Banknote facing recognition method and device, electronic equipment and storage medium

Abstract

This application discloses a banknote orientation recognition method, apparatus, electronic device, and storage medium, belonging to the field of image processing technology. The banknote orientation recognition method includes: acquiring a first image and a second image of the banknote to be recognized; determining four first feature point descriptors based on the first image and the second image; determining four first similarities based on the four first feature point descriptors and sample feature point descriptors, wherein the sample feature point descriptors describe feature points in sample feature regions of the sample banknote image; and determining the orientation recognition result of the banknote to be recognized based on the four first similarities and sample orientation information of the sample banknote image. This method improves the robustness of the banknote orientation recognition method and the accuracy of the orientation recognition result.

Description

Technical Field

[0001] This application belongs to the field of image processing technology, and in particular relates to a method, device, electronic device and storage medium for banknote orientation recognition. Background Technology

[0002] Optical character recognition technology is a fundamental technology for financial self-service equipment, serving as a prerequisite for optical character recognition and counterfeit detection of banknotes.

[0003] Currently, orientation recognition is usually based on the average grayscale value of an image or the differences in patterns within the image. However, these methods have poor robustness and the accuracy of orientation recognition results is low when faced with signal fluctuations or when the image being recognized is dirty. Summary of the Invention

[0004] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes a banknote orientation recognition method, apparatus, electronic device, and storage medium, which improves the robustness of the orientation recognition method and the accuracy of the orientation recognition results.

[0005] In a first aspect, this application provides a method for identifying the orientation of banknotes, the method comprising:

[0006] Acquire a first image and a second image of the banknote to be identified, wherein the first image includes pixel information of a first side of the banknote to be identified, and the second image includes pixel information of a second side of the banknote to be identified;

[0007] Based on the first image and the second image, four first feature point descriptors are determined. The four first feature point descriptors are used to describe the feature points of four first feature regions. Two of the four first feature regions are located in the first image, and the other two are located in the second image.

[0008] Based on the four first feature point descriptors and the sample feature point descriptors, four first similarities are determined. The sample feature point descriptors are used to describe the feature points of the sample feature region in the sample banknote image.

[0009] Based on the four first similarities and the sample orientation information of the sample banknote image, the orientation recognition result of the banknote to be identified is determined;

[0010] Wherein, the two first feature regions in the first image are centrally symmetrical about the center point of the first image, the two first feature regions in the second image are centrally symmetrical about the center point of the second image, one of the two first feature regions in the first image corresponds to the sample feature region and has the same image position, and one of the two first feature regions in the second image corresponds to the sample feature region and has the same image position.

[0011] According to the banknote orientation recognition method of this application, by determining the four first feature point descriptors in the image of the banknote to be recognized and the four first similarities with the sample feature point descriptors, and combining the preset orientation in the image of the banknote to be recognized and the sample orientation information of the sample banknote image with the four first similarities, the orientation recognition result of the banknote to be recognized is determined, thereby improving the robustness of the orientation recognition method and the accuracy of the orientation recognition result.

[0012] According to one embodiment of this application, the preset orientations of the four first feature regions are respectively front facing, front facing backward, back facing forward, and back facing backward. Determining the orientation recognition result of the banknote to be identified based on the four first similarities and the sample orientation information of the sample banknote image includes:

[0013] Based on the four first similarities, the target feature region of the banknote to be identified is determined. The target feature region is one of the four first feature regions, and the first similarity corresponding to the target feature region is greater than the first similarity corresponding to the other three first feature regions among the four first feature regions.

[0014] The orientation recognition result is determined based on the preset orientation of the target feature region and the orientation information of the sample.

[0015] According to one embodiment of this application, determining four first similarities based on the four first feature point descriptors and the sample feature point descriptors includes:

[0016] Based on the four first feature point descriptors and the sample feature point descriptors, four feature point pairs are determined. The four feature point pairs are used to characterize the matching relationship between the feature points of the four first feature regions and the feature points of the sample feature regions.

[0017] Identify the sample points and outliers in each pair of feature points;

[0018] The four first similarities are determined based on the sample points and the outliers in each of the feature point pairs.

[0019] According to one embodiment of this application, determining the four first similarities based on the sample points and outliers in each of the feature point pairs includes:

[0020] Application formula

[0021]

[0022] Determine the first similarity;

[0023] Among them, S ip represents the first similarity corresponding to the i-th first feature region. i q represents the number of sample points in the feature point pair corresponding to the i-th first feature region. i The number of outliers in the feature point pair corresponding to the i-th first feature region, i = 1, 2, 3 or 4.

[0024] According to one embodiment of this application, determining four first similarities based on the four first feature point descriptors and the sample feature point descriptors includes:

[0025] Based on the four first feature point descriptors and the N sample feature point descriptors, 4N first similarities are determined. The N sample feature point descriptors are used to describe the feature points of the sample feature regions of the N sample banknote images. The sample version information corresponding to the N sample banknote images is different, and N is a positive integer greater than 1.

[0026] The determination of the orientation recognition result of the banknote to be identified based on the four first similarities and the sample orientation information of the sample banknote image includes:

[0027] Based on the 4N first similarities, the sample orientation information, and the sample version information, the orientation recognition result and version recognition result of the banknote to be identified are determined.

[0028] According to one embodiment of this application, determining the orientation recognition result and version recognition result of the banknote to be identified based on the 4N first similarities, the sample orientation information, and the sample version information includes:

[0029] Based on the 4N first similarities, N target similarities are determined. The N target similarities correspond one-to-one with the N sample banknote images. The target similarity is used to characterize the similarity between the target feature region of the banknote to be identified and the sample feature region in the corresponding sample banknote image. The first similarity corresponding to the target feature region is greater than the first similarity corresponding to the other three first feature regions among the four first feature regions.

