Method for generating a palm print sample, apparatus for generating a palm print sample, computer device, and computer program

Abstract

This application discloses a method, device, apparatus, medium and program product for generating palm print samples, which relates to the field of machine learning. The method includes: generating positioning point data according to the distribution law of palm print main lines (step 210); generating adjustment point data according to the radian law of palm print main lines (step 220); generating palm print main lines based on the first data, the second data and the adjustment point data (step 230); and generating at least one palm print sample including the palm print main lines (step 240). This application can be applied to various situations such as cloud technology, artificial intelligence, smart transportation, etc.

Description

[Technical field]

[0001] This application claims priority to a Chinese patent application bearing application number 202210189742.8 and entitled "Method, apparatus, device, medium, and program product for generating palm print samples" filed with the China Patent Office on February 28, 2022, the entire contents of which are incorporated herein by reference.

[0002] The embodiments of the present application relate to the field of machine learning, in particular to a method for generating palm print samples; Palm print sample generation Device, computer device 、 and computer Program M Regarding. [Background technology]

[0003] With the rapid development of information technology, palm print authentication technology is increasingly widely used in various identity authentication situations due to its reliability and convenience. Palm print authentication is a method of identity verification based on the main lines, patterns, wrinkles, etc. of the palm, and is a non-invasive authentication method compared with face authentication, which is more easily accepted by users.

[0004] In related art, when authenticating palm prints, a technical solution based on deep learning is usually used to learn the inherent rules of palm print information in stored palm print images, train a model to learn discriminatory latent features, and use the trained model to analyze the palm prints to complete the identity authentication process.

[0005] However, when authenticating identity information using the above-mentioned deep learning, the deep web model usually relies on a large set of palmprint images and accurate label information. However, since palmprint information has high privacy and security requirements, there is a lack of a large public dataset for training the model in the field of palmprint recognition, which reduces the effectiveness of the model in authenticating identity information. Summary of the Invention [Problem to be solved by the invention]

[0006] The present application provides a method for generating a palm print sample, Palm print sample generation Device, computer device 、 and computer Program M to provide. [Means for solving the problem]

[0007] In one aspect, a method for generating palm print samples executed by a computer device is provided, the method including: generating positioning point data including first data corresponding to positioning points of a first main line and second data corresponding to positioning points of a second main line according to a distribution law of palm print main lines; generating adjustment point data according to a radian degree law of palm print main lines, wherein the adjustment points of the main line corresponding to the adjustment point data are used to control the radian degree of a main line composed of the positioning points of the first main line and the positioning points of the second main line; generating a palm print main line, which is a curve sequentially connecting the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, based on the first data, the second data, and the adjustment point data; and generating at least one palm print sample including the palm print main line, wherein the palm print sample is used to train a palm print authentication model, and the palm print authentication model is used to perform palm print authentication.

[0008] In another aspect, a palmprint sample generating device is provided, the device including: a positioning point generating module that generates positioning point data including first data corresponding to positioning points of a first main line and second data corresponding to positioning points of a second main line according to a distribution law of palmprint main lines; an adjustment point generating module that generates adjustment point data according to a radian degree law of palmprint main lines, the adjustment points of the main lines corresponding to the adjustment point data being used to control the radian degree of a main line consisting of the positioning points of the first main line and the positioning points of the second main line; a main line generating module that generates a palmprint main line, which is a curve that sequentially connects the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, based on the first data, the second data, and the adjustment point data; and a sample generating module that generates at least one palmprint sample including the palmprint main line, the palmprint sample being used to train a palmprint authentication model, and the palmprint authentication model being used to perform palmprint authentication.

[0009] In another aspect, a method and apparatus includes a processor and a memory, the memory storing at least one command, at least one program, code set, or command set, the at least one command, the at least one program, code set, or command set being loaded and executed by the processor. generating positioning point data including first data corresponding to the positioning points of a first main line and second data corresponding to the positioning points of a second main line according to a distribution law of palm-print main lines; generating adjustment point data according to a radian degree law of palm-print main lines, wherein the adjustment points of the main lines corresponding to the adjustment point data are used to control the radian degree of a main line consisting of the positioning points of the first main line and the positioning points of the second main line; generating a palm-print main line, which is a curve sequentially connecting the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, based on the first data, the second data, and the adjustment point data; and generating at least one palm-print sample including the palm-print main line, wherein the palm-print sample is used to train a palm-print authentication model, and the palm-print authentication model is used to perform palm-print authentication. Provide computer equipment.

[0010] In another embodiment, at least one command, at least one program, code set or command set is stored, and the at least one command, the at least one program, code set or command set is loaded and executed by a processor. generating positioning point data including first data corresponding to the positioning points of a first main line and second data corresponding to the positioning points of a second main line according to a distribution law of palm-print main lines; generating adjustment point data according to a radian degree law of palm-print main lines, wherein the adjustment points of the main lines corresponding to the adjustment point data are used to control the radian degree of a main line consisting of the positioning points of the first main line and the positioning points of the second main line; generating a palm-print main line, which is a curve sequentially connecting the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, based on the first data, the second data, and the adjustment point data; and generating at least one palm-print sample including the palm-print main line, wherein the palm-print sample is used to train a palm-print authentication model, and the palm-print authentication model is used to perform palm-print authentication. A computer-readable storage medium is provided.

[0011] In another embodiment, The present invention includes a computer program or a command, and when the computer program or the command is executed by a processor, the computer program or the command is for implementing the steps of: generating positioning point data including first data corresponding to positioning points of a first main line and second data corresponding to positioning points of a second main line according to a distribution law of palm-print main lines; generating adjustment point data according to a radian degree law of palm-print main lines, wherein the adjustment points of the main lines corresponding to the adjustment point data are used to control the radian degree of a main line composed of the positioning points of the first main line and the positioning points of the second main line; generating a palm-print main line, which is a curve that sequentially connects the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, based on the first data, the second data, and the adjustment point data; and generating at least one palm-print sample including the palm-print main line, wherein the palm-print sample is used to train a palm-print authentication model, and the palm-print authentication model is used to perform palm-print authentication. Computer Programming MA processor of a computing device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, thereby causing the computing device to perform the method for generating a palm print sample according to any of the above embodiments.

[0012] The details of one or more embodiments of the application are set forth in the drawings and description below. Other features, objects, and advantages of the application will become apparent from the description, drawings, and claims. [Brief description of the drawings]

[0013] In order to more clearly describe the technical solutions in the embodiments of the present application, the following briefly introduces the drawings used in the description of the embodiments. The drawings in the following description are only some embodiments of the present application, and it is obvious to those skilled in the art that other drawings can be obtained based on these drawings without using creative efforts.

[0014] [Figure 1] FIG. 1 is a schematic diagram of an implementation environment according to one exemplary embodiment of the present application. [Diagram 2] FIG. 2 is a flow chart of a method for generating a palm print sample according to one exemplary embodiment of the present application. [Diagram 3] FIG. 3 is a schematic diagram of a palm print according to one exemplary embodiment of the present application. [Figure 4] FIG. 4 is a schematic diagram of a plantar print according to one exemplary embodiment of the present application. [Diagram 5] FIG. 5 is a schematic diagram of a palm print partition according to one exemplary embodiment of the present application. [Figure 6] FIG. 6 is a schematic diagram of a Bezier curve according to one exemplary embodiment of the present application. [Figure 7] FIG. 7 is a flowchart of a method for generating a palm print sample according to another exemplary embodiment of the present application. [Figure 8] FIG. 8 is a schematic diagram of determining an area of ​​interest according to one exemplary embodiment of the present application. [Figure 9] FIG. 9 is a schematic diagram of a first area and a second area according to one exemplary embodiment of the present application. [Figure 10] FIG. 10 is a schematic diagram of a third area according to one exemplary embodiment of the present application. [Figure 11] FIG. 11 is a schematic diagram of determining a third area according to one exemplary embodiment of the present application. [Figure 12] FIG. 12 is a flowchart of a method for generating a palm print sample according to another exemplary embodiment of the present application. [Figure 13] FIG. 13 is a schematic diagram of a palm print sample according to one exemplary embodiment of the present application. [Figure 14] FIG. 14 is a schematic diagram of a target image according to one exemplary embodiment of the present application. [Figure 15] FIG. 15 is a flowchart of palm print recognition model training according to one exemplary embodiment of the present application. [Figure 16] FIG. 16 is a flow chart of a palm print authentication process according to one exemplary embodiment of the present application. [Figure 17] FIG. 17 is a structural block diagram of a palm print sample generating device according to one exemplary embodiment of the present application. [Figure 18] FIG. 18 is a structural block diagram of a palm print sample generating apparatus according to another exemplary embodiment of the present application. [Figure 19] FIG. 19 is a structural block diagram of a server according to one exemplary embodiment of the present application. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] In order to make the objectives, technical solutions and advantages of the present application clearer, the following embodiments of the present application are described in more detail in combination with the drawings.

[0016] In the related art, when authenticating palmprints, a technical solution based on deep learning is usually used. The inherent rules of palmprint information in stored palmprint images are learned, and the model is trained to learn discriminatory latent features, and the model obtained by training is used to analyze the palmprint, thereby performing the authentication process of identity information. However, when authenticating identity information using the above deep learning, the deep web model usually relies on a large set of palmprint images and accurate label information, but since palmprint information has high privacy and security, there is a lack of a large public data set for the model to learn in the field of palmprint authentication, which results in a low authentication effect of the model on identity information.

[0017] In an embodiment of the present application, a method for generating palm print samples is provided, which can enhance the robustness of a palm print recognition model trained on the palm print samples by having stronger diversity among the generated palm print samples. The method for generating palm print samples obtained by training of the present application includes at least one of the following situations when applied:

[0018] 1. Palm print recognition model training situation Due to the privacy of palm print data and the complexity of the method for acquiring palm print data, there is a small amount of palm print data stored in the palm print database, and it is difficult to obtain a good training effect when training a palm print authentication model based on the palm print data in the palm print database. For example, when the above palm print sample generating method is adopted, positioning point data is generated according to the distribution rule of the palm print main line, adjustment point data is generated according to the radian law of the palm print main line, and a plurality of palm print main lines are determined by the fixed point data and the adjustment point data, thereby obtaining a plurality of palm print samples including the palm print main lines. Thus, when training a palm print authentication model with palm print samples, the palm print authentication model can learn a variety of palm print features, and the accuracy of the palm print authentication model in the palm print authentication process can be improved.

[0019] II. Palmprint Encryption Situation For example, taking the palm print as an example, different palm print images can be obtained depending on the parameters such as the difference in the shape of the open palm, the change in light during shooting, and the noise of the shooting device. If the same palm is placed on the palm print encryption instrument at different times, it may cause a misjudgment. For example, by adopting the above palm print sample generation method, positioning point data is generated according to the distribution law of the palm print main line, adjustment point data is generated according to the radian law of the palm print main line, and multiple palm print main lines are determined by the fixed point data and the adjustment point data, thereby obtaining multiple palm print samples including the palm print main lines. By using multiple diverse palm print samples as the authentication standard, the display method of the palm print sample can be obtained in more situations. This can greatly overcome the problem of the difficulty of authentication due to the small number of palm prints, and realize the process of analyzing palm prints in more detail.

[0020] It should be noted that the above application scenarios are merely exemplary, and the palm print sample generating method of the present embodiment can also be applied to other scenarios, and the present embodiment is not limited thereto.

[0021] In addition, the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data for analysis, data for storage, data for display, etc.) and signals related to this application are all obtained with the permission of the user or sufficient permission from various parties, and the collection, use and processing of related data must comply with the relevant laws and standards of the relevant countries and regions. For example, the palm print data in this application is obtained under circumstances where sufficient permission has been obtained.

[0022] Next, an implementation environment according to an embodiment of the present application will be described. For example, referring to FIG. 1, the implementation environment involves a terminal 110 and a server 120, and the terminal 110 and the server 120 are connected via a communication network 130.

[0023] In some embodiments, the terminal 110 is installed with an application program having a palmprint data acquisition function. In some embodiments, the terminal 110 is used to send palmprint data to the server 120. Here, the palmprint data not only includes image data corresponding to the palmprint, but also includes geometric data corresponding to the palmprint, etc. The server 120 determines data information such as the distribution rule of the palmprint main line and the radian degree rule of the palmprint main line based on the palmprint data, and can authenticate the palmprint through the palmprint authentication model 121 based on the distribution rule of the palmprint main line and the radian degree rule of the palmprint main line, and selectively display the palmprint authentication result after the palmprint authentication on the terminal 110.

