CN117314897B - Method and device for discriminating distortion of banknote image - Google Patents
Method and device for discriminating distortion of banknote image Download PDFInfo
- Publication number
- CN117314897B CN117314897B CN202311595338.1A CN202311595338A CN117314897B CN 117314897 B CN117314897 B CN 117314897B CN 202311595338 A CN202311595338 A CN 202311595338A CN 117314897 B CN117314897 B CN 117314897B
- Authority
- CN
- China
- Prior art keywords
- image
- standard
- distortion
- images
- sample
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims description 35
- 238000012545 processing Methods 0.000 claims abstract description 21
- 238000010586 diagram Methods 0.000 claims description 52
- 238000004364 calculation method Methods 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 4
- 238000012850 discrimination method Methods 0.000 abstract description 3
- 238000002474 experimental method Methods 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000010186 staining Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity 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
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
Abstract
The invention provides a distortion discrimination method and a distortion discrimination device for paper money images, comprising the following steps: establishing a distortion recognition model according to the first sample paper money and the second sample paper money; processing paper currency to be identified based on standard image types of the standard images to obtain X contrast images; respectively calculating the similarity of the feature structures to be identified between X contrast images and corresponding X standard images, wherein M feature structure similarities to be identified exist between each contrast image and the corresponding standard image; if the similarity of any feature structure to be identified is greater than the similarity of the corresponding standard feature structure, the paper currency to be identified is non-distorted paper currency, otherwise, the paper currency to be identified is local distorted paper currency. The technology is suitable for recognizing the distortion condition of the banknote image with complex background, and has stronger universality, smaller calculated amount and higher stability.
Description
Technical Field
The invention relates to the technical field of image information processing, in particular to a method and a device for judging distortion of paper money images.
Background
"distortion" refers to the difficulty in distinguishing image features. Currently, the following three distortion discrimination methods are commonly used:
the method 1 comprises the steps of respectively carrying out mathematical transformation on characteristic values of a to-be-identified paper money image and an original paper money image in a large number of preset target areas to obtain a distortion judging result, specifically, respectively carrying out transformation on the characteristics of the to-be-identified paper money image and the characteristics of the original paper money image, judging differences between the to-be-identified paper money image and the original image through the transformation, and carrying out distortion judgment. The drawbacks of this approach are: a plurality of target areas are required to be preset to accurately realize distortion identification, and the calculated amount is large.
And 2, judging the distortion condition by calculating the similarity. In recent years, there are many research methods related to similarity calculation between features, such as MD5 value Algorithm (Message-Digest Algorithm), histogram similarity method, perceptual hash similarity method, and mean hash similarity method, but these algorithms have strict requirements on image size, and experiments prove that these methods have larger error in distortion discrimination in banknote images with complex background.
In addition, the method 3 is also commonly used for computing the similarity by using a LPIPS (Learned Perceptual Image Patch Similarity) algorithm for deeply learning the network verification index, and the method is close to human actual perception, so that the accuracy of computing the similarity is improved to a certain extent, but the computing amount is large.
Disclosure of Invention
Based on the above, the invention aims to provide a distortion judging method and a distortion judging device for banknote images, which are suitable for recognizing the distortion condition of banknote images with complicated background, and have the advantages of stronger universality, smaller calculated amount and higher stability.
In a first aspect, an embodiment of the present invention provides a distortion discriminating method of a banknote image, the distortion discriminating method including: s1: establishing a distortion recognition model according to a large number of first sample paper money and second sample paper money, wherein X standard images and standard feature structure similarity are stored in the distortion recognition model; each standard image corresponds to one standard image type, and each standard image corresponds to M standard feature structure similarity; s2: processing paper currency to be identified based on standard image types of the standard images to obtain X contrast images; s3: respectively calculating the similarity of the feature structures to be identified between X contrast images and corresponding X standard images, wherein M feature structure similarities to be identified exist between each contrast image and the corresponding standard image; s4: if the similarity of any feature structure to be identified is greater than the similarity of the corresponding standard feature structure, the paper currency to be identified is non-distorted paper currency, otherwise, the paper currency to be identified is local distorted paper currency.