[0030] Based on the N target similarities, a target sample banknote version image is determined, wherein the target similarity corresponding to the target sample banknote version image is greater than the target similarity corresponding to the other N-1 sample banknote version images among the N sample banknote images;

[0031] Based on the sample orientation information and sample version information corresponding to the target sample banknote version image, the orientation recognition result and version recognition result of the banknote to be identified are determined.

[0032] Secondly, this application provides a banknote orientation recognition device, the device comprising:

[0033] The acquisition module is used to acquire a first image and a second image of the banknote to be identified. The first image includes pixel information of a first side of the banknote to be identified, and the second image includes pixel information of a second side of the banknote to be identified.

[0034] A first processing module is configured to determine four first feature point descriptors based on the first image and the second image. The four first feature point descriptors are used to describe feature points of four first feature regions. Two of the four first feature regions are located in the first image, and the other two are located in the second image.

[0035] The second processing module is used to determine four first similarities based on the four first feature point descriptors and the sample feature point descriptors, wherein the sample feature point descriptors are used to describe the feature points of the sample feature region in the sample banknote image.

[0036] The third processing module is used to determine the orientation recognition result of the banknote to be identified based on the four first similarities and the sample orientation information of the sample banknote image;

[0037] Wherein, the two first feature regions in the first image are centrally symmetrical about the center point of the first image, the two first feature regions in the second image are centrally symmetrical about the center point of the second image, one of the two first feature regions in the first image corresponds to the sample feature region and has the same image position, and one of the two first feature regions in the second image corresponds to the sample feature region and has the same image position.

[0038] According to the banknote orientation recognition device of this application, by determining the four first feature point descriptors in the image of the banknote to be recognized and the four first similarities with the sample feature point descriptors, and combining the preset orientation in the image of the banknote to be recognized and the sample orientation information of the sample banknote image with the four first similarities, the orientation recognition result of the banknote to be recognized is determined, thereby improving the robustness of the orientation recognition method and the accuracy of the orientation recognition result.

[0039] Thirdly, this application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the banknote orientation recognition method as described in the first aspect above.

[0040] Fourthly, this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the banknote orientation recognition method as described in the first aspect above.

[0041] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the banknote orientation recognition method as described in the first aspect above.

[0042] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0043] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0044] Figure 1 This is one of the flowcharts illustrating the banknote orientation recognition method provided in the embodiments of this application;

[0045] Figure 2 This is a schematic diagram of the structure of the financial self-service device provided in the embodiments of this application;

[0046] Figure 3 This is a second schematic flowchart of the banknote orientation recognition method provided in the embodiments of this application;

[0047] Figure 4 This is a schematic diagram showing the distribution of the first feature region in the first image according to an embodiment of this application;

[0048] Figure 5 This is a schematic diagram of the structure of the banknote orientation recognition device provided in the embodiments of this application;

[0049] Figure 6 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application.

[0050] Reference numerals: signal acquisition module 210, image preprocessing module 220, orientation recognition module 230, first image 400, first feature region 401. Detailed Implementation

[0051] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0052] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0053] The following description, in conjunction with the accompanying drawings, details the banknote orientation recognition method, banknote orientation recognition device, electronic device, and readable storage medium provided in this application through specific embodiments and application scenarios.

[0054] The banknote orientation recognition method can be applied to a terminal, specifically executed by the hardware or software within the terminal.

[0055] The terminal includes, but is not limited to, portable communication devices such as mobile phones or tablets with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads). It should also be understood that, in some embodiments, the terminal may not be a portable communication device, but rather a desktop computer with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads).

[0056] The following embodiments describe a terminal including a display and a touch-sensitive surface. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, mouse, and joystick.

[0057] like Figure 1 As shown, the banknote orientation recognition method includes steps 110 to 140.

[0058] Step 110: Obtain the first image 400 and the second image of the banknote to be identified.

[0059] The first image 400 includes pixel information of the first side of the banknote to be identified, and the second image includes pixel information of the second side of the banknote to be identified.

[0060] The first and second sides of the banknote to be identified are set opposite each other. For example, the first side is the side with a portrait printed on it, and the second side is the side with numbers printed on it.

[0061] In this step, the banknote to be identified can be a banknote placed in the ticket slot of a financial self-service machine. The first image 400 and the second image of the banknote to be identified can be obtained by setting an image sensor in the ticket slot of the financial self-service machine.

[0062] The first image 400 and the second image can be acquired using a complementary metal-oxide-semiconductor (CMOS) image sensor.

[0063] In practice, two complementary metal-oxide-semiconductor (CMOS) image sensors can be installed at the ticket inlet of the self-service financial device. One CMOS image sensor is installed at the upper end of the ticket inlet, and the other is installed at the lower end. When the banknote to be identified is inserted into the ticket inlet, the two CMOS image sensors can contact both sides of the banknote to be identified, and the image sensors can capture the first image 400 and the second image of the banknote to be identified.

[0064] Step 120: Based on the first image 400 and the second image, determine four first feature point descriptors.

[0065] like Figure 4 As shown, four first feature point descriptors are used to describe the feature points of four first feature regions 401. The two first feature regions 401 in the first image 400 are centrally symmetrical about the center point of the first image 400. It can be understood that the two first feature regions 401 in the second image are also centrally symmetrical about the center point of the second image. One of the two first feature regions in the first image corresponds to the sample feature region and has the same image position. One of the two first feature regions in the second image corresponds to the sample feature region and has the same image position.

[0066] In this embodiment, two of the four first feature regions 401 are located in the first image 400, and the other two are located in the second image.

[0067] It is understandable that by acquiring two first feature regions 401 in the first image 400 and two first feature regions 401 in the second image, the four first feature regions 401 are located in pairs on different sides of the banknote to be identified, which can cover all possible orientation recognition results of the banknote to be identified, including front facing, front facing backward, back facing forward, and back facing backward.

[0068] In this embodiment, the first feature region 401 is the region in the first image 400 and the second image that corresponds to the sample feature region and is located at the same position.