[0024] Here, the palm print authentication model 121 generates first data corresponding to the positioning points of the first main line and second data corresponding to the positioning points of the second main line according to the distribution rule of the palm print main line, generates adjustment point data to control the arc degree of the main line according to the radian degree rule of the palm print main line, takes the curve obtained by sequentially connecting the positioning points of the first main line, the adjustment point of the main line, and the positioning point of the second main line as the palm print main line, generates at least one palm print sample including the palm print main line, and trains the palm print authentication model with the palm print sample. The above process is not the only example of the training process of the palm print authentication model 121.

[0025] The above terminals include, but are not limited to, mobile terminals such as mobile phones, tablet PCs, portable laptop computers, intelligent voice interaction devices, intelligent home appliances, and in-vehicle terminals, and may be realized as desktop computers, etc. The above server may be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, and may be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Networks (CDNs), and big data, artificial intelligence platforms, etc.

[0026] Here, cloud technology refers to a hosting technology that unifies a series of resources such as hardware, application programs, and networks in a wide area network or a local area network to realize data calculation, storage, processing, and sharing. Cloud technology is a collective term for network technology, information technology, integration technology, management platform technology, application technology, etc. that are applied based on the business model of cloud computing, forming a resource pool that can be used as needed, and is flexible and convenient. Cloud computing technology is an important support. For example, in video websites, image websites, and many portal sites, the background service of the technology network system requires a large amount of calculation and storage of resources. With the advanced development and application of the Internet industry, in the future, each item may have its own identification mark, which will be transmitted to the background system for logical processing. Different levels of data are processed separately, and various industry data all require the backing of a powerful system, which can only be achieved by cloud computing.

[0027] In some embodiments, the above servers may be implemented as nodes in a blockchain system.

[0028] It should be noted that the above implementation environment is merely a schematic example. The palm print sample generating method of the present application is specifically applied to a computer device, which may be a terminal or a server, and the method may be independently performed by the terminal or the server itself, or may be realized through the interaction between the terminal and the server.

[0029] In the method for generating palm print samples according to the present application, a first main line positioning point and a second main line positioning point are generated according to a distribution rule of the palm print main line, adjustment point data for controlling the arc degree of the main line is generated according to a palm print main line arc degree rule, the first main line positioning point, the main line adjustment point, and the second main line positioning point are connected in order to obtain a curve as the palm print main line, at least one palm print sample including the palm print main line is generated, and a palm print authentication model is trained with the palm print sample. By the above method, a plurality of palm print samples are obtained by simulating the distribution state of the main lines in the palm print. Since the palm print samples are determined by generating data (the first main line positioning point, the second main line positioning point, and the adjustment point data), the generated palm print samples are large in number, and since there is no upper limit on the quantity, the generated palm print samples can have stronger diversity. When a palm print recognition model is trained based on the generated palm print samples, the palm print recognition model can mine pattern-specific rules and information that do not exist in larger palm print datasets, break through the limited nature of the palm print dataset, and improve the robustness of the palm print recognition model.

[0030] Combining the above terminology introduction and application scenario, the palm print sample generating method of the present application will be described. Taking the method as applied to a server as an example, as shown in Figure 2, the method includes the following steps 210 to 240.

[0031] In step 210, positioning point data is generated according to the distribution law of the main palm lines.

[0032] Here, the positioning point data includes first data corresponding to the positioning points of the first main line and second data corresponding to the positioning points of the second main line.

[0033] The distribution law of the main palm lines is used to show the distribution situation of the main lines in the palm print. Optionally, by analyzing the palm prints of many palms of the organism itself, it can be found that the palm print of the palm mainly includes main palm lines and fine palm lines. Here, the general characteristics of the palm print of the palm are at least one of (1) the distribution situation of the main palm lines is relatively determined, and the distribution situation of the fine palm lines is relatively random, and (2) the main palm lines usually have longer, thicker, and deeper texture features in the palm print, and the fine palm lines have thinner, shorter, and shallower texture features compared with the main palm lines.

[0034] Optionally, when the palm print is the palm print of the palm, the distribution situation of the main palm lines usually shows a diagonal relationship, that is, the distribution law of the main palm lines of the palm print of the palm is the diagonal law. Typically, taking the palm of the left hand as an example, the main palm lines generally start from the upper left and end at the lower right. Taking the palm of the right hand as an example, the main palm lines generally start from the upper right and end at the lower left. Fig. 3 shows a schematic diagram of the palm print of the palm of the left hand. In the central area 310 of the palm, palm print main line 320 and palm print fine line 330 are included. The thick line is palm print the main line 320, and the thin line is palm print the fine line 330. That is, there are three palm print main lines 320 and thirteen palm print fine lines 330 in the central area 310 of the palm.

[0035] Optionally, when the palm print is the sole print, the distribution situation of the main palm lines usually appears in various forms of palm prints such as various "herringbone" lines, tortoise shell lines, vertical lines, etc. That is, the distribution law of the main palm lines of the sole can be determined in comparison with several common forms.

[0036] Typically, taking the "human" - shaped line on the sole as an example, when there are two main lines of the "human" - shaped line, the two main lines generally start from above and show a form of intersection or connection. Figure 4 shows a schematic diagram of the sole, which includes the "human" - shaped line. The "human" - shaped line includes two main lines. The two main lines intersect above, and the first main line 410 ends at the lower left, and the second main line 420 ends at the lower right. Optionally, when there is one main line of the "human" - shaped line, the main line starts from the lower left and ends at the lower right. Here, the degree of curvature of the main line is large, and at the starting point and the ending point of the main line, a "human" - shaped or inverted "human" - shaped pattern is shown. Or, the main line starts from above and ends at the lower right. Here, the degree of curvature of the main line is large, and at the starting point and the ending point of the main line, a horizontally placed "human" - shaped pattern or the like is shown.

[0037] Typically, taking the vertical line on the sole as an example, the main line of the vertical line generally includes at least one, and the main line generally starts from above and ends below or the like. It should be noted that the above are only schematic examples, and the embodiments of the present application are not limited thereto.

[0038] Optionally, the positioning point data is used to indicate the fixed situation of the main line and includes the first data and the second data. Based on the first data and the second data, the starting situation and the ending situation of the main line are determined, and the distribution situation of the main line is approximately determined. Typically, the first data corresponds to the positioning point of the first main line, and the second data corresponds to the positioning point of the second main line.

[0039] In an alternative embodiment, the starting point of the main line of the palm print is taken as the positioning point of the first main line, and the ending point of the main line of the palm print is taken as the positioning point of the second main line. Typically, the generation of the positioning point data includes at least one of the following methods.

[0040] (1) Random generation method Schematically, in any one coordinate area, the first data and the second data are generated by random generation, and according to the length restriction of the palm print main line, the line segment connecting the positioning point of the first main line and the positioning point of the second main line is scaled at an equal ratio to the coordinate area to determine the positioning point of the first main line corresponding to the first data and the positioning point of the second main line corresponding to the second data.

[0041] Or, the distribution range of the palm print main lines is determined according to the distribution rule of the palm print main lines. For example, the palm of an adult is about 16-22 cm, and the palm print main lines are distributed in the center area of ​​the palm. The center area of ​​the adult palm is about 8-12 cm. Take the size of the center area of ​​the adult palm as an example for analysis. The distribution area of ​​the palm print main lines is preset, and the size of the distribution area may be a square area with a side length of 8 cm, a rectangular area with a side length of 6 cm, a diamond area with a diagonal of 10 cm, etc. Optionally, the positioning points of the first main lines corresponding to the first data and the positioning points of the second main lines corresponding to the second data are generated in a random manner within the preset distribution area of ​​the palm print main lines.

[0042] (2) Generation method within a divided area Optionally, a predetermined distribution area of ​​the palm print main lines is taken as an example for explanation. Within the distribution area of ​​the palm print main lines, the distribution area is divided to obtain divided areas, and first data and second data are generated in the divided areas by a random generation method. Schematically, after dividing the distribution area, two divided areas are obtained, and a positioning point of the first main line corresponding to the first data is generated in the first divided area, and a positioning point of the second main line corresponding to the second data is generated in the second divided area.

[0043] Alternatively, the distribution area may be divided into a plurality of divided areas, and the sum of the areas of the plurality of divided areas may be realized as the entire distribution area or as a part of the distribution area. In other words, the divided area for generating the positioning point data may include the entire distribution area or may include only a part of the distribution area, etc.

[0044] 5, the palm of the left hand will be taken as an example. The palm of the left hand 510 is taken as the distribution area of ​​the palm-print main lines, and the distribution area is divided into four divided areas: upper left area 520, lower left area 530, upper right area 540, and lower right area 550. Usually, on the palm of the left hand, the starting point of the palm-print main lines (positioning point of the first main line) is located in the upper left area 520, and the ending point of the palm-print main lines (positioning point of the second main line) is located in the lower right area 550, so the upper left area 520 is taken as the divided area for generating the first data, and the lower right area 550 is taken as the divided area for generating the second data.

[0045] The above are merely schematic examples, and the embodiments of the present application are not limited thereto.

[0046] In step 220, adjustment point data is generated according to the radian law of palm print main lines.

[0047] Here, the adjustment point of the main line corresponding to the adjustment point data is used to control the arc degree of the main line constituted by the positioning point of the first main line and the positioning point of the second main line.

[0048] The palm print main line radian degree rule is used to describe the radian degree situation of the main line in the palm print. Schematically, the palm print main line is usually not a straight line segment but a curve with a certain radian degree. After determining the positioning point of the first main line and the positioning point of the second main line, the radian degree of the main line formed by the positioning point of the first main line and the positioning point of the second main line is adjusted according to the adjustment point of the generated main line.

[0049] In an optional embodiment, the adjustment point data is determined in the area between the positioning points of the first main line and the positioning points of the second main line according to the radian law of the palm print main lines.

[0050] Schematically, Bezier curves are used to parameterize and describe the geometrical appearance of a palm print. Optionally, at least one Bezier curve is used to describe the principal palm print lines of the palm.

[0051] Here, a Bezier curve is a mathematical curve applied to two-dimensional graphics application programs. A Bezier curve is composed of line segments and nodes, where the nodes are draggable fulcrums and the line segments are like elastic rubber bands. When controlling the shape of a line segment, the nodes control the arc degree of the line segment to obtain the corresponding curve.

[0052] Optionally, an arbitrary palm print main line will be taken as an example for explanation. A palm print main line is constructed using a quadratic Bezier curve. That is, one palm print main line (Bezier curve) is determined using three data (parameter points) on a two-dimensional (2D, 2-dimensional) plane. Here, the three data are first data representing a positioning point of a first main line, an adjustment point of the main line, and second data representing a positioning point of a second main line.

[0053] Optionally, FIG. 6 is a schematic diagram of determining palm print main lines using the method of Bezier curves. The area enclosed by the horizontal and vertical axes is the area of ​​generating palm print main lines, and the numbers written on the horizontal and vertical axes are used to assist in determining the coordinate positions of three data. Here, the schematic diagram includes three palm print main lines, each of which is determined by a first data, an adjustment point of the main lines, and a second data, and the "inverted triangle code" indicates the positioning point of the first main line corresponding to the first data, the "star code" indicates the adjustment point of the main line, and the "circle code" indicates the positioning point of the second main line corresponding to the second data.

[0054] Schematically, first generate a first main line positioning point corresponding to the first data and a second main line positioning point corresponding to the second data, and then generate a main line adjustment point between the first main line positioning point and the second main line positioning point. Optionally, generate the main line adjustment point according to a limiting relationship of the coordinate area between the first main line positioning point and the second main line positioning point.

[0055] For example, the coordinates of the positioning point of the first main line of the palm print main line 610 are (0.0,0.4), the coordinates of the adjustment point of the main line are (0.5,0.6), and the coordinates of the positioning point of the second main line are (0.6,1.0). The coordinates of the positioning point of the first main line of the palm print main line 620 are (0.0,0.0), the coordinates of the adjustment point of the main line are (0.7,0.3), and the coordinates of the positioning point of the second main line are (1.0,1.0). The coordinates of the positioning point of the first main line of the palm print main line 630 are (0.4,0.0), the coordinates of the adjustment point of the main line are (0.6,0.2), and the coordinates of the positioning point of the second main line are (1.0,0.3).

[0056] The above are merely schematic examples, and the embodiments of the present application are not limited thereto.