Further, S1 includes: s11: processing the front and back sides of each first sample banknote with a plurality of different lights to obtain a red light positive diagram, a red light negative diagram, a green light positive diagram, a green light negative diagram, a Lan Guangzheng diagram and a blue light negative diagram; s12: integrating the distortion areas of all the red light positive graphs of the first sample paper money to obtain a red light target area set locator RZ, wherein the distortion areas are rectangular and are represented by two coordinates of the lower left corner and the upper right corner, and the number of the distortion areas of the red light positive graphs is M; s13: the same steps as those of S12 are respectively carried out on the red light reverse graph, the green light forward graph, the green light reverse graph, the Lan Guangzheng graph and the blue light reverse graph of the first sample paper money, and locationRF, locationGZ, locationGF, locationBZ, locationBF are respectively obtained; s14: storing the X target area sets with the largest number of distortion areas stored in locationRZ, locationRF, locationGZ, locationGF, locationBZ, locationBF, and confirming the image type corresponding to each target area set, wherein the image type is defined as a target image type; s15: calculating the signal-to-noise ratio between any two second sample banknote images based on the X target image types to obtain X groups of first quasi-standard images; s16: based on X target image types, calculating the overall image similarity between any two second sample banknote images to obtain X groups of second quasi-standard images; s17: calculating the overall image similarity between any two of 2 first quasi-standard images and 2 second quasi-standard images corresponding to each target image type, and defining 2 second sample banknote images corresponding to the minimum overall image similarity as result P, wherein the standard image is any one of the result P; s18: and calculating the standard feature structure similarity between the two second sample banknote images in the resultPs based on the M distortion areas, wherein each group of resultPs corresponds to the M standard feature structure similarity.
Further, S12 includes: s121: presetting Q key attention areas; s122: for the red light positive diagram of any one first sample paper money, manually determining whether a key attention area of the red light positive diagram of the first sample paper money is a quasi-distortion area or not; s123: if the number of times that any one important attention area is confirmed as a quasi-distortion area is larger than a preset threshold value, confirming the important attention area as a distortion area; s124: and recording the coordinates of the left lower corner and the right upper corner of all the distortion areas to obtain a red light target area set locationRZ.
Further, if the target image types are red light inverse image, lan Guangzheng image and blue light inverse image respectively; s15 includes: s151: processing the back surface of each second sample banknote with red light to obtain a P Zhang Gongguang reverse chart, wherein the second sample banknote has P sheets; s152: calculating the signal-to-noise ratio between the red light inverse diagrams of any two second sample paper money to obtainA signal-to-noise ratio; s153: taking two second sample banknote images corresponding to the minimum signal-to-noise ratio as first pseudo-standard images; s154: processing the front surface of each second sample banknote by blue light to obtain a P Zhang Languang positive image, and processing the back surface by blue light to obtain a P Zhang Languang negative image; and repeating S152 and S153 on the P Zhang Languang positive diagram and the P Zhang Languang negative diagram to obtain a plurality of groups of first quasi-standard images, wherein each group has two second sample banknote images.
Further, the method for calculating the signal-to-noise ratio in S152 is as follows:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein m and n are the width and the height of the second sample paper money, I is the gray map matrix of the second sample paper money, K is the noise map matrix of the second sample paper money, and MSE is the mean square error;
the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>The PSNR is the signal-to-noise ratio for the maximum pixel value of the gray map of the second sample banknote.
Further, S16 includes: and sequentially and respectively calculating red light reverse graphs of any two second sample paper currencies, lan Guangzheng graphs of any two second sample paper currencies and overall graph similarity between blue light reverse graphs of any two second sample paper currencies, and respectively taking two paper currency images corresponding to the maximum overall graph similarity to obtain three groups of second quasi-standard images, wherein two second sample paper currency images exist in each group.
Further, S3 includes: and respectively calculating the similarity of the feature structures to be identified between the red light inverse diagram of the standard image and the red light inverse diagram of the paper currency to be identified, the blue light positive diagram of the standard image and the Lan Guangzheng diagram of the paper currency to be identified, and the blue light inverse diagram of the standard image and the blue light inverse diagram of the paper currency to be identified, and then calculating M feature structure similarities to be identified between each comparison image and the corresponding standard image.
Further, the banknote to be identified is a partially distorted banknote or a non-distorted banknote.
Further, the first sample note is distorted to a much higher degree than the second sample note.