[0069] For example, if the corresponding sample feature region is the first target image region of 100 RMB, then the regions corresponding to the four first feature regions 401 are the first target image, the 100 digit region image, the second target image, and the blank region image, respectively.

[0070] In this embodiment, the first image 400 selects a region with a large pixel difference and unique texture as the first feature region 401, which can improve the accuracy of banknote recognition when performing orientation recognition.

[0071] In practice, since the portrait and floral areas on the front of the 100 RMB note are quite large, feature extraction takes a long time. Therefore, we can select only the feature small image of the area with the larger pixel difference as the first feature area 401, which can improve the efficiency of face recognition while ensuring the accuracy of banknote face recognition.

[0072] A descriptor is a data structure that describes a feature. A feature point descriptor can be multi-dimensional and used to describe different attributes of the feature.

[0073] In this step, the four first feature point descriptors describing the feature points of the four first feature regions 401 can be determined by using the Scale-invariant feature transform (SIFT) algorithm or the Speeded Up Robust Features (SURF) algorithm.

[0074] Step 130: Determine the four first similarities based on the four first feature point descriptors and the sample feature point descriptors.

[0075] Among them, the sample feature point descriptor is used to describe the feature points of the sample feature region in the sample banknote image.

[0076] In this step, the sample banknote image is a banknote image with a known orientation, and the sample feature region can also be obtained by selecting feature maps with large pixel differences and unique textures in the sample banknote image.

[0077] In this embodiment, the sample banknote image can be a pre-processed banknote image, and the sample feature regions and sample feature point descriptors in the sample banknote image are all known values.

[0078] In practice, a similarity calculation can be performed between a first feature point descriptor and a sample feature point descriptor to determine four first similarities of the four first feature regions 401. Each first similarity corresponds to a first feature region 401, and the first similarity characterizes the degree of similarity between the first feature region 401 and the sample feature region.

[0079] It should be noted that the sample banknote image corresponds to the same version and denomination of the banknote to be identified.

[0080] For example, the banknote to be identified is the 2005 edition of the 100 RMB banknote, and the sample banknote is also the 2005 edition of the 100 RMB banknote.

[0081] Step 140: Based on the four first similarity scores and the sample orientation information of the sample banknote images, determine the orientation recognition result of the banknote to be identified.

[0082] The first similarity represents the degree of similarity between the first feature region 401 and the sample feature region. The four first similarities represent the degree of similarity between the four first feature regions 401 on both sides of the banknote to be identified and the sample feature regions of the sample banknote image.

[0083] Among them, the preset orientations of the four first feature regions 401 are front facing, front facing backward, back facing forward, and back facing backward, respectively.

[0084] In this embodiment, the orientation information of the sample feature region is determined, and the orientation of the sample feature region can be front-facing, front-facing-backward, back-facing-frontward, or back-facing-backward.

[0085] By using the largest first similarity among the four first similarities and the sample orientation information, it can be determined which side of the banknote to be identified has the greatest similarity to the sample feature region 401, and the orientation recognition result of the banknote to be identified can be determined based on the known sample orientation information.

[0086] For example, the first side of the banknote to be identified includes a first feature area 401 with the preset orientation being front-facing and a first feature area 401 with the preset orientation being front-back, and the second side includes a first feature area 401 with the preset orientation being back-facing and a first feature area 401 with the preset orientation being back-back.

[0087] Based on the sample feature point descriptor and the first feature point descriptor corresponding to the first feature region 401 with the preset facing front, the first similarity corresponding to the first feature region 401 with the preset facing front is calculated to be 0.26.

[0088] The sample feature point descriptor and the first feature point descriptor corresponding to the first feature region 401 with the preset orientation of front and back are calculated to have a first similarity of 0.28.

[0089] The sample feature point descriptor and the first feature point descriptor corresponding to the first feature region 401 with the preset orientation of the opposite side are calculated to have a first similarity of 0.97.

[0090] The sample feature point descriptor and the first feature point descriptor corresponding to the first feature region 401 with the preset orientation reversed are calculated to have a first similarity of 0.29.

[0091] Therefore, it can be seen that the first similarity of the first feature region 401 with the preset orientation of the opposite side is greater than the first similarity of the other three first feature regions 401 among the four first feature regions 401.

[0092] If the preset orientation of the first feature area 401 is reverse-facing, then the second side of the banknote to be identified is the reverse side, and the first side is the front side.

[0093] In related technologies, the average grayscale value of an image or the difference in patterns within the image is typically used as a basis for banknote orientation recognition. However, these orientation recognition methods have poor robustness and low accuracy when faced with signal fluctuations or dirty images.

[0094] In this embodiment, by determining the four first feature point descriptors in the image of the banknote to be identified and the four first similarities of the sample feature point descriptors, and combining the preset orientation in the image of the banknote to be identified and the sample orientation information of the sample banknote image with the four first similarities, the orientation recognition result of the banknote to be identified is determined. Compared with using the average gray value and pattern difference as the recognition basis, using feature point descriptors as the recognition basis for orientation recognition has better robustness and the orientation recognition result is more accurate.

[0095] According to the banknote orientation recognition method provided in the embodiments of this application, by determining the four first feature point descriptors in the image of the banknote to be recognized and the four first similarities with the sample feature point descriptors, and combining the preset orientation in the image of the banknote to be recognized and the sample orientation information of the sample banknote image with the four first similarities, the orientation recognition result of the banknote to be recognized is determined, thereby improving the robustness of the orientation recognition method and the accuracy of the orientation recognition result.

[0096] In some embodiments, the preset orientations of the four first feature regions 401 are respectively front facing, front facing backward, back facing forward, and back facing backward. Step 140, based on the four first similarities and the sample orientation information of the sample banknote image, determines the orientation recognition result of the banknote to be identified, including:

[0097] Based on the four first similarities, the target feature region of the banknote to be identified is determined. The target feature region is one of the four first feature regions 401. The first similarity corresponding to the target feature region is greater than the first similarity corresponding to the other three feature regions 401.