[0057] In step 230, a palm print main line is generated based on the first data, the second data, and the adjustment point data.

[0058] Here, the palm print main line is a curve that sequentially connects the positioning point of the first main line, the adjustment point of the main line, and the positioning point of the second main line.

[0059] 6, the palm print main line 610 is obtained by sequentially connecting the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, and is shown as a circular arc-shaped curve that protrudes slightly upward. The palm print main line 620 is obtained by sequentially connecting the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, and is shown as a circular arc-shaped curve that protrudes slightly upward. The palm print main line 630 is obtained by sequentially connecting the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, and is shown as a circular arc-shaped curve that recesses downward.

[0060] In step 240, at least one palm print sample including the main palm print lines is generated.

[0061] Here, the palm print samples are used to train a palm print recognition model, and the palm print recognition model is used for palm print recognition.

[0062] Schematically, one palm print sample includes at least one palm print main line, and for example, when observation is performed based on palm print data possessed by an organism itself, the number of palm print main lines is generally 2 to 5. That is, when the palm print situation possessed by an organism itself is used as the generation standard, one palm print sample usually includes two palm print main lines, or one palm print sample includes three palm print main lines, or one palm print sample includes four palm print main lines, or one palm print sample includes five palm print main lines.

[0063] In one optional embodiment, at least one palm print main line is obtained based on the above method of generating a palm print main line using positioning point data and adjustment points of the main line, and then multiple palm print main lines are generated using the same palm print main line generation method.

[0064] Alternatively, the positioning point data corresponding to the multiple palm print main lines may be the same or different. For example, the positioning points of the first main lines corresponding to the palm print main lines 1 and 2 are all point A, but the positioning point of the second main line of the palm print main lines 1 is point B, and the positioning point of the second main line of the palm print main lines 2 is point C. Or, the positioning point of the second main line corresponding to the palm print main lines 1 and 2 is point C, but the positioning point of the first main line of the palm print main lines 1 is point A, and the positioning point of the first main line of the palm print main lines 2 is point B. Or, the positioning points of the first main lines corresponding to the palm print main lines 1 and 2 are all point A, and the positioning points of the second main lines are all point B, but the adjustment point of the main line of the palm print main lines 1 is point C, and the adjustment point of the main line of the palm print main lines 2 is point D, etc. Schematically, 2 to 5 palm print main lines are randomly selected from the multiple generated palm print main lines to obtain a palm print sample.

[0065] In one alternative embodiment, a palm print sample includes three main lines, and the palm print sample includes not only the three main lines but also thin lines, which have shorter and shallower pattern characteristics than the main lines. A process for generating the thin lines is described below.

[0066] Optionally, positioning points of the at least two palm print thin lines are determined, and a palm print thin line is generated based on the positioning points of the at least two palm print thin lines.

[0067] Here, the positioning points of the palm print thin lines are used to determine the generation range of the thin lines, and typically, at least two positioning points of the palm print thin lines are generated according to the distribution rule of the palm print thin lines.

[0068] Optionally, the distribution rule of the palm print fine lines includes many kinds of distribution rules, such as the length rule, thickness rule, density rule, etc. of the palm print fine lines.

[0069] Schematically, the palm print thin line length rule is used to indicate a length restriction of the palm print thin line, for example, the length of the palm print thin line is shorter than the shortest palm print main line among multiple palm print main lines, or the length of the palm print thin line is shorter than a predetermined length threshold (e.g., 3 centimeters).

[0070] Schematically, the palm print thin line thickness rule is used to indicate a palm print thin line thickness restriction, for example, that the thickness of the palm print thin line is thinner than the thinnest palm print main line among multiple palm print main lines, or that the thickness of the palm print thin line is thinner than a predetermined thickness threshold (e.g., 1 millimeter).

[0071] Schematically, the density law of palm print fine lines is used to indicate the mutual distribution status of at least two palm print fine lines within a palm print generation area, for example, by specifying that at least three palm print fine lines are generated within a predetermined X area (predetermined area) within the palm print generation area, or by dividing the palm print generation area into a number of subareas of unit length (1 centimeter) and specifying that each subarea has at least two palm print fine lines.

[0072] The above-mentioned various distribution rules may be applied alone or in combination, for example, by determining palm print fine lines using only the length rule, or by comprehensively considering the length rule, the roughness rule, and the density rule to determine palm print fine lines. The above are merely schematic examples, and the embodiments of the present application are not limited thereto.

[0073] Optionally, in the distribution of palm prints, the distribution of palm print fine lines is relatively dispersed and random, and according to the distribution rule of the palm print fine lines and the palm print distribution situation of the palm print data stored in the palm print database, at least one of the following methods is used to generate at least one palm print fine line in the palm print generation area:

[0074] (1) After determining the positioning point of at least one palm print fine line, determine the positioning point of at least one remaining palm print fine line according to the distribution rule of the palm print fine lines.

[0075] Illustratively, the predetermined length threshold of the palm print thin line is pre-determined to be 3 centimeters, and the predetermined thickness threshold of the palm print thin line is pre-determined to be 1 millimeter. In the palm print generation area, a positioning point of one palm print thin line is randomly generated, and within the predetermined length threshold and the predetermined thickness threshold of the palm print thin line, a positioning point of at least one other palm print thin line is determined, and at least two palm print thin line positioning points are obtained. Optionally, based on the positioning point of at least one other palm print thin line, a positioning point of the other palm print thin line is further determined within the predetermined length threshold and the predetermined thickness threshold of the palm print thin line, etc.

[0076] (2) Based on the distribution rule of palm print fine lines, at least two palm print fine line positioning points are randomly determined.

[0077] Schematically, in the palmprint generation area, at least two points are randomly generated as positioning points of at least two palmprint thin lines based on the distribution rule of the palmprint thin lines; or, after dividing the palmprint generation area, at least two sub-areas are obtained, and the positioning points of the palmprint thin lines for generating the palmprint thin lines in the sub-areas are determined.

[0078] Optionally, after obtaining at least two palm print fine-line positioning points, the manner of obtaining palm print fine-line includes at least one of the following:

[0079] 1. Connect the positioning points of at least two palm print thin lines to obtain a palm print thin line.

[0080] Schematically, after obtaining the positioning points of at least two generated palm print thin lines, the positioning points of any two palm print thin lines are connected to obtain a palm print thin line in the form of a line segment; or after connecting the positioning points of any multiple palm print thin lines, a palm print thin line with an irregular line segment shape is obtained; or, taking into consideration the condition that the thin line length is short, the positioning points of any multiple (two or more) palm print thin lines are taken as a set of thin line positioning points, and a palm print thin line within a certain length range is obtained.

[0081] 2. Determining at least one palm print sample according to the palm print thin line adjustment points generated by the at least two palm print thin line positioning points.

[0082] In one alternative embodiment, an adjustment point of a palm print fine line is determined based on the positioning points of at least two palm print fine lines.

[0083] Here, the adjustment points of the palm print thin line are used to control the arc degree between the positioning points of at least two palm print thin lines. Schematically, when generating a palm print thin line, the method of Bezier curve is used to determine the palm print thin line. That is, after generating the positioning points of at least two palm print thin lines, the adjustment points of the palm print thin line are determined to adjust the arc degree of the line segment between the positioning points of at least two palm print thin lines, and through the process of moving the position of the adjustment points of the palm print thin line, different palm print thin lines corresponding to the palm print thin line adjustment points at different positions are obtained. For example, a thin line represented as an arc line form with a large arc degree, or an irregular curve form with a small curvature width, etc. is obtained.

[0084] In one alternative embodiment, the palm print fine lines are determined based on at least two palm print fine line positioning points and palm print fine line adjustment points in a preset palm print fine line quantity range.

[0085] Optionally, the number of palmprint main lines and palmprint thin lines is preset to make the palmprint situation in the generated palmprint sample similar to the palmprint situation of a living organism. Schematically, an observation result in a certain central area of ​​a palm is taken as an example. In the palmprint situation of a living organism, the number of palmprint main lines in the central area of ​​the palm is 2 to 5, and the number of palmprint thin lines is 5 to 15. For example, the number of palmprint main lines is preset to 2 to 5, and the number of palmprint thin lines is preset to 5 to 15, so that when determining the palmprint main lines, the number of palmprint main lines is controlled within the number range of 2 to 5, and at most four types of palmprint main lines are obtained, and when determining the palmprint thin lines, the number of palmprint thin lines is controlled within the number range of 5 to 15, and at most 11 types of palmprint thin lines are obtained.

[0086] In one alternative embodiment, at least one palm print sample including palm print main lines and palm print fine lines is generated within a preset palm print quantity range.

[0087] Here, the range of palmprint numbers includes at least one of the range of palmprint main lines and the range of palmprint fine lines. Optionally, the palmprint sample includes a certain number of palmprint main lines and palmprint fine lines, and when the palmprint sample is obtained based on the palmprint status of a living organism, the number of palmprint main lines and palmprint fine lines in the palmprint sample is similar to the number of palmprint main lines and palmprint fine lines in the living organism.

[0088] Schematically, in a certain central area of ​​a palm, the number of palm print main lines is controlled to 2 to 5, and the number of palm print thin lines is controlled to 5 to 15, to obtain a plurality of palm print samples. For example, the number of palm print main lines in the first palm print sample is 3, and the number of palm print thin lines is 12, the number of palm print main lines in the second palm print sample is 5, and the number of palm print thin lines is 10, the number of palm print main lines in the third palm print sample is 4, and the number of palm print thin lines is 5, etc.

[0089] In one optional embodiment, the generated palm print sample only includes palm print main lines, and the different palm print samples are determined according to the extending directions of different palm print main lines and the distribution relationship between the multiple palm print main lines, where the extending directions of the palm print main lines indicate the relationship between the positioning points of the first main lines, the adjustment points of the main lines and the positioning points of the second main lines, and the distribution relationship between the palm print main lines indicates the relationship (such as crossing relationship, parallel relationship, distance relationship, etc.) between the multiple palm print main lines belonging to one palm print sample.

[0090] Optionally, in consideration of factors such as palm posture, photographing angle, etc. in the distribution of palm prints, slight differences are generated in palm prints on different photographs of the same palm that are finally imaged. Schematically, a slight perturbation is applied to the generated palm print sample, and the palm print sample after the slight perturbation is applied is regarded as palm print data corresponding to the same identification. That is, multiple target samples obtained by perturbing the palm print sample within the target perturbation interval are regarded as palm print data corresponding to the same identification (ID, Identity Document).

[0091] Optionally, a perturbation interval of minor perturbation is predefined, and the perturbation operation performed on the palm print sample in the perturbation interval is the minor perturbation. Here, perturbing the palm print sample includes at least one of the following realization manners:

[0092] (1) Perturb the principal lines in the palm print sample. Here, the palm print main line in the palm print sample is a single smooth curve formed by sequentially connecting the positioning point of the first main line, the adjustment point of the main line, and the positioning point of the second main line. When perturbing the palm print main line in the palm print sample, the method includes perturbing any one of the positioning point of the first main line, the adjustment point of the main line, and the positioning point of the second main line on the palm print main line within the perturbation section, as well as perturbing any two of the positioning point of the first main line, the adjustment point of the main line, and the positioning point of the second main line on the palm print main line within the perturbation section, and further includes simultaneously perturbing the positioning point of the first main line, the adjustment point of the main line, and the positioning point of the second main line on the palm print main line within the perturbation section.

[0093] Optionally, the preset perturbation intervals include a perturbation interval X of the positioning points of the first main line, a perturbation interval Y of the adjustment points of the main line, and a perturbation interval Z of the positioning points of the second main line, and the predetermined perturbation ranges of the perturbation interval X, the perturbation interval Y, and the perturbation interval Z may be the same or different.

[0094] (2) Perturb the palm print fine lines in the palm print sample. Here, the palm print thin line includes a straight line formed by two palm print thin line positioning points, and also includes a curve formed by the palm print thin line positioning points and the palm print thin line adjustment points. When the palm print thin line is a straight line formed by two palm print thin line positioning points, the process of perturbing the palm print thin line in the palm print sample is realized by perturbing any one of the palm print thin line positioning points in the palm print thin line, or simultaneously perturbing the palm print thin line positioning points in the palm print thin line, within the perturbation interval. When the palm print thin line is a curve formed by the palm print thin line positioning points and the thin line adjustment points, the process of perturbing the palm print thin line in the palm print sample is realized by perturbing any one of the palm print thin line positioning points in the palm print sample, within the perturbation interval, or perturbing the thin line adjustment points in the palm print thin line, or simultaneously perturbing the palm print thin line positioning points in the palm print thin line and the palm print adjustment points in the perturbation interval, etc.