In a second aspect, an embodiment of the present invention provides a distortion discriminating apparatus for a banknote image, the distortion discriminating apparatus including: the model training module is used for establishing a distortion recognition model according to a large number of first sample paper money and second sample paper money, wherein X standard images and standard feature structure similarity are stored in the distortion recognition model; each standard image corresponds to one standard image type, and each standard image corresponds to M standard feature structure similarity; the banknote to be identified processing module is used for processing the banknote to be identified based on the standard image type of the standard image to obtain X contrast images; the similarity calculation module is used for calculating the similarity of the feature structures to be identified between the X comparison images and the corresponding X standard images respectively, wherein M feature structure similarities to be identified exist between each comparison image and the corresponding standard image; and the comparison module is used for judging whether the paper currency to be identified is non-distorted if the similarity of any feature structure to be identified is greater than the similarity of the corresponding standard feature structure, and judging that the paper currency to be identified is local distorted if the similarity of any feature structure to be identified is not the similarity of the corresponding standard feature structure.
The embodiment of the invention has the beneficial effects that:
the invention provides a distortion discrimination method and a distortion discrimination device for paper money images, comprising the following steps: s1: establishing a distortion recognition model according to the first sample paper money and the second sample paper money; s2: processing paper currency to be identified based on standard image types of the standard images to obtain X contrast images; s3: respectively calculating the similarity of the feature structures to be identified between X contrast images and corresponding X standard images, wherein M feature structure similarities to be identified exist between each contrast image and the corresponding standard image; s4: if the similarity of any feature structure to be identified is greater than the similarity of the corresponding standard feature structure, the paper currency to be identified is non-distorted paper currency, otherwise, the paper currency to be identified is local distorted paper currency. The technology is suitable for recognizing the distortion condition of the banknote image with complex background, and has stronger universality, smaller calculated amount and higher stability
Additional features and advantages of the invention will be set forth in the description which follows, or in part will be obvious from the description, or may be learned by practice of the invention.
In order to make the above objects, features and advantages of the present invention more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for discriminating distortion of a banknote image according to an embodiment of the present invention;
fig. 2 is a flowchart for establishing a distortion recognition model according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Examples
The invention provides a distortion discriminating method of paper money image.
The banknotes to be identified in this embodiment are partially distorted banknotes or non-distorted banknotes, but not entirely distorted banknotes. The embodiment of the invention is applied to identifying the paper currency with obvious distortion interference factors, wherein the distortion interference factors comprise: the graffiti covers the characteristics of the original pattern of the paper currency, the ink covers the characteristics of the original pattern of the paper currency, and the paper currency is damaged, so that the characteristics of the original pattern of the paper currency cannot be identified.
As shown in fig. 1, a flow chart of a distortion discriminating method of a banknote image according to the present embodiment is shown. A flow chart for creating a distortion recognition model is shown in fig. 2.
S1: establishing a distortion recognition model according to a large number of first sample paper money and second sample paper money, wherein X standard images and standard feature structure similarity are stored in the distortion recognition model; each standard image corresponds to a standard image type, and each standard image corresponds to M standard feature structure similarities.
The detailed procedure for creating the distortion recognition model is as follows.
S11: and processing the front side and the back side of each first sample paper money by using a plurality of different lights to obtain a red light positive diagram, a red light negative diagram, a green light positive diagram, a green light negative diagram, a Lan Guangzheng diagram and a blue light negative diagram.
Specifically, the front face of the first sample banknote is irradiated by red light, so that a red light positive image can be obtained, and other images can be obtained in the same way.
Specifically, the first sample banknote refers to a banknote with serious distortion in manual judgment, and each first sample banknote can be processed to obtain the 6 images.
Specifically, in the practical experimental process, the infrared positive, infrared negative, transmission positive, ultraviolet negative and the like of the paper money are modeled, but the effect is poor, so that the subsequent modeling operation is performed by using the red positive, red negative, green positive, green negative, lan Guangzheng and blue negative.
In particular, the characteristics of the distorted position are disturbed by different degrees, different categories of noise, where the noise comes from different degrees of graffiti and ink.
S12: and integrating the distortion areas of all the red light positive graphs of the first sample paper money to obtain a red light target area set locator RZ, wherein the distortion areas are rectangular and are represented by two coordinates of the lower left corner and the upper right corner, and the number of the distortion areas of the red light positive graphs is M.
Step S12 includes:
s121: q important attention areas are preset.
Here, the more Q is set, the more accurate the final result is, and the larger the calculation amount is, the more important the region of interest is preset as a rectangle.
S122: for the red light positive diagram of any one first sample banknote, manually determining whether a focus attention area of the red light positive diagram of the first sample banknote is a distortion simulating area.
Specifically, if the region is severely damaged, the distortion region is manually determined.