[0098] Based on the preset orientation of the target feature region and the orientation information of the sample, the orientation recognition result is determined.

[0099] Among them, the target feature region is the first feature region 401 that is most similar to the sample feature region among the four first feature regions 401. The preset orientation is the orientation information of the first feature region 401. This orientation information is not necessarily the real orientation, but only represents four possibilities of the orientation recognition result. The four possibilities of the orientation recognition result correspond to four different preset orientations.

[0100] In this embodiment, the preset orientations of the four first feature regions 401 are respectively front facing, front facing backward, back facing forward, and back facing backward.

[0101] Taking the preset orientation of the target feature region as the reverse and the sample orientation information as the forward as an example.

[0102] If the first similarity of the target feature region is greater than the first similarity of the other three feature regions among the four first feature regions 401, then the preset orientation of the target feature region should be consistent with the orientation information of the sample. Therefore, the actual orientation of the target feature region should be front-facing. The first feature region 401 with the preset orientation of back-facing should have the actual orientation of front-back. The first feature region 401 with the preset orientation of front-facing should have the actual orientation of back-back. The first feature region 401 with the preset orientation of front-back should have the actual orientation of back-facing.

[0103] In some embodiments, four first similarities are determined based on four first feature point descriptors and sample feature point descriptors, including:

[0104] Based on the four first feature point descriptors and the sample feature point descriptors, four feature point pairs are determined. These four feature point pairs are used to characterize the matching relationship between the feature points of the four first feature regions 401 and the feature points of the sample feature regions.

[0105] Identify the sample points and outliers in each feature point pair;

[0106] Based on the sample points and outliers in each feature point pair, four first similarities are determined.

[0107] In this embodiment, the Fast Approximate Nearest Neighbor Search (FLANN) matching algorithm can be used to match four first feature point descriptors and sample feature point descriptors. Each first feature point descriptor is matched with a sample feature point descriptor to obtain a feature point pair.

[0108] The first feature point descriptor describes the feature points of the first feature region 401, and the sample feature point descriptor describes the feature points of the sample feature region.

[0109] Among them, a feature point pair includes the pairwise matching relationship between multiple feature points in a first feature region 401 and multiple feature points in a sample feature region.

[0110] For example, a feature point pair can be obtained by matching feature points on the face in the first feature region 401 with feature points on the face in the sample feature region.

[0111] For example, a feature point pair can also be obtained by matching the feature points on the numbers in the first feature region 401 with the feature points on the numbers in the sample feature region.

[0112] By matching the first feature point descriptors in the four first feature regions 401 with the sample feature point descriptors in the sample feature regions, four feature point pairs can be determined.

[0113] In practice, after identifying feature point pairs, the two corresponding matching feature points in the feature point pair are compared. A matching threshold can be set. When the degree of matching of the position or pixel information between the feature points in the feature point pair is greater than the matching threshold, the two feature points are called sample points. When the degree of matching of the position or pixel information in the matching relationship of the feature point pair is less than the matching threshold, the two feature points are called outliers.

[0114] For example, if the matching threshold is set to 0.7, and the degree of matching of positional information between two feature points in a feature point pair is 0.88, then these two feature points are recorded as sample points.

[0115] For example, if the matching threshold is set to 0.9, and the degree of matching of the positional information between two feature points in a feature point pair is 0.86, then these two feature points are recorded as outliers.

[0116] In this embodiment, the outlier and sample points in the feature point pair can be determined by the Random Sampling Consensus (RANSAC) algorithm.

[0117] In practice, a mathematical model for interpreting or observing the first feature point descriptor and the sample feature point descriptor can be selected and input into the random sampling consensus algorithm. The parameters of the mathematical model are given, and the feature point pairs are input into the algorithm to obtain the outlier points and sample points output by the random sampling consensus algorithm.

[0118] Among them, feature points that can adapt to the mathematical model are called sample points, and feature points that cannot adapt to the mathematical model are called outliers.

[0119] By matching four first feature point descriptors with sample feature point descriptors, four feature point pairs are determined, and sample points and outliers within each feature point pair are identified. Then, the first similarity corresponding to the first feature region 401 of the four first feature point descriptors is calculated. Based on the first similarity and sample orientation information, the orientation of the banknote to be identified is determined, which can effectively improve the accuracy of orientation recognition results.

[0120] In some embodiments, four first similarities are determined based on the sample points and outliers in each feature point pair, including:

[0121] Application formula

[0122]

[0123] Determine the first similarity;

[0124] Among them, S i p represents the first similarity corresponding to the i-th first feature region 401. i q represents the number of sample points in the feature point pair corresponding to the i-th first feature region 401. i The number of outliers in the feature point pair corresponding to the i-th first feature region 401, i = 1, 2, 3 or 4.

[0125] In this embodiment, the first similarity between the first feature region 401 and the sample feature region can be determined by determining the number of sample points and the number of outliers in the feature point pair corresponding to the first feature region 401.

[0126] For example, assuming that the number of sample points in the feature point pair corresponding to the first feature region 401 with the facing direction is 90 and the number of outliers is 10, then the similarity between the first feature region 401 and the sample feature region is 0.9.

[0127] In some embodiments, four first similarities are determined based on four first feature point descriptors and sample feature point descriptors, including:

[0128] Based on four first feature point descriptors and N sample feature point descriptors, 4N first similarities are determined. The N sample feature point descriptors are used to describe the feature points of the sample regions of the N sample banknote images. The sample version information corresponding to the N sample banknote images is different, and N is a positive integer greater than 1.

[0129] Among them, the sample version information refers to the version information of the banknote. For example, there are two versions of RMB: the 1999 version and the 2005 version.