[0095] (3) Perturb the palm print main lines and fine lines in the palm print sample. Here, the palm print main line in the palm print sample is a smooth curve formed by sequentially connecting the positioning point of the first main line, the adjustment point of the main line, and the positioning point of the second main line, and the palm print thin line includes not only a straight line formed by the positioning points of the two palm print thin lines, but also a curve formed by the positioning point of the palm print thin line and the palm print adjustment point.

[0096] When perturbing the palm print main lines and palm print thin lines in a palm print sample, within the perturbation section, one or more of the positioning points of the first main line, the adjustment points of the main lines, and the positioning points of the second main line in the palm print main lines are perturbed, and one or more of the positioning points of the palm print thin lines and the adjustment points of the thin lines in the palm print thin lines are perturbed.

[0097] Optionally, based on the above perturbation process for the palmprint main lines and the palmprint thin lines in the perturbation interval range, perturbed palmprint samples corresponding to the multiple palmprint samples are obtained, and the multiple perturbed palmprint samples and the unperturbed palmprint sample are regarded as palmprint data corresponding to the same ID.

[0098] For example, within the perturbation interval range, palmprint sample A is slightly perturbed using the perturbation method described above to obtain palmprint sample B after perturbing the main lines in palmprint sample A, palmprint sample C, and palmprint sample D after perturbing the main lines and thin lines in palmprint sample A, and palmprint sample A, palmprint sample B, palmprint sample C, and palmprint sample D are treated as palmprint data corresponding to the same ID.

[0099] It should be noted that the above is merely a schematic example, and the embodiments of the present application are not limited thereto.

[0100] As described above, the first main line positioning point and the second main line positioning point are generated according to the distribution rule of the palm print main line, the adjustment point data for controlling the radian degree of the main line is generated according to the radian degree rule of the palm print main line, the curve obtained by sequentially connecting the first main line positioning point, the adjustment point of the main line, and the second main line positioning point is defined as the palm print main line, at least one palm print sample including the palm print main line is generated, and the palm print authentication model is trained with the palm print sample. Through the above method, a plurality of palm print samples are obtained by simulating the distribution state of the main lines in the palm print. Since the palm print samples are determined by generating data (the first main line positioning point, the second main line positioning point, and the adjustment point data), the generated palm print samples are large in number, and since there is no upper limit on the quantity, the generated palm print samples can have stronger diversity. When a palm print recognition model is trained based on the generated palm print samples, the palm print recognition model can mine pattern-specific rules and information that do not exist in larger palm print datasets, break through the limited nature of the palm print dataset, and improve the robustness of the palm print recognition model.

[0101] In one optional embodiment, as shown in FIG. 7, the process of generating the first main line positioning points, the second main line positioning points, and the main line adjustment points according to the palm print main line distribution rule and the palm print main line radian law can further realize the following steps 710 to 770.

[0102] In step 710, a first area and a second area corresponding to the extending direction of the palm print main line are determined according to the distribution rule of the palm print main line.

[0103] A palm print is taken as an example for a schematic explanation. In a palm print, the distribution of the palm print main lines follows a certain rule. FIG. 3 shows a schematic diagram of the distribution of the palm print on a palm. Here, the palm print main lines 320 are displayed in a display format in which the upper left is the positioning point of the first main line and the lower right is the positioning point of the second main line. Alternatively, the lower right is regarded as the positioning point of the first main line of the palm print main lines 320, and the upper left is regarded as the positioning point of the second main line of the palm print main lines 320. Alternatively, the palm print main lines are analyzed in a format in which the upper left is the positioning point of the first main line and the lower right is the positioning point of the second main line.

[0104] In one alternative embodiment, the palm print generating area is determined according to the distribution rule of the palm print main lines.

[0105] Here, the palm print generation area is used to limit the distribution range of the palm print main lines.

[0106] Schematically, the palmprint generation area is determined by a random generation method, for example, generating a rectangular area whose side length is a unit length (1 cm), or generating an irregularly shaped area whose maximum diagonal length is a predetermined length, etc.

[0107] Optionally, the palmprint generation area may be determined according to the distribution of the palmprint on the palm. Schematically, FIG. 8 shows a schematic diagram of a palmprint. The palmprint distribution of the palmprint shown in FIG. 8 is taken as an example for explanation. The process of determining the palmprint generation area according to the distribution rule of the palmprint main line includes the following processes:

[0108] (1) Locating key points between fingers Optionally, a target detector is used to detect the spaces between the fingers of the hand, and a key point A between the index finger 810 and the middle finger 820, a key point B between the middle finger 820 and the ring finger 830, and a key point C between the ring finger 830 and the little finger are determined, and the key point A, the key point B, and the key point C are determined as three inter-finger key points, which are the positioning results of the inter-finger key point positioning. Schematically, when a target detector is used to detect the spaces between the fingers of the hand, two fingers are finger The midpoint between the fingers is set as the inter-finger keypoint position.

[0109] (2) Determining the local coordinate system In one alternative embodiment, after determining the positions of the inter-finger keypoints according to the inter-finger spaces, a local coordinate system is established according to the keypoints.

[0110] Schematically, key point A between the index finger 810 and middle finger 820 and key point C between the ring finger 830 and little finger are connected, and the straight line obtained by connecting them is determined as the horizontal axis (x-axis) of the local coordinate system. Key point B between the middle finger 820 and ring finger 830 is set as the origin of the local coordinate system, and the vertical axis (y-axis) perpendicular to the horizontal axis is determined, thereby obtaining a local coordinate system constructed by key points A, B, and C.

[0111] Optionally, the center point D of the palmprint generation area is determined along the negative direction of the vertical axis in the local coordinate system. Here, the manner of determining the position of the center point D of the palmprint generation area includes determining according to the width and length of the center of the palm, or determining according to the distance between key points, etc.

[0112] Schematically, when determining the position of the center point D of the palmprint generation area according to the width and length of the center of the palm, the position is determined based on the width and length of the center of the palm. For example, a rectangle is constructed with the width and length of the center of the palm as the side lengths. The position of the center point D of the palmprint generation area is determined based on the intersection of the diagonals of the rectangle.

[0113] Schematically, when determining the position of the center point D of the palmprint generation area according to the distance between key points, first determine the BD distance between key point B and the center point D of the palmprint generation area, and the AC distance between key point A and key point C. Then, determine the position of the center point D of the palmprint generation area according to the length relationship between the BD distance and the AC distance. For example, by setting the BD distance as 1.5 times the AC distance, the center point D of the palmprint generation area is determined according to the AC distance and the constructed local coordinate system.

[0114] (3) Extraction of palm print generation area Alternatively, after determining the center point D of the palmprint generation area, the palmprint generation area is determined according to the center point of the palmprint generation area. For example, a rectangular area with a certain side length is constructed with the center point D of the palmprint generation area as the center, and this rectangular area is set as the palmprint generation area. Alternatively, an irregular palm-shaped area is constructed with the center point D of the palmprint generation area as the center of gravity, and this irregular palm-shaped area is set as the palmprint generation area, etc.

[0115] An example will be described in which a rectangular area constructed with the center point D of the palmprint generation area as the center is used as the palmprint generation area. The side lengths of the rectangular area are determined according to the length relationships between the key points.

[0116] Schematically, after determining the AC distance between key point A and key point C, the AC distance is set as the side length of the palmprint generation area, or 7 / 6 times the AC distance is set as the side length of the palmprint generation area, etc.

[0117] Alternatively, after determining the AB distance between key point A and key point B, the side length of the palm print generation area is set to twice the AB distance.

[0118] Alternatively, the palmprint generation area can be called a Region Of Interest (ROI), that is, the palmprint generation area is an area that is focused on in the process of generating a palmprint, and the palmprint generation process is carried out in the area.

[0119] Schematically, the direction of the palm print main lines follows a certain rule. For example, the direction of the palm print main lines in a palm print is from the upper left to the lower right. After obtaining the palm print generation area, the first area and the second area corresponding to each other in the palm print generation area are determined based on the direction of the palm print main lines, thereby generating the palm print start positioning point and the palm print end positioning point.

[0120] In one alternative embodiment, a first vertex and a second vertex that are diagonally related in the palm print generation area are determined.

[0121] Schematically, as shown in FIG. 9, the palmprint generation area is one rectangular area, and the vertices that are diagonally related in the rectangular area are determined as the first vertex and the second vertex. Here, the vertices that are diagonally related include vertices 910 and 940, and vertices 920 and 930. Schematically, if vertex 910 is the first vertex, vertex 940 is the second vertex, and if vertex 920 is the first vertex, vertex 930 is the second intersection. Schematically, when analyzing the distribution of palmprints on the left hand, the palmprints on the left hand generally show a direction in which the palmprint extends from the upper left to the lower right. That is, if the palmprint generation area is the rectangular area shown in FIG. 9, the direction in which the palmprint main line extends is from vertex 910 (the first vertex) to vertex 940 (the second vertex).

[0122] In one alternative embodiment, a first area is determined in the palm print generation area with a first vertex as a center and a first predetermined length as a radius.

[0123] Schematically, after the first vertex is determined, the first area determined based on the first vertex is a sector area, where a dot of the sector area is the first vertex, and the radius of the sector area is a first predetermined length. Or, the sector area is regarded as 1 / 4 of a circular area, where a center point of the circular area is the first vertex, and the radius of the circular area is a first predetermined length.

[0124] Alternatively, the first predetermined length may be a preset fixed value, or may be a value determined based on the palm print generation area.

[0125] For example, when the first predetermined length is a preset fixed value and the first area is determined based on the first vertex, the first area is determined in the palmprint generation area with the first vertex as the center and the preset fixed value as the radius. Alternatively, when the first predetermined length is a value determined based on the palmprint generation area (for example, the side length in the palmprint generation area is the diameter, or half the side length in the palmprint generation area is the diameter, etc.) and the first area is determined based on the first vertex, the first area is determined in the palmprint generation area with the first vertex as the center and the value determined based on the palmprint generation area as the radius.

[0126] Schematically, when the palm print generation area is a square area of ​​unit length, as shown in FIG. 9, the dot is the third vertex, and for the positioning point of the first main line, the coordinates of the positioning point of the first main line are defined as follows:

[0127]

number

[0128] Here, x indicates the horizontal coordinate and y indicates the vertical coordinate.

[0129] In one alternative embodiment, a second area is determined in the palm print generation area with the second vertex as the center and the second predetermined length as the radius.

[0130] Optionally, the process of determining the second area according to the second vertex is similar to the process of determining the first area according to the first vertex. Schematically, the length values ​​of the second predetermined length and the first predetermined length may be the same or different.

[0131] For example, as shown in FIG. 9, when the palmprint generation area is a square area and the second predetermined length and the first predetermined length have the same length values, the second area formed with the second vertex as its center and the second predetermined length as its radius, and the first area formed with the first vertex as its center and the first predetermined length as its radius are both sector-shaped areas, and the first area and the second area are areas of the same shape.

[0132] Schematically, when the palm print generation area is a square area of ​​unit length, as shown in FIG. 9, the dot is the third vertex, and for the positioning point of the second main line, the coordinates of the positioning point of the second main line are defined as follows:

[0133]

number

[0134] Here, x indicates the horizontal coordinate and y indicates the vertical coordinate.

[0135] Alternatively, if the second predetermined length and the first predetermined length have different length values, the second area formed with the second vertex as its center and the second predetermined length as its radius will have a different area shape from the first area formed with the first vertex as its center and the first predetermined length as its radius. For example, if the palmprint generation area is a square area, and the second predetermined length has a larger length value and the first predetermined length has a smaller length value, the second area corresponding to the second predetermined length will be larger than the first area corresponding to the first predetermined length.

[0136] It should be noted that the above is merely a schematic example, and the embodiments of the present application are not limited thereto.

[0137] In step 720, a first datum is determined that corresponds to a position point of a first main line within a first area.

[0138] Optionally, after determining the first area, determine, within the first area, first data corresponding to the positioning points of the first main line by a manner of random selection.

[0139] Here, the random selection refers to a selection method with equal probability. Schematically, in the first area, the coordinates corresponding to a certain coordinate point are randomly set as the coordinates of the positioning point of the first main line with equal probability, thereby realizing the process of generating the first data.