Specifically, here, how many first sample banknotes are collected, and how many red light positive charts can be obtained.
S123: if the number of times that any one important attention area is confirmed as the quasi-distorted area is larger than a preset threshold value, the important attention area is confirmed as the distorted area.
For example, there are 100 tens of thousands of red positive diagrams of first sample banknotes, in which one important region of interest is manually confirmed that the number of the pseudo-distorted regions is only 100, and the important region of interest is not the distorted region.
S124: and recording the coordinates of the left lower corner and the right upper corner of all the distortion areas to obtain a red light target area set locationRZ.
Specifically, herein, the number of distortion regions is defined as M.ltoreq.Q.
S13: the same steps as S12 are performed on the red reverse, green forward, green reverse, lan Guangzheng and blue reverse of the first sample banknote, respectively, resulting in locationRF, locationGZ, locationGF, locationBZ, locationBF.
S14: the X target area sets with the largest number of distortion areas stored in locationRZ, locationRF, locationGZ, locationGF, locationBZ, locationBF are saved, and the image type corresponding to each target area set is confirmed and defined as the target image type.
Specifically, a large number of experiments prove that x=3, S14 is 3 to reject, 3 operations are left to be performed subsequently, and a large number of experiments prove that locationRF, locationBZ, locationBF (the subsequent positions respectively correspond to position 1, position 2 and position 3) have a large number of distortion areas, and the subsequent operations are left to be performed. S15 and S16 are each performed based on x=3, locationRF, locationBZ, locationBF respectively.
S15: and calculating the signal-to-noise ratio between any two second sample banknote images based on the X target image types to obtain X groups of first quasi-standard images.
The second sample banknote is a newer banknote with a low degree of breakage and less staining.
S15 includes:
for example, x=3, and the 3 target image types are red inverse, lan Guangzheng, and blue inverse, respectively.
S151: the reverse side was treated with red light for each second sample banknote, resulting in a P Zhang Gongguang reverse image.
Wherein the second sample banknote has P sheets.
S152: calculating the signal-to-noise ratio between the red light inverse diagrams of any two second sample paper money to obtainAnd signal to noise ratio.
S153: and taking 2 second sample banknote images corresponding to the minimum signal-to-noise ratio as the first pseudo-standard image.
S154: the front side of each second sample banknote was treated with blue light to give a P Zhang Languang front side and the back side was treated with blue light to give a P Zhang Languang back side. And repeating S152 and S153 on the P Zhang Languang positive diagram and the P Zhang Languang negative diagram to obtain X (=3) groups of first pseudo-standard images, wherein each group comprises 2 second sample banknote images.
The method for calculating the signal-to-noise ratio in S152 is as follows:；
wherein m and n are the width and height of the second sample banknote, I is the gray map matrix of the second sample banknote, K is the noise map matrix of the second sample banknote, and MSE is the mean square error.
；
Wherein,the PSNR is the signal-to-noise ratio for the maximum pixel value of the gray map of the second sample banknote.
S16: and calculating the overall image similarity between any two second sample banknote images based on the X target image types to obtain X groups of second quasi-standard images.
S16 comprises the following steps: sequentially and respectively calculating red light reverse graphs of any two second sample paper currencies, lan Guangzheng graphs of any two second sample paper currencies and overall graph similarity between blue light reverse graphs of any two second sample paper currencies, and respectively taking 2 paper currency images corresponding to the maximum overall graph similarity, namely 3 groups of second quasi-standard images can be obtained, wherein 2 second sample paper currency images exist in each group. The calculation logic is the same as S15.
S17: and calculating the overall image similarity between any two of the 2 first quasi-standard images and 2 second quasi-standard images corresponding to each target image type, and defining 2 second sample banknote images corresponding to the minimum overall image similarity as a result P, wherein the standard image is any one of the result P.
In this way, x=3 standard images can be obtained, each standard image corresponds to a standard image type, and it should be noted that 3 standard images do not necessarily correspond to the same banknote.
The image corresponding to the minimum overall image similarity is taken as a standard image in S17 (i.e. the minimum similarity is found in the similar quasi-standard image to perform the subsequent operation), because the smaller the similarity is, the greater the distortion degree of the standard image is, and the more accurate the recognition result is obtained when the similarity is compared with the banknote to be recognized (i.e. steps S3 and S4).
S18: and calculating the standard feature structure similarity between the two second sample banknote images in the resultPs based on the M distortion areas, wherein each group of resultPs corresponds to the M standard feature structure similarity.