[0130] In practice, banknote orientation recognition needs to take into account the different characteristics of different versions of banknotes. In banknote orientation recognition, if the version information of the banknote is uncertain, the sample banknote image needs to include not only orientation information but also sample version information.

[0131] Taking RMB as an example, in this embodiment, based on four first feature point descriptors and two sample feature point descriptors, eight first similarities can be obtained.

[0132] In this embodiment, based on four first similarity scores and sample orientation information of the sample banknote images, the orientation recognition result of the banknote to be identified is determined, including:

[0133] Based on 4N first similarity scores, sample orientation information, and sample version information, the orientation recognition result and version recognition result of the banknote to be identified are determined.

[0134] Among them, the largest first similarity among the 4N first similarities can be determined. Based on the sample orientation information corresponding to the largest first similarity, the orientation recognition result of the banknote to be identified can be determined. Based on the version recognition result corresponding to the largest first similarity, the version recognition result of the banknote to be identified can be determined.

[0135] In some embodiments, based on 4N first similarities, sample orientation information, and sample version information, the orientation recognition result and version recognition result of the banknote to be identified are determined, including:

[0136] Based on 4N first similarities, N target similarities are determined. The N target similarities correspond one-to-one with the N sample banknote images. The target similarity is used to characterize the similarity between the target feature region of the banknote to be identified and the sample feature region in the corresponding sample banknote image. The first similarity corresponding to the target feature region is greater than the first similarity corresponding to the other three first feature regions 401 among the four first feature regions 401.

[0137] Based on N target similarities, a target sample banknote version image is determined. The target similarity corresponding to the target sample banknote version image is greater than the target similarity corresponding to the other N-1 sample banknote images among the N sample banknote images.

[0138] Based on the sample orientation information and sample version information corresponding to the target sample banknote version image, the orientation recognition result and version recognition result of the banknote to be identified are determined.

[0139] In this context, four first similarities can be determined between each version of the sample banknote image and the four first feature regions 401. The target similarity is the largest first similarity among the first similarities corresponding to each version of the sample banknote image. The target sample banknote version image is the sample banknote image corresponding to the largest target similarity among N target similarities.

[0140] For example, the similarity between the 2005 version of the RMB and the four first feature regions 401 are 0.4, 0.35, 0.37 and 0.98 respectively, and the target similarity of the 2005 version of the RMB is 0.98.

[0141] The similarity scores of the 1999 edition of the RMB with the four first feature regions 401 are 0.32, 0.31, 0.27 and 0.96 respectively, and the target similarity score of the 1999 edition of the RMB is 0.96.

[0142] Among them, the target similarity of the 1999 version of RMB is 0.96, and the target similarity of the 2005 version of RMB is 0.98. Therefore, the target sample banknote version image is the sample banknote image corresponding to the 2005 version of RMB.

[0143] Based on the target sample banknote version image and sample orientation information, the orientation recognition result of the banknote to be identified can be determined, and based on the sample version information, the version recognition result can be determined.

[0144] For example, if the target sample banknote version image is a 2005 edition of the Renminbi with the front facing forward, and the target feature region corresponding to the sample feature region of the target sample banknote version image is assumed to be facing forward, then the recognition result of the orientation of the banknote to be identified is facing forward, and the version information is the 2005 edition of the Renminbi.

[0145] By determining N sample feature point descriptors of the banknote to be identified, and determining 4N first similarities with four first feature point descriptors, the maximum target similarity among the 4N first similarities is further determined. Based on the sample orientation information and sample version information, combined with the preset orientation corresponding to the maximum target similarity, the orientation recognition result and version recognition result of the banknote to be identified are determined. This method can solve the orientation recognition and version recognition problems of banknotes of different versions.

[0146] The following is a specific embodiment used to describe the specific application scenario of the banknote orientation recognition method of this application.

[0147] like Figure 2 As shown, the financial self-service equipment may include a signal acquisition module 210, an image preprocessing module 220, and an orientation recognition module 230.

[0148] The signal acquisition module 210 can be an image sensor. The signal acquisition module 210 can be installed at the ticket inlet of the financial self-service equipment. The signal acquisition module 210 is used to acquire the image signal of the banknote to be recognized at the ticket inlet.

[0149] The signal acquisition module 210 can generate a first image 400 and a second image of the banknote to be identified based on the acquired image signal, and transmit the first image 400 and the second image to the image preprocessing module 220. The image preprocessing module 220 preprocesses the first image 400 and the second image and transmits the first image 400 and the second image to the orientation recognition module 230 so that the orientation recognition module 230 can perform feature extraction operations on the image.

[0150] like Figure 3 As shown, the orientation recognition module 230 selects four first feature regions 401 in the first image 400 and the second image, and assigns a preset orientation to each first feature region 401. It extracts feature points for each first feature region 401, calculates the first feature point descriptor, and obtains four first feature point descriptors. It matches the four first feature point descriptors with the sample feature point descriptors to obtain four feature point pairs. It uses a random sampling consensus algorithm to obtain the sample points and outliers in each feature point pair, and uses the proportion of the sample points in the sample points and outliers as the similarity between the first feature region 401 and the sample feature region to obtain four first similarities. The orientation of the first feature region 401 corresponding to the largest first similarity is set as the actual orientation.

[0151] This application also provides another method for banknote orientation recognition.

[0152] Methods for identifying the orientation of banknotes include:

[0153] Acquire the target image of the banknote to be identified;

[0154] Based on the target image, a first feature point descriptor is determined. The first feature point descriptor is used to describe the feature points of the first feature region 401 of the target image.

[0155] Based on the first feature point descriptor and four sample feature point descriptors, four first similarities are determined. The four sample feature point descriptors are used to describe the feature points of four sample feature regions in the sample banknote image. The sample banknote image includes a first sample image and a second sample image. The first sample image and the second sample image are two opposite sides of the sample banknote. Two of the four sample feature regions are located in the first sample image, and the other two are located in the second sample image.