[0140] Optionally, the first data corresponding to the positioning point of the first main line is determined in the first area by a selection method not having equal probability according to the generation rule of the palm print main line. For example, by analyzing the distribution state of the palm print, it is found that most of the palm print main lines start from point M, and when determining the first data in the first area, the probability of point M being the positioning point of the first main line is set to be high, thereby realizing the process of selecting the first data by a selection method not having equal probability.

[0141] In step 730, second data corresponding to position points of a second main line within a second area is determined.

[0142] Optionally, after determining the second area, determine second data corresponding to the positioning points of the second main line in the second area by a random selection manner.

[0143] Here, the random selection refers to a selection method with equal probability. Schematically, in the second area, the coordinates corresponding to a certain coordinate point are randomly set as the coordinates of the positioning point of the second main line with equal probability, thereby realizing a process of generating second data.

[0144] Optionally, the second data corresponding to the positioning point of the second main line is determined in the second area by a selection method not having equal probability according to the generation rule of the palm print main line. For example, by analyzing the palm print distribution situation of the palm print, it is found that most of the palm print main lines start from point N and end at point L. When determining the second data in the second area, the probability of the points N and L being the positioning points of the second main line is set to be high, thereby realizing the process of selecting the second data by a selection method not having equal probability.

[0145] It should be noted that the above is merely a schematic example, and the embodiments of the present application are not limited thereto.

[0146] In step 740, a third area is determined based on the positional relationship between the positioning points of the first main line and the positioning points of the second main line according to the radian law of the palm print main line.

[0147] Schematically, the palm print main line radian law indicates the curvature state of the palm print main line, for example, the radian degree of the palm print main line in the palm print is small, the curvature degree of the palm print main line in the palm print is smooth, etc.

[0148] Optionally, a third area for generating adjustment points of the main lines is determined based on a positional relationship between the positioning points of the first main line and the positioning points of the second main line.

[0149] In one alternative embodiment, the positioning points of the first main line and the positioning points of the second main line are connected to obtain a target line segment.

[0150] Schematically, as shown in FIG. 10, in a palm print generation area 1010, a positioning point 1020 of a first main line (shown as an inverted triangle) and a positioning point 1030 of a second main line (shown as a circle) are determined, and the positioning point 1020 of the first main line and the positioning point 1030 of the second main line are connected to obtain a target line segment 1040.

[0151] Optionally, a rectangular area having a predetermined side length and centered on the midpoint of the target line segment is set as the third area.

[0152] Schematically, the predetermined side length may be a fixed value set in advance, or may be a value determined based on the palm print generation area.

[0153] For example, when the predetermined side length is a preset fixed value and the third area is determined based on the midpoint of a line segment, a rectangular area determined within the palm print generation area is set as the third area, with the midpoint of the line segment as the center and the preset fixed value as the side length. For example, the preset fixed value includes a length a of the rectangle and a width b of the rectangle, and the third area is obtained by setting the midpoint of the line segment as the center, the length a of the rectangle as the length of the third area, and the width b of the rectangle as the width of the third area.

[0154] Alternatively, the specified side length is a numerical value determined based on the palmprint generation area (for example, the side length is half the side length in the palmprint generation area, or half the target line segment in the palmprint generation area), and when the first area is determined based on the midpoint of the line segment, a third area is defined within the palmprint generation area with the midpoint of the line segment as the center and the numerical value determined based on the palmprint generation area as the side length.

[0155] For example, as shown in FIG. 10, after acquiring a target line segment 1040, a third area 1050 is acquired by setting the midpoint of the target line segment 1040 as the center and a preset fixed value as the side length.

[0156] In one alternative embodiment, the coordinates of the midpoint of the target line segment connecting the positioning point of the first main line and the positioning point of the second main line are (x c ,y c ), and the midpoint coordinate (x c ,y c ) as the center and the target line as the criterion, the process of determining the third area is as follows:

[0157] Schematically, the length of the palmprint generating area is unit length 1, and the third area to be obtained is a square area. The side length of the square area is preset to 2 / 3. Then, the third area is located at the midpoint coordinate (x c ,y c ) and is a square area parallel to the target line segment with a side length of 2 / 3 of the target line.

[0158] Optionally, the third area is determined by a method of linear equations, the linear equation being 1. It is uniquely determined by the positioning point of the main line and the positioning point of the second main line. Schematically, the straight line on which the target line segment exists that connects the positioning point of the first main line and the positioning point of the second main line is defined as line A, and line A is y=k1x+b1. Similarly, the midpoint coordinates (x c ,y c ) and perpendicular to line A, define line B as y = k2x + b2.

[0159] Here, k1 indicates the slope corresponding to line A, b1 indicates the intercept corresponding to line A, k2 indicates the slope corresponding to line B, and b2 indicates the intercept corresponding to line B. Here, the relationship between k1, b1, k2, and b2 is as follows:

[0160]

number

[0161] That is, the line A and the midpoint coordinate (x c ,y c ) the straight line B can be determined uniquely.

[0162] Alternatively, as shown in FIG. 11, the midpoint coordinate 1110 (x c ,y c ), and based on that the predetermined extension of the third area is 2 / 3, two straight lines A11120 and A21130 are determined that are parallel to straight line A and have a vertical distance of 1 / 3 from straight line A, and two straight lines B11140 and B21150 are determined that are parallel to straight line B and have a vertical distance of 1 / 3 from straight line B.

[0163] The equation of line A11120 is y = k1x + b3, the equation of line A21130 is y = k1x + b4, the equation of line B11140 is y = k2x + b5, and the equation of line B21150 is y = k2x + b6.

[0164] The above are merely schematic examples, and the embodiments of the present application are not limited thereto.

[0165] Schematically, as shown in FIG. 11, the values ​​of the four straight lines, namely, line A11120, line A21130, line B11140, and line B21150, can be uniquely determined, and a square area surrounded by the four straight lines is determined, and this square area is regarded as the third area.

[0166] In step 750, adjustment point data is generated within the third area.

[0167] Optionally, after the third area is determined, adjustment point data for adjusting the main line radian angle within the third area is generated. Schematically, the range of coordinate values ​​of the adjustment point data is defined as follows.

[0168]

number

[0169] Here, k1 indicates the slope corresponding to line A11120 and line A21130, b3 indicates the intercept corresponding to line A11120, b4 indicates the intercept corresponding to line A21130, k2 indicates the slope corresponding to line B11140 and line B21150, b5 indicates the intercept corresponding to line B11140, and b6 indicates the intercept corresponding to line B21150.

[0170] In step 760, a palm print main line is generated based on the first data, the second data, and the adjustment point data.

[0171] Here, the palm print main line is a curve that sequentially connects the positioning point of the first main line, the adjustment point of the main line, and the positioning point of the second main line.

[0172] Schematically, the palm print main line is a curve obtained by sequentially connecting the positioning point of the first main line, the adjustment point of the main line, and the positioning point of the second main line. Here, the coordinate positions of the positioning point of the first main line and the positioning point of the second main line are determined, and the adjustment point of the main line is the positioning point of the first main line and the positioning point of the second main line. Connected By adjusting the radii of the line segments, different palm print main lines can be obtained. That is, the positions of the adjustment points of the main lines and the formation of the palm print main lines are closely related.

[0173] In step 770, at least one palm print sample including palm print main lines and palm print fine lines is generated.

[0174] The palm print samples are used to train a palm print recognition model, and the palm print recognition model is used for palm print recognition.

[0175] Schematically, a palm print sample includes palm print main lines and palm print thin lines. After obtaining at least one palm print sample including palm print main lines, a palm print recognition model is trained based on the at least one palm print sample, and the palm print model learns the relationship and difference between different palm print samples, so that the recognition efficiency in the process of the palm print recognition model performing palm print recognition can be improved.

[0176] Optionally, after obtaining the generated palm print sample, a slight perturbation is applied to the palm print sample within a predetermined perturbation area, and the palm print sample to which the slight perturbation is applied is regarded as palm print data corresponding to the same ID. Schematically, in a predetermined perturbation area, a number of different perturbation operations are performed on one palm print sample to obtain a number of palm print data corresponding to the palm print sample. For example, in a predetermined perturbation area, for one palm print sample, a palm print main line in the palm print sample is perturbed to obtain one palm print data, and a palm print thin line in the palm print sample is perturbed to obtain another palm print data. The palm print sample and the palm print data obtained based on the palm print sample are regarded as palm print data corresponding to the same ID. Through the process of applying a slight perturbation to the palm print sample within a certain perturbation area, more diverse palm print samples can be formed. Alternatively, in the process of perturbing the palm print sample in the perturbation area, the palm print sample may be perturbed once to obtain palm print data belonging to the same ID as the palm print sample, or the palm print sample may be perturbed multiple times to obtain palm print data belonging to the same ID as the palm print sample. Wherein, when the palm print sample is perturbed multiple times, the sum of the perturbations of the multiple perturbations is located within the perturbation interval. The above is merely a schematic example, and the embodiment of the present application is not limited thereto.

[0177] As described above, the curve obtained by sequentially connecting the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line is taken as the palm print main line, and at least one palm print sample including the palm print main line is obtained, and the palm print sample is used to train the palm print recognition model. Through the above method, a large number of palm print samples are simulated and obtained according to the distribution of the main lines in the palm print, so that the generated palm print samples have stronger diversity. Based on the palm print sample, the palm print recognition model is trained, which breaks through the localization of the palm print data set and improves the robustness of the palm print recognition model.

[0178] In the method according to the embodiment of the present application, the process of obtaining palm print samples according to sections is described. According to the distribution rule of the palm print main line, a first area and a second area corresponding to the direction of the palm print main line are determined, a first data is determined in the first area, a second data is determined in the second area, and according to the radian law of the palm print main line, a third area is determined, and adjustment point data is generated in the third area. The positioning point of the first main line corresponding to the first data, the adjustment point of the main line, and the positioning point of the second main line corresponding to the second data are sequentially connected to generate at least one palm print sample including the palm print main line. Through the above method, the distribution rule of the palm print main line and the radian law of the palm print main line are displayed in a section manner, so that the position information of the positioning point of the first main line and the positioning point of the second main line in the palm print main line can be more realistically determined, and the adjustment point of the main line can be determined. A palm print sample including the palm print main line is obtained by sequentially connecting the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, and a slight perturbation is then performed on the generated palm print sample within a predetermined perturbation interval (for example, perturbing one or several points among the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line), thereby obtaining multiple palm print data belonging to the same identification and having greater diversity, and further increasing the diversity of the palm print sample.

[0179] In one optional embodiment, after obtaining at least one palm print sample, training a palm print recognition model with the at least one palm print sample. Schematically, as shown in Fig. 12, after step 240 shown in Fig. 2 above, the method includes the following steps 1210 to 1230.

[0180] In step 1210, a sample image set is obtained.

[0181] Here, at least one sample image is stored in the sample image set.

[0182] Schematically, the sample images in the sample image set include various types, such as landscape images, architectural images, animal images, plant images, etc. Optionally, the sample image set is a large-scale classification image dataset, such as the ImageNet dataset.

[0183] In step 1220, the palm print sample is embedded into the sample image with the sample image as the background to obtain a target image.

[0184] Schematically, a plurality of sample images are selected from a sample image set, and the generated palm print sample is embedded onto the selected sample images with the selected sample images as backgrounds to obtain a target sample including the palm print sample and the sample images.

[0185] Optionally, making the selected sample image the background indicates placing the selected sample image below, and embedding the generated palm print sample on the sample image indicates placing the palm print sample above. For example, the embedding relationship between the sample image and the palm print sample is displayed in a layered form, with layer 1 below layer 2, i.e. the sample image is layer 1, and the target sample is layer 2.

[0186] In one optional embodiment, the color and width of the palm print main lines, palm print fine lines, etc. in the palm print sample are set as c and w, respectively, and the palm print sample is placed in the upper layer, and a sample image I selected from the sample image set is placed in the lower layer, that is, the palm print sample is embedded on the sample image I to obtain a target image. Schematically, the process of embedding and obtaining the target image is as follows:

[0187]

number

[0188] Here, S denotes a target image obtained by incorporating a palmprint sample (including palmprint main lines and palmprint fine lines) onto sample image I, synthesize denotes a process of incorporating a palmprint sample onto sample image I to generate a target image, P denotes the palmprint main lines corresponding to the palmprint sample, and Q denotes the palmprint fine lines corresponding to the palmprint sample.