S2: and processing the paper currency to be identified based on the standard image type of the standard image to obtain X contrast images.
For example, if x=3, for example, the obtained standard image is a red inverse image, a Lan Guangzheng image and a blue inverse image, then the red inverse image, the Lan Guangzheng image and the blue inverse image of the banknote to be identified are also obtained correspondingly as comparison images.
S3: and respectively calculating the similarity of the feature structures to be identified between the X contrast images and the corresponding X standard images, wherein M feature structure similarities to be identified exist between each contrast image and the corresponding standard image.
For example, following the example of S2, the specific operation of S3 is: and respectively calculating the similarity of the feature structures to be identified between the red light inverse diagram of the standard image and the red light inverse diagram of the paper currency to be identified, the blue light positive diagram of the standard image and the Lan Guangzheng diagram of the paper currency to be identified, and the blue light inverse diagram of the standard image and the blue light inverse diagram of the paper currency to be identified, so that 3M feature structure similarities to be identified can be obtained, and specifically, M feature structure similarities to be identified can be calculated between each comparison image and the corresponding standard image.
S4: if the similarity of any feature structure to be identified is greater than the similarity of the corresponding standard feature structure, the paper currency to be identified is non-distorted paper currency, otherwise, the paper currency to be identified is local distorted paper currency.
Specifically, the "corresponding" refers to that the comparison is performed according to the image type and the distortion areas, and if there are M distortion areas for any image type, the comparison can be performed M times, and if there are 1 times of feature structure similarity to be identified that is greater than the corresponding standard feature structure similarity, the banknote to be identified is a non-distorted banknote.
The beneficial effects of this embodiment are:
1. through the calculation and comparison of the signal-to-noise ratio and the structural similarity in the steps S15-S17, the use of a constant threshold value with non-universality can be reduced, the calculated amount is reduced, and meanwhile, the recognition accuracy is improved.
2. Step S3 and step S4 increase the robustness of the distortion recognition algorithm by calculating the similarity between each distortion region of the standard image and the image to be recognized.
3. The modeling process S11, S15, S16 makes the recognition result more accurate by different utilization of a large number of older first sample notes and newer second sample notes.
4. S12, S13 and S14 accurately reject the images based on the number of the distortion areas, so that the calculated amount can be reduced, and the accuracy of the judging result is ensured.
Examples
The present embodiment provides a distortion discriminating apparatus of a banknote image, the distortion discriminating apparatus including:
the model training module is used for establishing a distortion recognition model according to a large number of first sample paper money and second sample paper money, wherein X standard images and standard feature structure similarity are stored in the distortion recognition model; each standard image corresponds to a standard image type, and each standard image corresponds to M standard feature structure similarities.
And the banknote processing module to be identified is used for processing the banknote to be identified based on the standard image type of the standard image to obtain X contrast images.
The similarity calculation module is used for calculating the similarity of the feature structures to be identified between the X comparison images and the corresponding X standard images respectively, wherein M feature structure similarities to be identified exist between each comparison image and the corresponding standard image.
And the comparison module is used for judging whether the paper currency to be identified is non-distorted if the similarity of any feature structure to be identified is greater than the similarity of the corresponding standard feature structure, and judging that the paper currency to be identified is local distorted if the similarity of any feature structure to be identified is not the similarity of the corresponding standard feature structure.
The distortion discriminating device for banknote images provided by the embodiment of the invention has the same implementation principle and technical effects as those of the embodiment of the distortion discriminating method for banknote images, and for the sake of brief description, reference may be made to the corresponding contents of the embodiment of the method.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention.