[0156] Based on the first similarity scores and the sample orientation information of the sample banknote images, the orientation recognition result of the banknote to be identified is determined.

[0157] In the first sample image, two sample feature regions are centrally symmetrical about the center point of the first sample image, and in the second sample image, two sample feature regions are centrally symmetrical about the center point of the second sample image. The image positions of the two sample feature regions in the first sample image are the same as those in the second sample image.

[0158] In this embodiment, the first feature region 401 can be determined by using a scale-invariant feature transform (SIFT) algorithm or a speed-up robust features (SURF) algorithm to determine the first feature point descriptor of the feature points describing the first feature region 401 of the target image.

[0159] In this embodiment, the first feature region 401 is selected as the region with a large pixel difference and unique texture. This can improve the accuracy of banknote recognition when performing orientation recognition.

[0160] In practice, since the portrait and floral areas on the front of the 100 RMB note are quite large, feature extraction takes a long time. Therefore, we can select only the feature small images of the areas with large pixel differences and unique textures as the first feature area 401, which can improve the efficiency of face recognition while ensuring the accuracy of banknote face recognition.

[0161] For example, the four sample feature regions can be the first target image of 100 RMB, the 100 digit region image, the second target image, and the blank region image, respectively. Thus, the image positions of the sample feature regions can be obtained. Based on the image positions of the sample feature regions, the first feature region 401 corresponding to the image position of the sample feature region is determined in the first image 400 or the second image corresponding to the banknote to be identified.

[0162] In this embodiment, the sample banknote image is a banknote image with a known orientation, and the four sample feature regions can also be obtained by selecting feature thumbnails with large pixel differences and unique textures from the banknote.

[0163] In this embodiment, the sample banknote image can be a pre-processed banknote image, and the sample feature regions and sample feature point descriptors in the sample banknote image are all known values.

[0164] It is understandable that by acquiring two sample feature regions in the first sample image and two sample feature regions in the second sample image, the four sample feature regions are located in pairs on different sides of the sample banknote, and the corresponding orientations include front facing, front facing backward, back facing forward, and back facing backward.

[0165] In practice, the similarity between the four sample feature point descriptors and the first feature point descriptor can be calculated respectively, thereby determining the four first similarities between the first feature region 401 and the four corresponding sample feature regions. The first similarity characterizes the degree of similarity between the first feature region 401 and the sample feature regions.

[0166] It should be noted that the sample banknote image corresponds to the same version and denomination of the banknote to be identified.

[0167] For example, the banknote to be identified is the 2005 edition of the 100 RMB banknote, and the sample banknote is also the 2005 edition of the 100 RMB banknote.

[0168] In this embodiment, the first similarity represents the degree of similarity between the first feature region 401 and the sample feature region, and the four first similarities represent the degree of similarity between the first feature region 401 of the banknote to be identified and the four sample feature regions of the sample banknote image.

[0169] Among them, the sample orientations of the four sample feature regions are front-facing, front-facing-back, back-facing-front, and back-facing-back, respectively.

[0170] By using the maximum first similarity among the four first similarities and the sample orientation information, it can be determined which sample feature region of the first feature region 401 in the banknote to be identified has the greatest similarity, and the orientation recognition result of the banknote to be identified can be determined based on the known sample orientation information.

[0171] For example, the first side of the sample banknote includes a sample feature area facing forward and a sample feature area facing backward, and the second side includes a sample feature area facing backward and a sample feature area facing backward.

[0172] Based on the first feature point descriptor and the sample feature point descriptor corresponding to the sample feature region facing forward, the first similarity between the sample feature region facing forward and the first feature region 401 is calculated to be 0.26.

[0173] The first feature point descriptor and the sample feature point descriptor corresponding to the sample feature region facing forward and backward are used to calculate the first similarity between the sample feature region facing forward and backward and the first feature region 401, which is 0.28.

[0174] The first feature point descriptor and the sample feature point descriptor corresponding to the sample feature region facing the negative side are used to calculate the first similarity between the sample feature region facing the negative side and the first feature region 401, which is 0.97.

[0175] The first feature point descriptor and the sample feature point descriptor corresponding to the sample feature region facing the opposite side are used to calculate the first similarity between the sample feature region facing the opposite side and the first feature region 401, which is 0.29.

[0176] Therefore, it can be seen that the first similarity between the sample feature region facing the negative side and the first feature region 401 is greater than the first similarity between the other three sample feature regions among the four sample feature regions.

[0177] The target image of the first feature region 401 is oriented as the reverse side. When it is determined that the target image represents the first side of the banknote to be identified, the first side of the banknote to be identified is the reverse side, and the second side of the banknote to be identified is the front side.

[0178] According to the banknote orientation recognition method provided in the embodiments of this application, by determining the first feature point descriptor in the image of the banknote to be recognized and the four first similarities of the four sample feature point descriptors, and combining the target image of the banknote to be recognized and the sample orientation information of the sample banknote images with the four first similarities, the orientation recognition result of the banknote to be recognized is determined, thereby improving the robustness of the orientation recognition method and the accuracy of the orientation recognition result.

[0179] It should be noted that for orientation recognition, you can select four regions to be recognized and one sample region of the banknote to be recognized, or you can select one region to be recognized and four sample regions of the banknote to be recognized. In both cases, the orientation information is determined by methods such as calculating similarity using descriptors and matching feature points.

[0180] When selecting one identification area and four sample areas of the banknote to be identified for oriented recognition, the method of similarity calculation and feature point matching can be the same as when selecting four identification areas and one sample area of ​​the banknote to be identified, and will not be described in detail in the embodiments of this application.

[0181] The banknote orientation recognition method provided in this application embodiment can be executed by an electronic device or a functional module or entity in an electronic device that can implement the banknote orientation recognition method. The electronic devices mentioned in this application embodiment include, but are not limited to, mobile phones, tablets, computers, cameras, and wearable devices. The banknote orientation recognition method provided in this application embodiment will be described below using an electronic device as the execution subject.