[0189] Schematically, after a palm print sample is generated, pattern feature information such as color, length, width, etc. of the palm print main lines and palm print thin lines corresponding to the palm print sample is determined. Pattern position information of the palm print main lines and palm print thin lines is also relatively determined. In a target image obtained by incorporating the palm print sample into a sample image based on the above pattern information, the incorporated palm print sample and the non-incorporated palm print sample are the same, that is, the pattern feature information and pattern position information corresponding to the incorporated palm print sample and the non-incorporated palm print sample are the same.

[0190] Schematically, in the process of incorporating the palm print sample into the sample image, when the size of the sample image and the size of the palm print sample generation area are different, the incorporating process may be different. For example, when the size of the sample image is larger than the size of the palm print sample generation area, the palm print sample is directly incorporated on the sample image, or the sample image is reduced to a certain size (for example, the size of the palm print sample generation area), and then the palm print sample is incorporated on the sample image. When the size of the sample image is smaller than the size of the palm print sample generation area, the palm print sample is directly incorporated on the sample image, or the sample image is enlarged to a certain size (for example, the size of the palm print sample generation area), and then the palm print sample is incorporated on the sample image.

[0191] It should be noted that the above is merely a schematic example, and the embodiments of the present application are not limited thereto.

[0192] In one alternative embodiment, at least one palm print sample is perturbed within a target perturbation interval to obtain a target sample.

[0193] Schematically, the target perturbation section includes a palm-print main line perturbation section and a palm-print thin line perturbation section.

[0194] Here, the palm-print main line perturbation section indicates a section range in which the palm-print main lines are perturbed, and the palm-print thin line perturbation section indicates a section range in which the palm-print thin lines are perturbed.

[0195] Optionally, the palm print sample is perturbed within the target perturbation interval to obtain a plurality of target samples. For example, within the palm print main line perturbation interval, the palm print main lines in the palm print sample are perturbed to obtain a plurality of target samples with slight changes in the palm print main lines. Or, within the palm print fine line perturbation interval, the palm print fine lines in the palm print sample are perturbed to obtain a plurality of target samples with slight changes in the palm print sample. Or, within the target perturbation interval, the palm print main lines and palm print fine lines in the palm print sample are perturbed to obtain a plurality of target samples with slight changes in the palm print main lines and palm print fine lines.

[0196] Optionally, in the palm print distribution situation, factors such as palm posture, photographing angle, photographing position, etc. will all cause slight differences in the palm prints on different photographs of the same palm that are finally imaged. Schematically, based on the consideration of improving the robustness of the model, a slight perturbation is added to the generated palm print sample, and the palm print sample after the slight perturbation is considered as palm print data corresponding to the same identity.

[0197] In one alternative embodiment, perturbing the at least one palm print sample is achieved by adding noise to a palm print main line corresponding to the at least one palm print sample, or by adding noise to a palm print thin line corresponding to the at least one palm print sample.

[0198] Schematically, the process of adding perturbation noise to the generated palm print samples is as follows.

[0199]

number

[0200] Here, P i denotes the palm print main line in the i-th palm print sample, and P j i denotes the j-th palm print sample with perturbation noise added based on the palm print main line, and N p denotes the perturbation noise added to the palm line, and Q i denotes the palm print fine line in the i-th palm print sample, and Q j i is palm print thin line Let us denote the j-th palm print sample with perturbation noise based on N q denotes the perturbation noise added to the palm print fine lines. Optionally, the perturbation noise N p ~N(μ,0.04), perturbation noise N q ∼N(μ,0.01) are all very small Gaussian noises.

[0201] In one optional embodiment, within a palm print main line perturbation section, a palm print main line corresponding to at least one palm print sample is perturbed to obtain a perturbed main line; within a palm print thin line perturbation section, a palm print thin line corresponding to at least one palm print sample is perturbed to obtain a perturbed thin line; and based on the perturbed main line and the perturbed thin line, a target sample is obtained.

[0202] Schematically, the above perturbation noise N p is the palm print main line perturbation section corresponding to the palm print main line, and the above perturbation noise N q is the palm-print thin line perturbation interval corresponding to the palm-print thin line, and the perturbation noise N p and the perturbation noise N q These are collectively referred to as the target perturbation interval.

[0203] 13 is a schematic diagram of a set of palm print samples obtained after adding perturbation noise to the palm print samples. Optionally, the palm print sample 1310, the palm print sample 1320, the palm print sample 1330 and the palm print sample 1340 are palm print samples generated according to the above palm print sample generating method.

[0204] Schematically, an example is taken of perturbing a palm print sample 1310. Within a target perturbation interval, a slight Gaussian noise is added to the palm print main lines and palm print thin lines of the palm print sample 1310 to obtain target samples 1311, 1312, and 1313, and the target samples 1311, 1312, and 1313 are palm print data corresponding to the same identification.

[0205] Alternatively, take the example of perturbing the palm print sample 1320. Within the target perturbation interval, add a slight Gaussian noise to the palm print main lines and palm print thin lines of the palm print sample 1320 to obtain target samples 1321, 1322, and 1323, and the target samples 1321, 1322, and 1323 are palm print data corresponding to the same identification.

[0206] Based on the same method as above, target samples 1331, 1332, and 1333 corresponding to palm print sample 1330 are obtained, and target samples 1331, 1332, and 1333 are regarded as palm print data corresponding to the same identification. Furthermore, target samples 1341, 1342, and 1343 corresponding to palm print sample 1340 are obtained, and target samples 1341, 1342, and 1343 are regarded as palm print data corresponding to the same identification.

[0207] That is, in the above process, the number of palm print main lines and palm print fine lines in the target samples corresponding to the same identity is determined, and the slight changes caused by adding perturbation noise within the target perturbation range are allowed, that is, the palm print data obtained after adding perturbation noise within the target perturbation range is still regarded as the palm print data of the same identity.

[0208] It should be noted that the above is merely a schematic example, and the embodiments of the present application are not limited thereto.

[0209] In one alternative embodiment, the sample image is used as a background, the target sample is embedded on the sample image, and the target image is acquired.

[0210] Optionally, making the sample image the background indicates placing the sample image below, and embedding the target sample on the sample image indicates placing the target sample above. For example, if layers show the embedding relationship between images, and layer 1 is below layer 2, then the sample image is layer 1 and the target sample is layer 2.

[0211] 14 is a schematic diagram of a target image obtained by incorporating a target sample onto a sample image. For example, target image 1410 is an image obtained by incorporating target sample 1400 onto a landscape image 1411, and landscape image 1411 is a sample image. Or, target image 1420 is an image obtained by incorporating target sample 1400 onto an animal image 1421, and animal image 1421 is a sample image.

[0212] Alternatively, the sample images may be images having different parameters such as a variety of image sizes, image qualities, etc. For example, the sample images may be images with high clarity or images with low clarity.

[0213] In step 1230, a palm print recognition model is trained with the target images.

[0214] In one optional embodiment, a palm print recognition model is first trained on a target image to obtain a candidate palm print recognition model.

[0215] Schematically, after obtaining at least one target image corresponding to each of a plurality of identification documents, the target image is used as an input to a palm print recognition model, and a first training is performed on the palm print recognition model.

[0216] For example, after obtaining one palmprint sample by the above palmprint generation method, the palmprint main lines and palmprint thin lines in the palmprint sample are perturbed to obtain multiple target samples, the multiple target samples correspond to identification A, and each target sample is placed on a different type of sample image to obtain multiple target images, which also correspond to identification A, and the multiple target images corresponding to identification A are input into the palmprint authentication model, and then the palmprint authentication model is trained on the multiple target images, thereby preventing the occurrence of overtraining problems with the color, width, and background content of the palmprint main lines, palmprint thin lines, etc. in the palmprint sample.

[0217] Optionally, the color and width of the palm print main lines, thin lines, etc. in the palm print sample are set as c and w, respectively, and a sample image I is randomly selected from the sample image set as the background of the target image. Schematically, the outline of the generated palm print pattern is as follows:

[0218]

number

[0219] Here, synthesize refers to the process of generating the target image, and S j i denotes the target image obtained by embedding the generated palm print sample (including palm print main lines and palm print fine lines) on the sample image I, and P j i denotes the j-th palm print sample with perturbation noise added based on the palm print main line, and Q j i denotes the j-th palm print sample to which perturbation noise is added based on the palm print main line.

[0220] In one alternative embodiment, a palm print data set is acquired.

[0221] Here, at least one palm print data is stored in the palm print data set, and the at least one palm print data is correspondingly marked with a data label.

[0222] Schematically, the palm print data stored in the palm print data set is palm print data obtained through legal authorization. Optionally, the data label marked corresponding to the palm print data is used to distinguish different palm print data. For example, palm print data 1 is a palm print corresponding to user 1, and user 1 is the data label of palm print data 1, or palm print data 2 is a palm print obtained from family palm print database 2, and family palm print database 2 is the data label of palm print data 1.

[0223] The above are merely schematic examples, and the embodiments of the present application are not limited thereto.

[0224] In one optional embodiment, a second training is performed on the candidate palmprint recognition model with the palmprint data and the data labels corresponding to the palmprint data to obtain a target palmprint recognition model.

[0225] Here, the target palm print recognition model is a model obtained by training the palm print recognition model.

[0226] Schematically, the generated palm print sample is used to improve model performance in a training stage of a palm print recognition model, i.e., a palm print sample or a target image corresponding to the palm print sample is used to perform a first training on the palm print recognition model to obtain a candidate palm print recognition model; certification The model can better learn the geometric information of the generated palm print samples.

[0227] In one alternative embodiment, the candidate palm print certificationConsidering the application of the model to real-world situations, candidate palm prints are generated using palm print data and corresponding data labels. certification A second training is performed on the model. Here, the second training is performed by using palm print data of a living organism to generate candidate palm prints. certification The model is used to improve the authentication model against real palm prints.

[0228] Schematically, as shown in FIG. 15, the training process of performing the first training and the second training on the palm print model is as follows.

[0229] In step 1510, a palm print sample is generated.

[0230] Optionally, the above-mentioned palm print sample generating method is used to obtain a palm print sample including palm print main lines and palm print fine lines, which can support the palm print recognition model to pay attention to the subtle changes between the patterns such as palm print main lines and palm print fine lines, and can facilitate the palm print recognition model to learn more discriminatory features.

[0231] In step 1520, the target image is bulk composited.

[0232] Schematically, after obtaining a palm print sample, the palm print sample is perturbed to obtain a target sample, and a sample image arbitrarily obtained from a sample image set is used as a background, and the target sample is embedded onto the sample image to obtain a target image.

[0233] In step 1530, a first training session is performed.

[0234] Optionally, a large number of synthesized target images are inputted into the palm print recognition model, and then a first training is performed on the palm print recognition model based on the target images.

[0235] Schematically, the mass-combined target images include a target images of identification A and b target images of identification B, and the a target images and the b target images are input into the palmprint recognition model, so that the palmprint recognition model learns the palmprint pattern information (palmprint main line information, palmprint thin line information) of the palmprint samples corresponding to different identifications in different target images. For example, the palmprint recognition model learns the similarity between the a target images of identification A and the similarity between the b target images of identification B. In addition, the palmprint recognition model further learns the difference between the a target images of identification A and the b target images of identification B, and determines the similarity of the palmprint pattern information of the palmprint samples corresponding to the same identification in different target images, and the difference of the palmprint pattern information of the palmprint samples corresponding to different identifications in different target images, etc. Optionally, a candidate palmprint recognition model is obtained based on the above first training.

[0236] In step 1540, a publicly available palm print data set is obtained.

[0237] Schematically, a palm print dataset storing a plurality of palm print data that has been made public based on a legal method is obtained, and the palm print dataset stores a plurality of palm print data marked with a data label.

[0238] In step 1550, a second training is performed.

[0239] Optionally, taking into consideration the possibility that there may be a certain difference between the palm print sample and the palm print data corresponding to a living organism, after a first training is performed on the palm print recognition model, a second training is performed on the candidate palm print recognition model using the palm print data corresponding to the living organism.

[0240] In one optional embodiment, palm print data is input into a candidate palm print recognition model, a loss value corresponding to the palm print data is determined according to the output data of the candidate palm print recognition model and the data label corresponding to the palm print data, the candidate palm print recognition model is trained with the loss value, and a target palm print recognition model is obtained according to the training of the candidate palm print recognition model reaching a training goal.