Claims (8)
1. A distortion discriminating method of a banknote image, the distortion discriminating method comprising:
s1: establishing a distortion recognition model according to the first sample paper money and the second sample paper money, wherein X standard images and standard feature structure similarity are stored in the distortion recognition model; each standard image corresponds to one standard image type, and each standard image corresponds to M standard feature structure similarity; the distortion degree of the first sample paper money is far higher than that of the second sample paper money;
s2: processing paper currency to be identified based on standard image types of the standard images to obtain X contrast images;
s3: respectively calculating the similarity of the feature structures to be identified between X contrast images and corresponding X standard images, wherein M feature structure similarities to be identified exist between each contrast image and the corresponding standard image;
s4: if the similarity of any feature structure to be identified is greater than the similarity of the corresponding standard feature structure, the paper currency to be identified is non-distorted paper currency, otherwise, the paper currency to be identified is local distorted paper currency;
s1 comprises the following steps:
s11: processing the front and back sides of each first sample banknote with a plurality of different lights to obtain a red light positive diagram, a red light negative diagram, a green light positive diagram, a green light negative diagram, a Lan Guangzheng diagram and a blue light negative diagram;
s12: integrating the distortion areas of all the red light positive graphs of the first sample paper money to obtain a red light target area set locator RZ, wherein the distortion areas are rectangular and are represented by two coordinates of the lower left corner and the upper right corner, and the number of the distortion areas of the red light positive graphs is M;
s13: the same steps as those of S12 are respectively carried out on the red light reverse graph, the green light forward graph, the green light reverse graph, the Lan Guangzheng graph and the blue light reverse graph of the first sample paper money, and locationRF, locationGZ, locationGF, locationBZ, locationBF are respectively obtained;
s14: storing the X target area sets with the largest number of distortion areas stored in locationRZ, locationRF, locationGZ, locationGF, locationBZ, locationBF, and confirming the image type corresponding to each target area set, wherein the image type is defined as a target image type;
s15: calculating the signal-to-noise ratio between any two second sample banknote images based on the X target image types to obtain X groups of first quasi-standard images;
s16: based on X target image types, calculating the overall image similarity between any two second sample banknote images to obtain X groups of second quasi-standard images;
s17: calculating the overall image similarity between any two of 2 first quasi-standard images and 2 second quasi-standard images corresponding to each target image type, and defining 2 second sample banknote images corresponding to the minimum overall image similarity as result P, wherein the standard image is any one of the result P;
s18: and calculating the standard feature structure similarity between the two second sample banknote images in the resultPs based on the M distortion areas, wherein each group of resultPs corresponds to the M standard feature structure similarity.
2. The method of discriminating distortion of a banknote image according to claim 1 wherein S12 includes:
s121: presetting Q key attention areas;
s122: for the red light positive diagram of any one first sample paper money, manually determining whether a key attention area of the red light positive diagram of the first sample paper money is a quasi-distortion area or not;
s123: if the number of times that any one important attention area is confirmed as a quasi-distortion area is larger than a preset threshold value, confirming the important attention area as a distortion area;
s124: and recording the coordinates of the left lower corner and the right upper corner of all the distortion areas to obtain a red light target area set locationRZ.
3. The method for discriminating distortion of a banknote image according to claim 2 wherein if the type of the target image is red, lan Guangzheng, blue; characterized in that S15 comprises:
s151: processing the back surface of each second sample banknote with red light to obtain a P Zhang Gongguang reverse chart, wherein the second sample banknote has P sheets;
s152: calculating the signal-to-noise ratio between the red light inverse diagrams of any two second sample paper money to obtainA signal-to-noise ratio;
s153: taking two second sample banknote images corresponding to the minimum signal-to-noise ratio as first pseudo-standard images;
s154: processing the front surface of each second sample banknote by blue light to obtain a P Zhang Languang positive image, and processing the back surface by blue light to obtain a P Zhang Languang negative image; and repeating S152 and S153 on the P Zhang Languang positive diagram and the P Zhang Languang negative diagram to obtain a plurality of groups of first quasi-standard images, wherein each group has two second sample banknote images.
4. A method for discriminating distortion of a banknote image according to claim 3 wherein the method for calculating the signal to noise ratio in S152 is as follows:
；
wherein m and n are the width and the height of the second sample paper money, I is the gray map matrix of the second sample paper money, K is the noise map matrix of the second sample paper money, and MSE is the mean square error;
；
the PSNR is the signal-to-noise ratio for the maximum pixel value of the gray map of the second sample banknote.
5. The method of discriminating distortion of a banknote image according to claim 4 wherein S16 includes: and sequentially and respectively calculating red light reverse graphs of any two second sample paper currencies, lan Guangzheng graphs of any two second sample paper currencies and overall graph similarity between blue light reverse graphs of any two second sample paper currencies, and respectively taking two paper currency images corresponding to the maximum overall graph similarity to obtain three groups of second quasi-standard images, wherein two second sample paper currency images exist in each group.
6. The method for discriminating distortion of a banknote image according to claim 5 wherein S3 includes:
and respectively calculating the similarity of the feature structures to be identified between the red light inverse diagram of the standard image and the red light inverse diagram of the paper currency to be identified, the blue light positive diagram of the standard image and the Lan Guangzheng diagram of the paper currency to be identified, and the blue light inverse diagram of the standard image and the blue light inverse diagram of the paper currency to be identified, and then calculating M feature structure similarities to be identified between each comparison image and the corresponding standard image.
