of the detection of guiding against false of paper currency mark |
-
2020
- 2020-04-08 CN CN202010270998.2A patent/CN111462392A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101533536A (en) * | 2009-04-15 | 2009-09-16 | 谭一舟 | Duble light path imaging device used for discriminating spectrum difference and gravure characteristics of notes |
| AU2009238260A1 (en) * | 2009-11-13 | 2011-06-02 | Canon Kabushiki Kaisha | Forgery detection using finger print |
| CN108320373A (en) * | 2017-01-17 | 2018-07-24 | 深圳怡化电脑股份有限公司 | A kind of method and device of the detection of guiding against false of paper currency mark |
| CN107067533A (en) * | 2017-04-14 | 2017-08-18 | 深圳怡化电脑股份有限公司 | The method and device that a kind of bank note differentiates |
Non-Patent Citations (1)
| Title |
|---|
| 蒋天瑜: "基于多光谱图像的纸币防伪鉴定技术研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114821888A (en) * | 2022-05-06 | 2022-07-29 | 广州中智融通金融科技有限公司 | Counterfeit money management method, system, electronic device and readable medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP1490828B1 (en) | Currency verification | |
| US9824268B2 (en) | Media item validation | |
| CN106952393B (en) | Paper money identification method and device, electronic equipment and storage medium | |
| JP2006301881A (en) | Currency identifying device, currency identifying method and currency identifying program | |
| CN108921831B (en) | Stained coin identification method based on image processing technology | |
| CN107134047B (en) | White watermark detection method and device | |
| KR20070093210A (en) | Recognizing the denomination of a note using wavelet transform | |
| JP5174917B2 (en) | Coin damage determination apparatus and damage determination method | |
| CN101583978B (en) | Apparatus for media recognition and method for media kind distinction with the same | |
| CN111462392A (en) | Method and device for identifying paper money based on multispectral image similarity algorithm | |
| CN106296975B (en) | method and device for identifying face value of dollar paper money | |
| CN106447904B (en) | Method and device for identifying paper money | |
| CN106447908B (en) | Paper money counterfeit distinguishing method and device | |
| KR101740592B1 (en) | Method for recoginizing universal bill | |
| KR101385358B1 (en) | Apparatus and method for medium recognition, auto teller machine | |
| KR100888674B1 (en) | Method for measuring similarity using frenquency range | |
| US20060115140A1 (en) | Pattern identification method and identification device | |
| CN110969757A (en) | Multi-country banknote type rapid identification technology | |
| JP3718613B2 (en) | Method and apparatus for determining coin authenticity based on image | |
| CN108428279B (en) | Paper money identification method and device | |
| JP2004220205A (en) | Identification method and identification device for circular matter | |
| CN111627145B (en) | Method and device for identifying fine hollow image-text of image | |
| WO2017046561A1 (en) | Method and system for detecting staining | |
| JP3854962B2 (en) | Pattern identification method and identification apparatus | |
| JP4691644B2 (en) | Medium position detection method and medium recognition apparatus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200728 |