[0241] Schematically, a second training is performed on the candidate palmprint authentication model based on a plurality of palmprint data in the palmprint data set and data labels corresponding to the plurality of palmprint data. For example, a loss value is calculated using the output of the candidate palmprint authentication model for the palmprint data and the data labels corresponding to the palmprint data, and a second training is performed on the candidate palmprint authentication model based on the loss value calculation result, with the training goal being to reduce the loss value, thereby realizing a process of fine-tuning the candidate palmprint authentication model.

[0242] For example, palm print data is input to a candidate palm print recognition model, and the candidate palm print recognition model outputs a predicted label corresponding to the palm print data, and a loss value corresponding to the palm print data is determined based on the difference between the predicted label corresponding to the palm print data and the data label. It can be understood that the data label is a training label, which is an accurate label corresponding to the palm print data, and the predicted label is a label obtained by predicting after the model processes the input data. Training the model allows the model to learn the similarity between palm print data corresponding to the same data label in different palm print data, and the difference between palm print data corresponding to different data labels in different palm print data, so that the model finally outputs an accurate label or a label very close to the accurate label, and has palm print recognition ability.

[0243] Optionally, in the process of training the candidate palm print recognition model with the loss value, the training of the candidate palm print recognition model reaches a training goal to obtain a target palm print recognition model. Schematically, the training goal includes at least one of the following situations:

[0244] 1. According to the loss value reaching the convergence state, the candidate palmprint recognition model obtained by the closest one iteration of training is set as the target palmprint recognition model.

[0245] Schematically, the loss value reaching the convergence state indicates that the numerical value of the loss value obtained by the loss function does not change further or the change range is smaller than a predetermined threshold. For example, if the loss value corresponding to the nth palmprint data is 0.1 and the loss value corresponding to the n+1th palmprint data is also 0.1, it is considered that the loss value has reached the convergence state, and the candidate palmprint recognition model with the adjusted loss value corresponding to the nth palmprint data or the n+1th palmprint data is set as the target palmprint recognition model, thereby realizing the training process for the candidate palmprint recognition model.

[0246] 2. When the number of times the loss value is acquired reaches a threshold number, the candidate palm print recognition model obtained by the closest one iterative training is set as the target palm print recognition model.

[0247] Schematically, one loss value is obtained in one acquisition, and the number of acquisitions of the loss value for training the candidate palmprint authentication model is preset, and when one palmprint data corresponds to one loss value, the number of acquisitions of the loss value is the number of palmprint data, or when one palmprint data corresponds to multiple loss values, the number of acquisitions of the loss value is the number of loss values. For example, one loss value is obtained in one acquisition, and the threshold of the number of acquisitions of the loss value is preset to 10 times, when the threshold of the number of acquisitions is reached, the candidate palmprint authentication model adjusted to the closest one loss value is set as the target palmprint authentication model, or the candidate palmprint authentication model adjusted to the minimum loss value in the process of adjusting the loss value 10 times is set as the target palmprint authentication model, thereby realizing the training process for the candidate palmprint authentication model.

[0248] In one alternative embodiment, after obtaining the target palm print authentication model, the palm print is authenticated. Schematically, FIG. 16 is a flow diagram of taking a palm print.

[0249] First, take a palm print authentication photo 1610 and a palm print enrollment photo 1620, and detect the palm corresponding to the palm print based on the palm print authentication photo 1610 and the palm print enrollment photo 1620, and extract the palm print area of ​​interest 1630. Then, transmit the palm print area of ​​interest 1630 to a background 1640 (e.g., a server), and the background 1640 adds the palm print authentication photo 1610, the palm print enrollment photo 1620, and the palm print area of ​​interest 1630 to a palm print registry. Finally, the background 1640 authenticates the palm print based on the target palm print authentication model on its own side. Optionally, after authenticating the palm print, display the palm print authentication result on a front end (e.g., a terminal).

[0250] Schematically, the palm print authentication result may be displayed not only in the form of "yes" or "no", but also in the form of a numerical value such as probability. The above is merely a schematic example, and the embodiments of the present application are not limited thereto.

[0251] As shown in the table below, it includes the authentication data of the target palm print authentication model and the authentication data obtained by other palm print authentication techniques.

[0252] [Table 1]

[0253] Table 1 shows the recognition effect of the target recognition model and other methods in the palmprint recognition field on five public datasets. The five public datasets are Chinese sentiment vocabulary database (CASIA, Institute of Automation, Chinese Academy of Sciences), Indian Institute of Technology Delhi dataset (IITD, Indian Institute of Technology Delhi), Hong Kong Polytechnic University dataset (PolyU, Polytechnic University), Trinity College Dublin (TCD, Trinity College Dublin), and Maintenance Planning Document (MPD, Maintenance Planning Document). The evaluation metrics are Top-1 accuracy rate and Equal Error Rate (EER, Equal Error Rate), respectively. Communication Protocol (PalmNet) is a state-of-the-art method in the palmprint field and is used for comparison here.

[0254] Here, the higher the Top-1 index, the better the authentication effect, and the smaller the EER index, the better the authentication effect. Schematically, ArcFace authentication technology using the baseline method (ArcFace) is adopted, and its backbone network is the mobile phone network (MobileFaceNet). From the above table, we can see that the effect of the target palmprint authentication model on palmprint authentication is superior to other methods.

[0255] As described above, the curve obtained by sequentially connecting the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line is taken as the palm print main line, and at least one palm print sample including the palm print main line is obtained, and the palm print sample is used to train the palm print recognition model. Through the above method, a large number of palm print samples are simulated and obtained according to the distribution of the main lines in the palm print, so that the generated palm print samples have stronger diversity. Based on the palm print sample, the palm print recognition model is trained, which breaks through the localization of the palm print data set and improves the robustness of the palm print recognition model.

[0256] In the method according to the embodiment of the present application, a process of training a palmprint recognition model with at least one palmprint sample is described. First, a set of sample images is obtained, then, a palmprint sample is embedded on the sample image with a sample image in the set of sample images as a background, a target image is obtained, and a palmprint recognition model is trained with the target image to obtain a target palmprint recognition model. The above method can improve the palmprint recognition performance of the target palmprint recognition model in real situations, and using the target palmprint recognition model does not increase the amount of extra calculation or training burden, which is a simple and effective training optimization solution.

[0257] 17 is a structural block diagram of a palm print sample generating device according to an exemplary embodiment of the present application. As shown in FIG. 17, the device includes a positioning point generating module 1710 for generating positioning point data according to a distribution rule of palm print main lines, where the positioning point data includes first data corresponding to the positioning points of the first main line and second data corresponding to the positioning points of the second main line, and an adjustment point generating module 1720 for generating adjustment point data according to a radian degree rule of palm print main lines, where the adjustment points of the main lines corresponding to the adjustment point data are used to control the radian degree of the main lines composed of the positioning points of the first main line and the positioning points of the second main line. The palm print authentication module 1700 includes an adjustment point generation module 1720, a main line generation module 1730 for generating a palm print main line based on the first data, the second data, and the adjustment point data, where the palm print main line is a curve sequentially connecting a position point of the first main line, an adjustment point of the main line, and a position point of the second main line, and a sample generation module 1740 for generating at least one palm print sample including the palm print main line, where the palm print sample is used to train a palm print authentication model, and the palm print authentication model is used to perform palm print authentication.

[0258] In one optional embodiment, the positioning point generation module 1710 further determines a first area corresponding to the extension direction of the palm print main line and a second area according to the distribution rule of the palm print main line, determines first data corresponding to the positioning point of the first main line in the first area, and determines second data corresponding to the positioning point of the second main line in the second area.

[0259] In one optional embodiment, the positioning point generation module 1710 further determines a palmprint generation area, which determines a distribution range of the palmprint main lines according to the distribution rule of the palmprint main lines, determines a first vertex and a second vertex that are in a diagonal relationship within the palmprint generation area, determines a first area within the palmprint generation area with the first vertex as a center and a first predetermined length as a radius, and determines a second area within the palmprint generation area with the second vertex as a center and a second predetermined length as a radius.

[0260] In one optional embodiment, the positioning point generation module 1710 further determines, in a random selection manner, first data corresponding to a positioning point of a first main line in a first area, and determines, in a random selection manner, second data corresponding to a positioning point of a second main line in a second area.

[0261] In one optional embodiment, the adjustment point generation module 1720 further determines a third area based on the positional relationship between the positioning points of the first main line and the positioning points of the second main line according to the radian law of the palm print main line, and generates adjustment point data in the third area.

[0262] In one optional embodiment, the adjustment point generation module 1720 further connects the positioning point of the first main line and the positioning point of the second main line to obtain a target line segment, and sets a rectangular area with a specified side length as the third area, centered on the midpoint of the target line segment.

[0263] In one optional embodiment, the palm print sample further includes palm print fine lines.

[0264] The sample generating module 1740 further determines positioning points of at least two palm print thin lines, generates palm print thin lines based on the positioning points of the at least two palm print thin lines, and generates at least one palm print sample including palm print main lines and palm print thin lines.

[0265] In one optional embodiment, the sample generation module 1740 further connects the positioning points of at least two palm print thin lines to obtain a palm print thin line.

[0266] In one optional embodiment, the sample generation module 1740 further determines, based on the positioning points of the at least two palm print thin lines, an adjustment point of the palm print thin lines for controlling an arc between the positioning points of the at least two palm print thin lines, and determines, within a predetermined palm print thin line quantity range, a palm print thin line based on the positioning points of the at least two palm print thin lines and the adjustment point of the palm print thin lines.

[0267] In one optional embodiment, as shown in FIG. 18 , the apparatus further includes: an acquisition module 1750 for acquiring a sample image set in which at least one sample image is stored; an embedding module 1760 for embedding a palm print sample into the sample image against a background of the sample image to acquire a target image; and a training module 1770 for training a palm print recognition model with the target image.

[0268] In one optional embodiment, the embedding module 1760 further perturbs at least one palm print sample within the target perturbation interval to obtain a target sample, and embeds the target sample on the sample image with the sample image as a background to obtain a target image.

[0269] In one alternative embodiment, the target perturbation interval includes a palm print main line perturbation interval and a palm print fine line perturbation interval.

[0270] The embedding module 1760 further perturbs a palm print main line corresponding to the at least one palm print sample within a palm print main line perturbation interval to obtain a perturbed main line, perturbs a palm print thin line corresponding to the at least one palm print sample within a palm print thin line perturbation interval to obtain a perturbed thin line, and obtains a target sample based on the perturbed main line and the perturbed thin line.

[0271] In one optional embodiment, the embedding module 1760 further adds noise to the palm print main lines corresponding to the at least one palm print sample.

[0272] In one optional embodiment, the embedding module 1760 further adds noise to the palm print thin lines corresponding to the at least one palm print sample.

[0273] In one optional embodiment, the training module 1770 further performs a first training on the palm print authentication model with the target image to obtain a candidate palm print authentication model, obtains a palm print dataset in which at least one palm print data correspondingly marked with a data label is stored, and performs a second training on the candidate palm print authentication model with the palm print data and the data label corresponding to the palm print data, to obtain a target palm print authentication model which is a model obtained by training on the palm print authentication model.

[0274] In one optional embodiment, the training module 1770 further inputs palm print data into a candidate palm print recognition model, determines a loss value corresponding to the palm print data according to the output data of the candidate palm print recognition model and the data label corresponding to the palm print data, trains the candidate palm print recognition model with the loss value, and obtains a target palm print recognition model according to the training for the candidate palm print recognition model reaching a training goal.

[0275] As described above, the positioning points of the first main line and the positioning points of the second main line are generated according to the distribution rule of the palm print main line, the adjustment point data for controlling the radian degree of the main line is generated according to the radian degree rule of the palm print main line, the curve obtained by sequentially connecting the positioning points of the first main line, the adjustment point of the main line, and the positioning point of the second main line is defined as the palm print main line, at least one palm print sample including the palm print main line is generated, and the palm print authentication model is trained with the palm print sample. The above device simulates and obtains multiple palm print samples in the distribution state of the main lines in the palm print. The palm print samples are determined by generating data (the positioning points of the first main line, the positioning points of the second main line, and the adjustment point data), so that the generated palm print samples are large in number, and since there is no upper limit on the quantity, the generated palm print samples can have stronger diversity. When a palm print recognition model is trained based on the generated palm print samples, the palm print recognition model can mine pattern-specific rules and information that do not exist in larger palm print datasets, break through the limited nature of the palm print dataset, and improve the robustness of the palm print recognition model.