[0182] The banknote orientation recognition method provided in this application can be implemented by a banknote orientation recognition device. This application uses a banknote orientation recognition device to implement the banknote orientation recognition method as an example to illustrate the banknote orientation recognition device provided in this application.

[0183] This application also provides a banknote orientation recognition device.

[0184] like Figure 5 As shown, the banknote orientation recognition device includes:

[0185] The acquisition module 510 is used to acquire a first image 400 and a second image of the banknote to be identified. The first image 400 includes pixel information of a first side of the banknote to be identified, and the second image includes pixel information of a second side of the banknote to be identified.

[0186] The first processing module 520 is used to determine four first feature point descriptors based on the first image 400 and the second image. The four first feature point descriptors are used to describe the feature points of four first feature regions 401. Two of the four first feature regions 401 are located in the first image 400 and the other two are located in the second image.

[0187] The second processing module 530 is used to determine four first similarities based on four first feature point descriptors and sample feature point descriptors. The sample feature point descriptors are used to describe the feature points of the sample feature region in the sample banknote image.

[0188] The third processing module 540 is used to determine the orientation recognition result of the banknote to be identified based on the four first similarity scores and the sample orientation information of the sample banknote images.

[0189] In the first image 400, the two first feature regions 401 are centrally symmetrical about the center point of the first image 400, and the two first feature regions 401 in the second image are also centrally symmetrical about the center point of the second image. One of the two first feature regions in the first image corresponds to the sample feature region and has the same image position, and one of the two first feature regions in the second image corresponds to the sample feature region and has the same image position.

[0190] According to the banknote orientation recognition device provided in the embodiments of this application, by determining the four first feature point descriptors in the image of the banknote to be recognized and the four first similarities with the sample feature point descriptors, and combining the preset orientation in the image of the banknote to be recognized and the sample orientation information of the sample banknote image with the four first similarities, the orientation recognition result of the banknote to be recognized is determined, thereby improving the robustness of the orientation recognition method and the accuracy of the orientation recognition result.

[0191] In some embodiments, the third processing module 540 is used to determine the target feature region of the banknote to be identified based on four first similarities. The target feature region is one of the four first feature regions 401, and the first similarity corresponding to the target feature region is greater than the first similarity corresponding to the other three first feature regions 401.

[0192] Based on the preset orientation of the target feature region and the orientation information of the sample, the orientation recognition result is determined.

[0193] In some embodiments, the second processing module 530 is used to determine four feature point pairs based on four first feature point descriptors and sample feature point descriptors. The four feature point pairs are used to characterize the matching relationship between the feature points of the four first feature regions 401 and the feature points of the sample feature regions.

[0194] Identify the sample points and outliers in each feature point pair;

[0195] Based on the sample points and outliers in each feature point pair, four first similarities are determined.

[0196] In some embodiments, the second processing module 530 is further configured to apply formulas.

[0197]

[0198] Determine the first similarity;

[0199] Among them, S i p represents the first similarity corresponding to the i-th first feature region 401. i q represents the number of sample points in the feature point pair corresponding to the i-th first feature region 401. i The number of outliers in the feature point pair corresponding to the i-th first feature region 401, i = 1, 2, 3 or 5.

[0200] In some embodiments, the second processing module 530 is further configured to determine 4N first similarities based on four first feature point descriptors and N sample feature point descriptors, wherein the N sample feature point descriptors are used to describe the feature points of the sample feature regions of the N sample banknote images, and the sample version information corresponding to the N sample banknote images is different.

[0201] N is a positive integer greater than 1.

[0202] The third processing module 540 is also used to determine the orientation recognition result and version recognition result of the banknote to be identified based on 4N first similarity scores, sample orientation information and sample version information.

[0203] In some embodiments, the third processing module 540 is further configured to determine N target similarities based on 4N first similarities, wherein the N target similarities correspond one-to-one with the N sample banknote images, and the target similarity is used to characterize the similarity between the target feature region of the banknote to be identified and the sample feature region in the corresponding sample banknote image, wherein the first similarity corresponding to the target feature region is greater than the first similarity corresponding to the other three first feature regions 401 among the four first feature regions 401.

[0204] Based on N target similarities, a target sample banknote version image is determined. The target similarity corresponding to the target sample banknote version image is greater than the target similarity corresponding to the other N-1 sample banknote images among the N sample banknote images.

[0205] Based on the sample orientation information and sample version information corresponding to the target sample banknote version image, the orientation recognition result and version recognition result of the banknote to be identified are determined.

[0206] The banknote orientation recognition device in this application embodiment can be an electronic device or a component of an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices besides a terminal. For example, the electronic device can be a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc. It can also be a server, network attached storage (NAS), personal computer (PC), television set (TV), ATM, or self-service machine, etc. This application embodiment does not specifically limit the device.

[0207] The banknote orientation recognition device in this application embodiment can be a device with an operating system. This operating system can be Android, iOS, or other possible operating systems; this application embodiment does not specifically limit it.

[0208] The banknote orientation recognition device provided in this application embodiment can achieve... Figures 1 to 4 The various processes implemented in the method implementation examples will not be described again here to avoid repetition.

[0209] In some embodiments, such as Figure 6 As shown, this application embodiment also provides an electronic device 600, including a processor 601, a memory 602, and a computer program stored in the memory 602 and executable on the processor 601. When the program is executed by the processor 601, it implements the various processes of the above-described banknote orientation recognition method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0210] It should be noted that the electronic devices in the embodiments of this application include the mobile electronic devices and non-mobile electronic devices described above.

[0211] This application also provides a non-transitory computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described banknote orientation recognition method embodiment and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0212] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0213] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described banknote orientation recognition method.