7. The method according to claim 1, wherein the bill to be recognized is a partially distorted bill or a non-distorted bill.
8. A distortion discriminating apparatus for a banknote image, the distortion discriminating apparatus comprising:
the model training module is used for establishing a distortion recognition model according to a large number of first sample paper money and second sample paper money, wherein X standard images and standard feature structure similarity are stored in the distortion recognition model; each standard image corresponds to one standard image type, each standard image corresponds to M standard feature structure similarity, and the distortion degree of the first sample paper money is far higher than that of the second sample paper money;
the banknote to be identified processing module is used for processing the banknote to be identified based on the standard image type of the standard image to obtain X contrast images;
the similarity calculation module is used for calculating the similarity of the feature structures to be identified between the X comparison images and the corresponding X standard images respectively, wherein M feature structure similarities to be identified exist between each comparison image and the corresponding standard image;
the comparison module is used for judging whether the paper currency to be identified is non-distorted paper currency if the similarity of any feature structure to be identified is larger than the similarity of the corresponding standard feature structure, and if not, judging that the paper currency to be identified is local distorted paper currency;
the model training module is also used for processing the front side and the back side of each first sample paper money by using a plurality of different lights to obtain a red light positive image, a red light negative image, a green light positive image, a green light negative image, a Lan Guangzheng image and a blue light negative image; integrating the distortion areas of all the red light positive graphs of the first sample paper money to obtain a red light target area set locator RZ, wherein the distortion areas are rectangular and are represented by two coordinates of the lower left corner and the upper right corner, and the number of the distortion areas of the red light positive graphs is M; the same steps as those of S12 are respectively carried out on the red light reverse graph, the green light forward graph, the green light reverse graph, the Lan Guangzheng graph and the blue light reverse graph of the first sample paper money, and locationRF, locationGZ, locationGF, locationBZ, locationBF are respectively obtained; storing the X target area sets with the largest number of distortion areas stored in locationRZ, locationRF, locationGZ, locationGF, locationBZ, locationBF, and confirming the image type corresponding to each target area set, wherein the image type is defined as a target image type; calculating the signal-to-noise ratio between any two second sample banknote images based on the X target image types to obtain X groups of first quasi-standard images; based on X target image types, calculating the overall image similarity between any two second sample banknote images to obtain X groups of second quasi-standard images; calculating the overall image similarity between any two of 2 first quasi-standard images and 2 second quasi-standard images corresponding to each target image type, and defining 2 second sample banknote images corresponding to the minimum overall image similarity as result P, wherein the standard image is any one of the result P; and calculating the standard feature structure similarity between the two second sample banknote images in the resultPs based on the M distortion areas, wherein each group of resultPs corresponds to the M standard feature structure similarity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311595338.1A CN117314897B (en) | 2023-11-28 | 2023-11-28 | Method and device for discriminating distortion of banknote image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311595338.1A CN117314897B (en) | 2023-11-28 | 2023-11-28 | Method and device for discriminating distortion of banknote image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117314897A CN117314897A (en) | 2023-12-29 |
| CN117314897B true CN117314897B (en) | 2024-03-19 |
Family
ID=89286870
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311595338.1A Active CN117314897B (en) | 2023-11-28 | 2023-11-28 | Method and device for discriminating distortion of banknote image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117314897B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107067000A (en) * | 2017-03-31 | 2017-08-18 | 深圳怡化电脑股份有限公司 | A kind of recognition methods of bank note degree and device |
| KR101781009B1 (en) * | 2016-08-31 | 2017-10-23 | 노틸러스효성 주식회사 | Soiled banknote discrimination method |