[0276] It should be noted that the palmprint sample generating device according to the above embodiment is only described by way of example according to the division of each functional module, but in actual use, the above functions may be distributed to different functional modules to be completed as necessary. That is, the internal structure of the device may be divided into different functional modules to complete all or part of the above functions. It should be noted that the palmprint sample generating device according to the above embodiment belongs to the same concept as the embodiment of the palmprint sample generating method, and the specific realization process thereof may be referred to the embodiment of the method, and the description thereof will be omitted here.

[0277] 19 shows a structural schematic diagram of a server according to one exemplary embodiment of the present application. The server 1900 includes a central processing unit (CPU) 1901, a system memory 1904 including a random access memory (RAM) 1902 and a read only memory (ROM) 1903, and a system bus 1905 connecting the system memory 1904 and the central processing unit 1901. The server 1900 further includes an operating system 1913 and mass storage 1906 for storing application programs 1914 and other program modules 1915.

[0278] Mass storage 1906 is connected to a mass memory controller (not shown) of system bus 1905, and thereby to central processing unit 1901. Mass storage 1906 and its associated computer-readable media provide non-volatile storage for server 1900. That is, mass storage 1906 may include a computer-readable medium (not shown), such as a hard disk, or a Compact Disc Read Only Memory (CD-ROM) drive.

[0279] Without loss of generality, computer readable media may include computer storage media and communication media. Computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information such as computer readable commands, data structures, program modules, or other data. Computer storage media includes RAM, ROM, Erasable Programmable Read Only Memory (EPROM), Electrically Erasable Programmable Read Only Memory (EEPROM), Flash memory or other solid-state drive technology, CD-ROM, Digital Versatile Disc (DVD) or other optical storage, cassettes, magnetic tape, disk storage or other magnetic storage devices. Of course, those skilled in the art will appreciate that computer storage media is not limited to the above several types. The above system memory 1904 and mass storage 1906 may be collectively referred to as memory.

[0280] Depending on the embodiment of the present application, the server 1900 may further be connected to a remote computer on a network, such as the Internet, and may operate in that manner, i.e., the server 1900 may be connected to a network 1912 by connecting to a network interface unit 1911 on the system bus 1905, or the server 1900 may use the network interface unit 1911 to connect to other types of networks or remote computer systems (not shown).

[0281] The memory further includes one or more programs, the one or more programs being stored in the memory and configured to be executed by the CPU.

[0282] An embodiment of the present application further provides a computer device, the computer device including a processor and a memory, the memory storing at least one command, at least one program, code set or command set, the at least one command, at least one program, code set or command set being loaded and executed by the processor to realize the method for generating a palm print sample according to each embodiment of the method described above.

[0283] An embodiment of the present application further provides a computer-readable storage medium, the computer-readable storage medium storing at least one command, at least one program, code set or command set, the at least one command, at least one program, code set or command set being loaded and executed by a processor to realize the method for generating a palm print sample according to each of the above method embodiments.

[0284] An embodiment of the present application further provides a computer program product or a computer program comprising computer instructions stored in a computer readable storage medium, the computer instructions being read by a processor of a computing device from the computer readable storage medium and executed by the processor to cause the computing device to perform the method for generating a palm print sample according to any of the above embodiments.

[0285] Alternatively, the computer-readable storage medium may include a read only memory (ROM), a random access memory (RAM), a solid state drive (SSD), or an optical disk. Here, the random access memory may include a resistance random access memory (ReRAM) and a dynamic random access memory (DRAM). The numbers of the above embodiments of the present application are for illustrative purposes only and do not represent the superiority or inferiority of the embodiments.

[0286] Those skilled in the art can understand that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by instructing related hardware by a program. The above programs may be stored in a computer-readable storage medium, and the above storage medium may be a read-only memory, a magnetic disk, an optical disk, or the like.

[0287] The above are only selective embodiments of the present application, and do not limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application are within the scope of protection of the present application.

Claims

1. 1. A method for generating palm print samples implemented by a computing device, comprising: generating positioning point data including first data corresponding to positioning points of the first main line and second data corresponding to positioning points of the second main line according to a distribution rule of the palm print main lines; generating adjustment point data according to a rule of radian degree of a palm print main line, and the adjustment point of the main line corresponding to the adjustment point data is used to control the radian degree of a main line formed by the positioning point of the first main line and the positioning point of the second main line; generating a palm print main line, which is a curve that sequentially connects the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, based on the first data, the second data, and the adjustment point data; generating at least one palm print sample including the palm print main lines, the palm print sample being used to train a palm print recognition model, and the palm print recognition model being used to perform palm print authentication.

13. A method for generating a palm print sample comprising:

2. The generating of positioning point data according to a distribution rule of the palm print main line includes: determining a first area and a second area corresponding to a direction in which the palm print main line extends according to a distribution rule of the palm print main line; determining the first data corresponding to a position point of the first main line within the first area; determining the second data corresponding to a position point of the second main line within the second area.

2. The method of claim 1, wherein the palm print sample is generated based on the palm print sample.

3. Determining a first area and a second area corresponding to a direction in which the palm print main line extends according to a distribution rule of the palm print main line; determining a palm print generating area that defines a distribution range of the palm print main lines according to a distribution rule of the palm print main lines; determining a first vertex and a second vertex that are diagonally related within the palm print generation area; determining the first area within the palm print generation area with the first vertex as a center and a first predetermined length as a radius; determining the second area within the palm print generation area with the second vertex as a center and a second predetermined length as a radius.

3. The method of claim 2, wherein the palm print sample is generated based on the palm print sample.

4. Determining the first data corresponding to positioning points of the first main line in the first area and determining the second data corresponding to positioning points of the second main line in the second area includes: determining the first data corresponding to the positioning point of the first main line in the first area by random selection; determining, within the second area, the second data corresponding to the positioning points of the second main line in a random selection manner; 3. The method of claim 2, wherein the palm print sample is generated based on the palm print sample.

5. The generation of the adjustment point data according to the radian law of the palm print main line is determining a third area based on a positional relationship between the positioning points of the first main line and the positioning points of the second main line according to the radian law of the palm print main line; generating the adjustment point data within the third area.

2. The method of claim 1, wherein the palm print sample is generated based on the palm print sample.

6. Determining a third area based on a positional relationship between a positioning point of the first main line and a positioning point of the second main line acquiring a target line segment by connecting the positioning point of the first main line and the positioning point of the second main line; and determining, as the third area, a rectangular area having a predetermined side length and centered on a midpoint of the target line segment.

6. The method of claim 5, wherein the palm print sample is generated based on the palm print sample.

7. The palm print sample further includes palm print fine lines, Generating at least one palm print sample including the palm print main lines includes: determining at least two palm print fine line positioning points; generating the palm print thin line based on the positioning points of the at least two palm print thin lines; generating at least one palm print sample including the palm print main lines and the palm print fine lines within a predetermined palm print quantity range, the palm print quantity range including at least one of a palm print main line quantity range and a palm print fine line quantity range; The method for generating a palm print sample according to any one of claims 1 to 6.

8. generating the palm print thin line based on the positioning points of the at least two palm print thin lines, and connecting the positioning points of the at least two palm print thin lines to obtain a palm print thin line.

8. The method of claim 7, wherein the palm print sample is generated based on the palm print sample.

9. generating the palm print thin line based on the positioning points of the at least two palm print thin lines, determining an adjustment point of the palm print thin line for controlling an arc between the positioning points of the at least two palm print thin lines based on the positioning points of the at least two palm print thin lines; and determining a palm print thin line based on the positioning points of the at least two palm print thin lines and the adjustment point of the palm print thin line within a predetermined quantity range of the palm print thin lines.

8. The method of claim 7, wherein the palm print sample is generated based on the palm print sample.

10. After generating at least one palm print sample including the palm print main lines, acquiring a sample image set having at least one sample image stored therein; Incorporating the palm print sample onto the sample image as a background of the sample image to obtain a target image; training the palm print recognition model with the target image. The method for generating a palm print sample according to any one of claims 1 to 6.

11. Incorporating the palm print sample onto the sample image as a background of the sample image to obtain a target image includes: Perturbing the at least one palm print sample within a target perturbation interval to obtain a target sample; and incorporating the target sample onto the sample image against the sample image to obtain a target image.

11. The method of claim 10, wherein the palm print sample is generated based on the palm print sample.

12. the target perturbation section includes a palm print main line perturbation section and a palm print thin line perturbation section; Perturbing the at least one palm print sample within the target perturbation interval to obtain a target sample includes: Perturbing a palm print main line corresponding to the at least one palm print sample within the palm print main line perturbation section to obtain a perturbed main line; Perturbing a palm print thin line corresponding to the at least one palm print sample within the palm print thin line perturbation section to obtain a perturbed thin line; and obtaining the target sample based on the perturbation main line and the perturbation thin line. The method of claim 11, wherein the palm print sample is generated based on the palm print sample.

13. Perturbing the at least one palm print sample comprises: adding noise to the palm print main lines corresponding to the at least one palm print sample. The method of claim 11, wherein the palm print sample is generated based on the palm print sample.

14. Perturbing the at least one palm print sample comprises: adding noise to the palm print fine lines corresponding to the at least one palm print sample. The method of claim 11, wherein the palm print sample is generated based on the palm print sample.

15. Training the palm print recognition model with the target image includes: A first training is performed on the palm print recognition model with the target image to obtain a candidate palm print recognition model; obtaining a palm print data set, in which at least one palm print data correspondingly marked with a data label is stored; and performing a second training on the candidate palm print authentication model using the palm print data and a data label corresponding to the palm print data to obtain a target palm print authentication model, the target palm print authentication model being a model obtained by training the palm print authentication model.

11. The method of claim 10, wherein the palm print sample is generated based on the palm print sample.

16. performing a second training on the candidate palm print recognition model with the palm print data and a data label corresponding to the palm print data to obtain a target palm print recognition model; inputting the palm print data into the candidate palm print authentication model, and determining a loss value corresponding to the palm print data according to output data of the candidate palm print authentication model and a data label corresponding to the palm print data; training the candidate palm print recognition model with the loss value; and obtaining the target palm print recognition model in response to the training of the candidate palm print recognition model reaching a training goal.

16. The method of claim 15, wherein the palm print sample is generated based on the palm print sample.

17. a positioning point generating module for generating positioning point data including first data corresponding to the positioning points of the first main line and second data corresponding to the positioning points of the second main line according to a distribution rule of the palm print main lines; an adjustment point generating module for generating adjustment point data according to a rule of radian degree of a palm print main line, the adjustment point of the main line corresponding to the adjustment point data being used to control the radian degree of a main line formed by a positioning point of the first main line and a positioning point of the second main line; a main line generating module that generates a palm print main line, which is a curve that sequentially connects the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, based on the first data, the second data, and the adjustment point data; a sample generation module for generating at least one palm print sample including the palm print main lines, the palm print sample being used to train a palm print recognition model, and the palm print recognition model being used to perform palm print recognition.

13. A palm print sample generating device comprising:

18. a processor and a memory, the memory storing at least one command, at least one program, code set, or command set, the at least one command, the at least one program, code set, or command set being loaded and executed by the processor, generating positioning point data including first data corresponding to positioning points of a first main line and second data corresponding to positioning points of a second main line according to a distribution rule of palm print main lines; generating adjustment point data according to a palm print main line radian degree rule, the adjustment point of the main line corresponding to the adjustment point data being used to control the radian degree of the main line formed by the positioning point of the first main line and the positioning point of the second main line; generating a palm print main line, which is a curve that sequentially connects the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, based on the first data, the second data, and the adjustment point data; generating at least one palm print sample including the palm print main lines, the palm print sample being used to train a palm print recognition model, and the palm print recognition model being used to perform palm print recognition; A computer device implementing the method for generating palm print samples according to any one of claims 1 to 6.

19. The computer program or command is executed by a processor to generating positioning point data including first data corresponding to positioning points of a first main line and second data corresponding to positioning points of a second main line according to a distribution rule of palm print main lines; generating adjustment point data according to a palm print main line radian degree rule, the adjustment point of the main line corresponding to the adjustment point data being used to control the radian degree of the main line formed by the positioning point of the first main line and the positioning point of the second main line; generating a palm print main line, which is a curve that sequentially connects the positioning points of the first main line, the adjustment points of the main line, and the positioning points of the second main line, based on the first data, the second data, and the adjustment point data; generating at least one palm print sample including the palm print main lines, the palm print sample being used to train a palm print recognition model, and the palm print recognition model being used to perform palm print recognition; A computer program for implementing the method for generating a palm print sample according to any one of claims 1 to 6.

Images