[0214] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0215] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0216] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0217] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

[0218] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0219] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for recognizing the orientation of banknotes, characterized in that, include: Acquire a first image and a second image of the banknote to be identified, wherein the first image includes pixel information of a first side of the banknote to be identified, and the second image includes pixel information of a second side of the banknote to be identified; Based on the first image and the second image, four first feature point descriptors are determined. The four first feature point descriptors are used to describe the feature points of four first feature regions. Two of the four first feature regions are located in the first image, and the other two are located in the second image. Based on the four first feature point descriptors and the sample feature point descriptors, four first similarities are determined. The sample feature point descriptors are used to describe the feature points of the sample feature region in the sample banknote image. Based on the four first similarities and the sample orientation information of the sample banknote image, the orientation recognition result of the banknote to be identified is determined; Wherein, the two first feature regions in the first image are centrally symmetrical about the center point of the first image, the two first feature regions in the second image are centrally symmetrical about the center point of the second image, one of the two first feature regions in the first image corresponds to the sample feature region and has the same image position, and one of the two first feature regions in the second image corresponds to the sample feature region and has the same image position; The determination of four first similarities based on the four first feature point descriptors and the sample feature point descriptors includes: Based on the four first feature point descriptors and the sample feature point descriptors, four feature point pairs are determined. The four feature point pairs are used to characterize the matching relationship between the feature points of the four first feature regions and the feature points of the sample feature regions. Identify the sample points and outliers in each pair of feature points; The four first similarities are determined based on the sample points and the outliers in each of the feature point pairs.

2. The banknote orientation recognition method according to claim 1, characterized in that, The preset orientations of the four first feature regions are respectively front-facing, front-back, back-facing, and back-back. The determination of the orientation recognition result of the banknote to be identified based on the four first similarities and the sample orientation information of the sample banknote image includes: Based on the four first similarities, the target feature region of the banknote to be identified is determined. The target feature region is one of the four first feature regions, and the first similarity corresponding to the target feature region is greater than the first similarity corresponding to the other three first feature regions among the four first feature regions. The orientation recognition result is determined based on the preset orientation of the target feature region and the orientation information of the sample.

3. The banknote orientation recognition method according to claim 1, characterized in that, The determination of the four first similarities based on the sample points and outliers in each feature point pair includes: Application formula Determine the first similarity; Among them, S i p represents the first similarity corresponding to the i-th first feature region. i q represents the number of sample points in the feature point pair corresponding to the i-th first feature region. i The number of outliers in the feature point pair corresponding to the i-th first feature region, i = 1, 2, 3 or 4.

4. The banknote orientation recognition method according to any one of claims 1-3, characterized in that, The determination of four first similarities based on the four first feature point descriptors and the sample feature point descriptors includes: Based on the four first feature point descriptors and the N sample feature point descriptors, 4N first similarities are determined. The N sample feature point descriptors are used to describe the feature points of the sample feature regions of the N sample banknote images. The sample version information corresponding to the N sample banknote images is different, and N is a positive integer greater than 1. The determination of the orientation recognition result of the banknote to be identified based on the four first similarities and the sample orientation information of the sample banknote image includes: Based on the 4N first similarities, the sample orientation information, and the sample version information, the orientation recognition result and version recognition result of the banknote to be identified are determined.

5. The banknote orientation recognition method according to claim 4, characterized in that, The determination of the orientation recognition result and version recognition result of the banknote to be identified based on the 4N first similarities, the sample orientation information, and the sample version information includes: Based on the 4N first similarities, N target similarities are determined. The N target similarities correspond one-to-one with the N sample banknote images. The target similarity is used to characterize the similarity between the target feature region of the banknote to be identified and the sample feature region in the corresponding sample banknote image. The first similarity corresponding to the target feature region is greater than the first similarity corresponding to the other three first feature regions among the four first feature regions. Based on the N target similarities, a target sample banknote version image is determined, wherein the target similarity corresponding to the target sample banknote version image is greater than the target similarity corresponding to the other N-1 sample banknote version images among the N sample banknote images; Based on the sample orientation information and sample version information corresponding to the target sample banknote version image, the orientation recognition result and version recognition result of the banknote to be identified are determined.

6. A banknote orientation recognition device, characterized in that, include: The acquisition module is used to acquire a first image and a second image of the banknote to be identified. The first image includes pixel information of a first side of the banknote to be identified, and the second image includes pixel information of a second side of the banknote to be identified. A first processing module is configured to determine four first feature point descriptors based on the first image and the second image. The four first feature point descriptors are used to describe feature points of four first feature regions. Two of the four first feature regions are located in the first image, and the other two are located in the second image. The second processing module is used to determine four first similarities based on the four first feature point descriptors and the sample feature point descriptors, wherein the sample feature point descriptors are used to describe the feature points of the sample feature region in the sample banknote image. The third processing module is used to determine the orientation recognition result of the banknote to be identified based on the four first similarities and the sample orientation information of the sample banknote image; Wherein, the two first feature regions in the first image are centrally symmetrical about the center point of the first image, the two first feature regions in the second image are centrally symmetrical about the center point of the second image, one of the two first feature regions in the first image corresponds to the sample feature region and has the same image position, and one of the two first feature regions in the second image corresponds to the sample feature region and has the same image position; The determination of four first similarities based on the four first feature point descriptors and the sample feature point descriptors includes: Based on the four first feature point descriptors and the sample feature point descriptors, four feature point pairs are determined. The four feature point pairs are used to characterize the matching relationship between the feature points of the four first feature regions and the feature points of the sample feature regions. Identify the sample points and outliers in each pair of feature points; The four first similarities are determined based on the sample points and the outliers in each of the feature point pairs.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the banknote orientation recognition method as described in any one of claims 1-5.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the banknote orientation recognition method as described in any one of claims 1-5.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the banknote orientation recognition method as described in any one of claims 1-5.

Images