| CN107610316A (en) * | 2017-08-28 | 2018-01-19 | 深圳怡化电脑股份有限公司 | Detect the method, apparatus and terminal device of bank note defect |
| CN108604397A (en) * | 2016-03-29 | 2018-09-28 | 光荣株式会社 | Paper representing value processing unit and paper representing value processing method |
| CN110910561A (en) * | 2018-09-18 | 2020-03-24 | 深圳怡化电脑股份有限公司 | Banknote contamination identification method and device, storage medium and financial equipment |
| EP3723054A1 (en) * | 2019-03-27 | 2020-10-14 | Glory Ltd. | Banknote recognition unit, banknote handling device, and banknote recognition method |
| CN114255540A (en) * | 2022-01-25 | 2022-03-29 | 中国农业银行股份有限公司 | Method, device, equipment and storage medium for identifying stained paper money |
| CN115346306A (en) * | 2022-08-26 | 2022-11-15 | 中国银行股份有限公司 | Method and device for identifying stained paper money |
| CN117115486A (en) * | 2023-10-19 | 2023-11-24 | 恒银金融科技股份有限公司 | Banknote version identification method and device and electronic equipment |
-
2023
- 2023-11-28 CN CN202311595338.1A patent/CN117314897B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108604397A (en) * | 2016-03-29 | 2018-09-28 | 光荣株式会社 | Paper representing value processing unit and paper representing value processing method |
| KR101781009B1 (en) * | 2016-08-31 | 2017-10-23 | 노틸러스효성 주식회사 | Soiled banknote discrimination method |
| CN107067000A (en) * | 2017-03-31 | 2017-08-18 | 深圳怡化电脑股份有限公司 | A kind of recognition methods of bank note degree and device |
| CN107610316A (en) * | 2017-08-28 | 2018-01-19 | 深圳怡化电脑股份有限公司 | Detect the method, apparatus and terminal device of bank note defect |
| CN110910561A (en) * | 2018-09-18 | 2020-03-24 | 深圳怡化电脑股份有限公司 | Banknote contamination identification method and device, storage medium and financial equipment |
| EP3723054A1 (en) * | 2019-03-27 | 2020-10-14 | Glory Ltd. | Banknote recognition unit, banknote handling device, and banknote recognition method |
| CN114255540A (en) * | 2022-01-25 | 2022-03-29 | 中国农业银行股份有限公司 | Method, device, equipment and storage medium for identifying stained paper money |
| CN115346306A (en) * | 2022-08-26 | 2022-11-15 | 中国银行股份有限公司 | Method and device for identifying stained paper money |
| CN117115486A (en) * | 2023-10-19 | 2023-11-24 | 恒银金融科技股份有限公司 | Banknote version identification method and device and electronic equipment |
Non-Patent Citations (1)
| Title |
|---|
| 一种快速纸币图像污损检测算法研究与仿真;张颖 等;计算机仿真;第33卷(第06期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117314897A (en) | 2023-12-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Fast and effective image copy-move forgery detection via hierarchical feature point matching | |
| CN109086714B (en) | Form recognition method, recognition system and computer device | |
| AU2011288069B2 (en) | Valuable file identification method and identification system, device thereof | |
| CN112686812B (en) | Bank card inclination correction detection method and device, readable storage medium and terminal | |
| CN102667810B (en) | Face recognition in digital images | |
| Gill et al. | A review paper on digital image forgery detection techniques | |
| CN108681735A (en) | Optical character recognition method based on convolutional neural networks deep learning model | |
| CN111310628A (en) | Paper currency forming mode inspection and identification method based on paper currency printing pattern characteristics | |
| CN109409158B (en) | Anti-counterfeiting method based on two-dimensional code edge roughness | |
| Pawade et al. | Comparative study of different paper currency and coin currency recognition method | |
| CN111767877A (en) | Living body detection method based on infrared features | |
| CN111738079A (en) | Banknote denomination recognition method and device | |
| CN115311746A (en) | Off-line signature authenticity detection method based on multi-feature fusion | |
| WO2022121025A1 (en) | Certificate category increase and decrease detection method and apparatus, readable storage medium, and terminal | |
| CN117314897B (en) | Method and device for discriminating distortion of banknote image | |
| CN111582115B (en) | Financial bill processing method, device, equipment and readable storage medium | |
| CN112200075A (en) | Face anti-counterfeiting method based on anomaly detection | |
| CN112215225A (en) | KYC certificate verification method based on computer vision technology | |
| WO2021102770A1 (en) | Method and device for verifying authenticity of product | |
| Tahaoglu et al. | Robust Copy-Move Forgery Detection Technique Against Image Degradation and Geometric Distortion Attacks | |
| CN110969757A (en) | Multi-country banknote type rapid identification technology | |
| CN113269136A (en) | Offline signature verification method based on triplet loss | |
| CN108074323B (en) | Paper money facing identification method and device thereof | |
| CN112597810A (en) | Identity document authentication method and system | |
| Shinde et al. | Identification of fake currency using soft computing |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |