Biometric methods and systems

Abstract

The biometric identification method and system provided in this specification acquire a cloud-side feature set stored on the cloud, which corresponds to a local feature set stored locally. The target biometric feature is then compared with features in a preset feature set to obtain feature comparison information corresponding to the target biometric feature. This preset feature set includes both the local feature set and the cloud-side feature set. Based on the feature comparison information, the local feature set, and the cloud-side feature set, the target object is identified. This allows the cloud-side feature set to supplement the deficiencies of the local feature set. Furthermore, by distributing cloud-side data to support edge-side decision-making, the accuracy of edge-side decision-making is further improved. Therefore, the accuracy of biometric identification can be enhanced.

Description

Technical Field

[0001] This specification relates to the field of biometrics, and more particularly to a biometric method and system. Background Technology

[0002] In recent years, with the rapid development of internet technology, the application of biometrics has become increasingly widespread. For example, in scenarios such as payment or security, it is often necessary to identify individuals through facial recognition or other biometric features. For these scenarios, existing biometric methods typically involve capturing biological images on the terminal and then performing biometric identification directly at the device.

[0003] In the process of researching and practicing existing technologies, the inventors of this invention discovered that the limited local computing power and storage space on the device side reduce the accuracy of biometric identification, thus leading to a decrease in the accuracy rate of biometric identification. Summary of the Invention

[0004] This specification provides a more accurate biometric identification method and system.

[0005] In a first aspect, this specification provides a biometric identification method, comprising: acquiring a cloud-side feature set stored on the cloud side, the cloud-side feature set corresponding to a local feature set stored locally; acquiring a target biometric image of a target object, and determining the target biometric features of the target object based on the target biometric image; performing feature comparison between the target biometric features and features in a preset feature set to obtain feature comparison information corresponding to the target biometric features, the preset feature set including the local feature set and the cloud-side feature set; and identifying the target object based on the feature comparison information, the local feature set, and the cloud-side feature set.

[0006] In some embodiments, the step of comparing the target biometric feature with features in a preset feature set to obtain feature comparison information corresponding to the target biometric feature includes: filtering biometric features corresponding to local candidate objects in the local feature set to obtain a local candidate biometric feature set; filtering biometric features corresponding to cloud candidate objects in the cloud-side feature set to obtain a cloud-side associated biometric feature set, wherein the cloud-side candidate objects are associated with the local candidate objects; and comparing the target biometric feature with biometric features in the target biometric feature set to obtain feature comparison information corresponding to the target biometric feature, wherein the target biometric feature set includes the local candidate biometric feature set and the cloud-side associated biometric feature set.

[0007] In some embodiments, the association includes at least one of appearance similarity, blood relationship, kinship, social relationship, and pre-defined rights and responsibilities.

[0008] In some embodiments, the step of filtering out the biometric features corresponding to cloud-side candidate objects from the cloud-side feature set to obtain a cloud-side associated biometric feature set includes: obtaining associated object information; determining the cloud-side candidate object corresponding to the local candidate object based on the associated object information; and filtering out the biometric features corresponding to the cloud-side candidate object from the cloud-side feature set to obtain a cloud-side associated biometric feature set.

[0009] In some embodiments, determining the cloud-side candidate object corresponding to the local candidate object based on the associated object information includes: extracting at least one pair of associated objects from the associated object information, wherein the pair of associated objects includes any two associated objects that have an association relationship; and filtering the associated objects corresponding to the local candidate object from the at least one pair of associated objects to obtain the cloud-side candidate object.

[0010] In some embodiments, comparing the target biometric feature with biometric features in a target biometric feature set to obtain feature comparison information corresponding to the target biometric feature includes: comparing the target biometric feature with biometric features in a local candidate biometric feature set to obtain local comparison information; comparing the target biometric feature with biometric features in a cloud-side associated biometric feature set to obtain cloud-side comparison information; and fusing the local comparison information and the cloud-side comparison information to obtain feature comparison information corresponding to the target biometric feature.

[0011] In some embodiments, identifying the target object based on the feature comparison information, the local feature set, and the cloud-side feature set includes: determining local target features based on the feature comparison information and the local feature set; and identifying the target object based on the local target features and the cloud-side feature set.

[0012] In some embodiments, determining the local target feature based on the feature comparison information and the local feature set includes: extracting the local current feature from the local feature set; and fusing the local current feature with the feature comparison information to obtain the local target feature.

[0013] In some embodiments, identifying the target object based on the local target features and the cloud-side feature set includes: extracting historical behavioral features of the local candidate object from the cloud-side feature set; fusing the historical behavioral features with the local target features to obtain the edge-side decision features corresponding to the target object; and determining the biometric result of the target object based on the edge-side decision features.

[0014] In some embodiments, obtaining the cloud-side feature set stored on the cloud side includes: sending a feature acquisition request to a cloud server, the feature acquisition request including the identification information of local candidate objects; and receiving the associated object information generated based on the identification information and the cloud-side feature set corresponding to the local feature set returned by the cloud server.

[0015] Secondly, this specification provides a biometric identification method applied to a remote server communicating with a local device, comprising: receiving a feature acquisition request from the local device, the feature acquisition request including identification information of a local candidate object, the local device storing biometric information of the local candidate object and the identification information; based on the identification information, filtering a cloud-side associated biometric set from a preset biometric set, the cloud-side associated biometric set including biometrics of objects associated with the local candidate object; and sending the cloud-side associated biometric set as a cloud-side feature set to the local device, for the local device to execute the biometric identification method described in the first aspect of this specification, and identify the target object based on the cloud-side feature set.

[0016] In some embodiments, the step of filtering out a cloud-side associated biometric set from a preset biometric set based on the identification information includes: determining associated object information based on the preset biometric set; determining a cloud-side candidate object corresponding to the local candidate object based on the identification information and the associated object information; and filtering out the biometrics corresponding to the cloud-side candidate object from the preset biometric set to obtain the cloud-side associated biometric set.

[0017] In some embodiments, determining the associated object information based on the preset biometric set includes: obtaining the feature similarity between biometric features in the preset biometric set; selecting at least one biometric pair from the preset biometric set, wherein the biometric pair is any two biometric features whose similarity exceeds a preset similarity threshold; and determining the associated object information based on the at least one biometric pair.

[0018] In some embodiments, determining the associated object information based on the at least one biometric pair includes: obtaining the target object identifier of at least one candidate object corresponding to the at least one biometric pair; determining at least one associated object pair based on the target object identifier, wherein the at least one associated object pair corresponds to the at least one biometric pair; and fusing the at least one associated object pair to obtain the associated object information.

[0019] In some embodiments, determining the cloud-side candidate object corresponding to the local candidate object based on the identification information and the associated object information includes: obtaining a set of object identifier pairs corresponding to the associated object information; comparing the object identifier in the identification information with the object identifier in the set of object identifier pairs to obtain the associated object identifier; and determining the cloud-side candidate object corresponding to the local candidate object based on the associated object identifier.

[0020] In some embodiments, comparing the object identifier in the identification information with the object identifiers in the object identifier pair set to obtain the associated object identifier includes: extracting the local object identifier of the local candidate object from the identification information; filtering at least one object identifier corresponding to the local object identifier from the object identifier pair set to obtain a candidate associated object identifier set; and deduplicating the object identifiers in the candidate associated object identifier set based on the local object identifier to obtain the associated object identifier.

[0021] In some embodiments, after receiving the feature acquisition request from the local device, the method further includes: generating historical behavioral features of the local candidate object based on the identification information; and sending the cloud-side associated biometric set as a cloud-side feature set to the local device, which includes: sending the historical behavioral features and the cloud-side associated biometric set as the cloud-side feature set to the local device.

[0022] Thirdly, this specification also provides a biometric system, comprising: at least one storage medium storing at least one instruction set for performing biometric identification; and at least one processor communicatively connected to the at least one storage medium, wherein, when the biometric system is running, the at least one processor reads the at least one instruction set and executes the biometric identification methods described in the first and second aspects of this specification according to the instructions of the at least one instruction set.

[0023] As can be seen from the above technical solutions, the biometric identification method and system provided in this specification acquire a cloud-side feature set stored on the cloud side, which corresponds to a local feature set stored locally. Then, a target biometric image of the target object is acquired, and based on the target biometric image, the target biometric features of the target object are determined. Then, the target biometric features are compared with features in a preset feature set to obtain feature comparison information corresponding to the target biometric features. The preset feature set includes the local feature set and the cloud-side feature set. Based on the feature comparison information, the local feature set, and the cloud-side feature set, the target object is identified. Since this solution can acquire the cloud-side feature set from the cloud side, and the cloud-side feature set corresponds to the local feature set, the cloud-side feature set can supplement the deficiencies of the local feature set. In addition, by sending cloud-side data to support end-side decision-making, the accuracy of end-side decision-making is further improved. Therefore, the accuracy of biometric identification can be improved.

[0024] Other functionalities of the biometric methods and systems provided in this specification will be partially listed in the following description. The figures and examples described below will be readily apparent to those skilled in the art. The inventive aspects of the biometric methods and systems provided in this specification can be fully understood through practice or use of the methods, apparatus, and combinations described in the detailed examples below. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 A schematic diagram illustrating an application scenario of a biometric system provided according to an embodiment of this specification is shown.

[0027] Figure 2 A hardware structure diagram of a computing device provided according to an embodiment of this specification is shown;

[0028] Figure 3 A flowchart of a biometric identification method according to an embodiment of this specification is shown;

[0029] Figure 4 A flowchart of another biometric method provided according to embodiments of this specification is shown; and

[0030] Figure 5 A schematic flowchart of a face recognition method in a face recognition scenario provided by an embodiment of this specification is shown. Detailed Implementation

[0031] The following description provides specific application scenarios and requirements for this specification, intended to enable those skilled in the art to make and use the contents of this specification. Various partial modifications to the disclosed embodiments will be apparent to those skilled in the art, and the general principles defined herein can be applied to other embodiments and applications without departing from the spirit and scope of this specification. Therefore, this specification is not limited to the embodiments shown, but rather to the widest scope consistent with the claims.

[0032] The terminology used herein is for the purpose of describing particular exemplary embodiments only and is not restrictive. For example, unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “the” used herein may also include the plural forms. When used in this specification, the terms “comprising,” “including,” and / or “containing” mean that the associated integers, steps, operations, elements, and / or components are present, but do not exclude the presence of one or more other features, integers, steps, operations, elements, components, and / or groups, or that other features, integers, steps, operations, elements, components, and / or groups may be added to the system / method.

[0033] Considering the following description, these and other features of this specification, as well as the operation and function of the related components of the structure, and the economy of assembly and manufacture of the parts, can be significantly improved. All of these form part of this specification with reference to the accompanying drawings. However, it should be clearly understood that the drawings are for illustrative and descriptive purposes only and are not intended to limit the scope of this specification. It should also be understood that the drawings are not drawn to scale.

[0034] The flowcharts used in this specification illustrate operations implemented according to some embodiments of this specification. It should be clearly understood that the operations in the flowcharts may not be implemented in a sequential order. Instead, the operations may be implemented in reverse order or simultaneously. Furthermore, one or more additional operations may be added to the flowcharts. One or more operations may be removed from the flowcharts.

[0035] Before describing the specific embodiments in this specification, the application scenarios of this specification will be introduced as follows:

[0036] In any scenario requiring facial recognition, such as facial recognition for access control, payment, or unlocking, the local device captures and extracts facial features, then searches the local facial feature database. The retrieved features correspond to the user ID, which becomes the final recognition decision. However, due to limitations in local computing and storage space, the performance of the local facial feature extraction algorithm is significantly lower than that of the cloud-based algorithm. The overall accuracy of purely local retrieval and decision-making is also low. Introducing cloud-based data features can compensate for the limited and missing features of the local device data, further enhancing the capabilities of the decision model and thus improving the accuracy of facial recognition.

[0037] For ease of description, the terms that will appear in the following descriptions will be explained as follows:

[0038] Decision-making: In the process of biometric identification of a target object, the step of determining the final object identifier or identity based on the search results of the collected biometric features in the biometric database is called decision-making. Taking facial recognition as an example, in the offline facial recognition process, this refers to the step of determining the final person ID based on the search results of facial features in the facial recognition database.

[0039] End-side: The local capabilities of biometric devices (equipment). In the case of local identification, the local device (equipment) completes the acquisition of biological images of the target object, extraction of biometric features, feature comparison (feature retrieval), and identification decision.

[0040] Cloud-side: Biometric capabilities in the cloud. In end-to-end identification scenarios, after local devices (equipment) acquire biological images of the target object locally, biometric feature extraction, feature comparison (feature retrieval), and identification decisions are completed in the cloud.

[0041] Edge-side feature library: A local database of biometric features for candidate objects stored on the edge. Due to limitations in actual memory space on the edge, edge-side feature libraries are typically small and have limited capacity.

[0042] Cloud-side feature library: A full biometric database stored on the cloud side. The cloud-side feature library can contain the biometric features of all candidate objects in the storage service.

[0043] It should be noted that the above-described face recognition scenario is just one of the many application scenarios provided in this specification. The biometric methods and systems described in this specification can be applied not only to face recognition scenarios but also to all biometric scenarios, such as human body recognition, animal recognition, or recognition of other organisms, etc. Those skilled in the art should understand that the application of the biometric methods and systems described in this specification to other application scenarios is also within the scope of protection of this specification.

[0044] Figure 1 This diagram illustrates an application scenario of a biometric system 001 provided according to an embodiment of this specification. The biometric system 001 (hereinafter referred to as System 001) can be applied to biometric recognition in any scenario, such as face recognition, body recognition, animal recognition, other biometric recognition, etc. Figure 1 As shown, system 001 may include target object 100, client (end-side) 200, server (cloud-side) 300 and network 400.

[0045] The target object 100 can be an object waiting to undergo biometric identification or an object that is currently undergoing biometric identification. Taking a face recognition scenario as an example, the target object can be a user logging in for face recognition or a user currently undergoing face recognition. The target object 100 can undergo biometric identification on the client 200.

[0046] Client 200 can be a device for acquiring biometric images of target object 100. In some embodiments, the biometric method can be executed on client 200. In this case, client 200 may store data or instructions for executing the biometric method described herein, and may execute or be used to execute said data or instructions. In some embodiments, client 200 may include a hardware device with data processing capabilities and the necessary programs required to drive the hardware device. Figure 1As shown, client 200 can communicate with server 300. In some embodiments, server 300 can communicate with multiple clients 200. In some embodiments, client 200 can interact with server 300 through network 400 to receive or send messages, such as receiving or sending target biological images or target biometric features. In some embodiments, client 200 may include mobile devices, tablets, laptops, built-in devices in motor vehicles, or similar content, or any combination thereof. In some embodiments, the mobile device may include smart home devices, smart mobile devices, virtual reality devices, augmented reality devices, or similar devices, or any combination thereof. In some embodiments, the smart home device may include smart TVs, desktop computers, etc., or any combination thereof. In some embodiments, the smart mobile device may include smartphones, personal digital assistants, gaming devices, navigation devices, etc., or any combination thereof. In some embodiments, the virtual reality device or augmented reality device may include virtual reality headsets, virtual reality glasses, virtual reality patches, augmented reality headsets, augmented reality glasses, augmented reality patches, or similar content, or any combination thereof. For example, the virtual reality device or the augmented reality device may include Google Glass, head-mounted displays, VR, etc. In some embodiments, the built-in device in the motor vehicle may include an in-vehicle computer, an in-vehicle TV, etc. In some embodiments, the client 200 may include an image acquisition device for acquiring biological images of the target object 100, thereby obtaining a target biological image. In some embodiments, the image acquisition device may be a two-dimensional image acquisition device (such as an RGB camera), or a combination of a two-dimensional image acquisition device (such as an RGB camera) and a depth image acquisition device (such as a 3D structured light camera, a laser detector, etc.). In some embodiments, the client 200 may be a device with positioning technology for locating the position of the client 200.

[0047] In some embodiments, the client 200 may have one or more applications (APPs) installed. The APPs provide the target object 100 with the ability and interface to interact with the outside world via the network 400. The APPs include, but are not limited to: web browser APPs, search APPs, chat APPs, shopping APPs, video APPs, financial management APPs, instant messaging tools, email clients, social media platform software, etc. In some embodiments, the client 200 may have a target APP installed. The target APP can collect video or image information within a target space for the client 200, thereby obtaining a target image. In some embodiments, the target object 100 can also trigger a biometric request through the target APP. The target APP can respond to the biometric request and execute the biometric method described in this specification. The biometric method will be described in detail later.

[0048] Server 300 can be a server providing various services, such as a backend server supporting target biometric images acquired on client 200, or a cloud server. In some embodiments, the biometric identification method can be executed on server 300. In this case, server 300 can store data or instructions for executing the biometric identification method described herein, and can execute or be used to execute said data or instructions. In some embodiments, server 300 may include hardware devices with data processing capabilities and the necessary programs to drive the hardware devices. Server 300 can communicate with multiple clients 200 and receive data sent by clients 200.

[0049] Network 400 serves as a medium to provide a communication connection between client 200 and server 300. Network 400 facilitates the exchange of information or data. For example... Figure 1As shown, client 200 and server 300 can connect to network 400 and transmit information or data to each other through network 400. In some embodiments, network 400 can be any type of wired or wireless network, or a combination thereof. For example, network 400 may include cable networks, wired networks, fiber optic networks, telecommunications networks, intranets, the Internet, local area networks (LANs), wide area networks (WANs), wireless local area networks (WLANs), metropolitan area networks (MANs), public switched telephone networks (PSTNs), Bluetooth networks, ZigBee networks, near field communication (NFC) networks, or similar networks. In some embodiments, network 400 may include one or more network access points. For example, network 400 may include wired or wireless network access points, such as base stations or Internet switching points, through which one or more components of client 200 and server 300 can connect to network 400 to exchange data or information.

[0050] It should be understood that Figure 1 The number of clients 200, servers 300, and networks 400 shown is merely illustrative. Depending on implementation needs, there can be any number of clients 200, servers 300, and networks 400.

[0051] It should be noted that the biometric identification method can be executed entirely on the client 200, entirely on the server 300, or partially on the client 200 and partially on the server 300.

[0052] Figure 2 A hardware structure diagram of a computing device 600 provided according to an embodiment of this specification is shown. The computing device 600 can execute the biometric identification methods described in this specification. The biometric identification methods are described in other parts of this specification. When the biometric identification methods are executed on a client 200, the computing device 600 can be the client 200. When the biometric identification methods are executed on a server 300, the computing device 600 can be the server 300. When the biometric identification methods can be executed partly on the client 200 and partly on the server 300, the computing device 600 can be both the client 200 and the server 300.

[0053] like Figure 2 As shown, the computing device 600 may include at least one storage medium 630 and at least one processor 620. In some embodiments, the computing device 600 may also include a communication port 650 and an internal communication bus 610. Additionally, the computing device 600 may include I / O components 660.

[0054] The internal communication bus 610 can connect different system components, including storage medium 630, processor 620 and communication port 650.

[0055] I / O component 660 supports input / output between computing device 600 and other components.

[0056] Communication port 650 is used for data communication between computing device 600 and external sources. For example, communication port 650 can be used for data communication between computing device 600 and network 400. Communication port 650 can be a wired communication port or a wireless communication port.

[0057] Storage medium 630 may include a data storage device. The data storage device may be a non-transitory storage medium or a temporary storage medium. For example, the data storage device may include one or more of a disk 632, a read-only storage medium (ROM) 634, or a random access storage medium (RAM) 636. Storage medium 630 also includes at least one instruction set stored in the data storage device. The instructions are computer program code, which may include programs, routines, objects, components, data structures, procedures, modules, etc., that perform the biometric methods provided in this specification.

[0058] At least one processor 620 can be communicatively connected to at least one storage medium 630 and a communication port 650 via an internal communication bus 610. The at least one processor 620 is used to execute the at least one instruction set described above. When the computing device 600 is running, the at least one processor 620 reads the at least one instruction set and, according to the instructions of the at least one instruction set, executes the biometric method provided in this specification. The processor 620 can execute all the steps included in the biometric method. The processor 620 can be in the form of one or more processors. In some embodiments, the processor 620 may include one or more hardware processors, such as a microcontroller, microprocessor, reduced instruction set computer (RISC), application-specific integrated circuit (ASIC), application-specific instruction set processor (ASIP), central processing unit (CPU), graphics processing unit (GPU), physical processing unit (PPU), microcontroller unit, digital signal processor (DSP), field-programmable gate array (FPGA), advanced RISC machine (ARM), programmable logic device (PLD), any circuit or processor capable of performing one or more functions, or any combination thereof. For illustrative purposes only, only one processor 620 is described in this specification for the computing device 600. However, it should be noted that the computing device 600 in this specification may also include multiple processors. Therefore, the operation and / or method steps disclosed in this specification may be executed by one processor as described in this specification, or they may be executed jointly by multiple processors. For example, if the processor 620 of the computing device 600 in this specification executes steps A and B, it should be understood that steps A and B may also be executed jointly or separately by two different processors 620 (e.g., the first processor executes step A, the second processor executes step B, or the first and second processors jointly execute steps A and B).

[0059] Figure 3 A flowchart of a biometric method P100 according to an embodiment of this specification is shown. As previously described, computing device 600 can execute the biometric method P100 of this specification. Specifically, processor 620 can read an instruction set stored in its local storage medium and then execute the biometric method P100 of this specification according to the instructions in the instruction set. Figure 3 As shown, method P100 may include:

[0060] S110: Obtain the set of cloud-side features stored on the cloud side.

[0061] The cloud-side feature set corresponds to the local feature set stored locally. The cloud-side feature set can be a set of features stored in the cloud for biometric identification. The cloud side can be understood as a remote server relative to the local machine, also referred to as a cloud server.

[0062] There are several ways to obtain the cloud-side feature set that exists on the cloud side, including the following:

[0063] For example, the processor 620 can directly obtain cloud-side feature sets from the cloud side, or it can indirectly obtain cloud-side feature sets from the cloud side, as follows:

[0064] (1) The processor 620 directly obtains the cloud-side feature set from the cloud side.

[0065] For example, the processor 620 can send a feature acquisition request to the cloud server (server 300), which includes the identification information of local candidate objects, and receive the associated object information generated based on the identification information and the cloud-side feature set corresponding to the local feature set returned by the cloud server (server 300).

[0066] Upon receiving a feature acquisition request, the cloud server (300) can generate a cloud-side feature set on the cloud side and send the cloud-side feature set to the local device, enabling the local device to perform biometric identification. Therefore, another biometric identification method can be provided for the cloud side. Figure 4 A flowchart of another biometric method P200 provided in an embodiment of this specification is presented. As previously described, computing device 600 can execute the biometric method P200 of this specification. Specifically, processor 620 can read an instruction set stored in its local storage medium and then execute the biometric method P200 of this specification according to the provisions of the instruction set. Figure 4 As shown, method P200 may include:

[0067] S210: Receive a feature acquisition request from the local device.

[0068] The feature acquisition request includes the identification information of local candidate objects. Local candidate objects can be candidate objects corresponding to biological (object) identifiers stored locally for biometric identification. The identification information can be object identifiers or identity identifiers indicating the local candidate objects.

[0069] The local device can be a device corresponding to the cloud side. This local device can be the client 200, or other local devices used for biometric identification.

[0070] There are several ways to receive feature acquisition requests from local devices, including the following:

[0071] For example, the processor 620 can directly receive feature acquisition requests sent by the local device, or it can indirectly receive feature acquisition requests sent by the local device.

[0072] There are several ways to indirectly receive feature acquisition requests sent by local devices. For example, the processor 620 can receive biometric requests sent by third-party devices and extract feature acquisition requests from the biometric requests. Alternatively, when there are many identification information of local candidate objects or a large amount of memory, the processor 620 can also receive storage addresses sent by third-party devices and obtain feature acquisition requests sent by local devices based on the storage addresses.

[0073] S220: Based on the identification information, select the cloud-side associated biometric set from the preset biometric set.

[0074] The preset biometric set can be a cloud-based feature library, used to store the biometrics of all candidate objects on the cloud. Biometrics can be the characteristic information that characterizes the object from a biological perspective. There are many types of biometrics, such as at least one of facial features, limb features, blood features, body shape features, hair features, voice features, odor features, and organ features.

[0075] The cloud-side associated biometric set is a collection of biometric features of candidate objects stored in the cloud that are associated with local candidate objects. The term "association" indicates a specific relationship between candidate objects. This association can be of various types, including at least one of the following: facial similarity, blood relation, kinship, pre-defined rights and responsibilities, social relationship, etc.

[0076] There are several ways to select cloud-side associated biometric sets from a preset set of biometric features based on identification information, as follows:

[0077] For example, the processor 620 determines the associated object information based on the preset biometric set, determines the cloud-side candidate object corresponding to the local candidate object based on the identification information and the associated object information, and filters out the biometrics corresponding to the cloud-side candidate object from the preset biometric set to obtain the cloud-side associated biometric set.

[0078] The associated object information can be information indicating the association relationship between candidate objects corresponding to biometric features in a preset biometric set. The associated object information may include at least one pair of candidate objects that are associated, or the identification information of that pair. There are multiple ways to determine the associated object information based on the preset biometric set. For example, the processor 620 can obtain the feature similarity between biometric features in the preset biometric set, filter out at least one pair of biometric features in the preset biometric set (where the pair is any two biometric features with a similarity exceeding a similarity threshold), and determine the associated object information based on at least one pair of biometric features.

[0079] There are several ways to obtain the feature similarity between biometric features in the preset biometric feature set. For example, the processor 620 can compare all or part of the biometric features in the preset biometric feature set, and then calculate the similarity between any two biometric features to obtain the feature similarity between biometric features in the preset biometric feature set.

[0080] After obtaining the feature similarity between biometrics in a preset biometric set, at least one biometric pair can be selected from the preset biometric set. This biometric pair consists of any two biometrics whose similarity threshold exceeds a preset similarity threshold. There are various methods for selecting at least one biometric pair. For example, the processor 620 can select feature similarities exceeding the preset similarity threshold from the specific similarity values ​​to obtain at least one target feature similarity. Then, it can select the biometric corresponding to at least one target feature similarity from the preset biometric set to obtain at least one biometric pair.

[0081] After selecting at least one pair of biometric features, the associated object information can be determined based on that pair. There are several ways to determine the associated object information. For example, the processor 620 can obtain the target object identifier of at least one candidate object corresponding to at least one pair of biometric features, determine at least one pair of associated objects based on the target object identifier, and determine that the at least one pair of associated objects corresponds to one of its own biometric features. Additionally, the processor can fuse the at least one pair of associated objects to obtain the associated object information.

[0082] Here, an associated object pair can be understood as a pair of candidate objects that have a relationship. There are multiple ways to determine at least one associated object pair based on the target object identifier. For example, the processor 620 can filter out the object identifier pairs corresponding to each biometric pair from the target object identifiers, and consider the two candidate objects corresponding to each object identifier pair as an associated object pair. In this case, it can be observed that the associated objects have a one-to-one correspondence with the biometric pairs.

[0083] After determining the associated object information, the cloud-side candidate object corresponding to the local candidate object can be determined based on the identification information and the associated object information. There are multiple ways to determine the cloud-side candidate object. For example, the processor 620 can obtain the set of object identifier pairs corresponding to the associated object information, compare the object identifier in the identification information with the object identifier in the set of object identifier pairs to obtain the associated object identifier, and determine the cloud-side candidate object corresponding to the local candidate object based on the associated object identifier.

[0084] The object identifier pair set can be a set of object identifier pairs corresponding to each associated object pair in the associated object information. There are multiple ways to compare the object identifiers in the identifier information with the object identifiers in the object identifier pair set. For example, the processor 620 can extract the local object identifier of the local candidate object from the identifier information, filter at least one object identifier corresponding to the local object identifier in the object identifier set to obtain the candidate associated object identifier set, and based on the local object identifier, remove duplicate object identifiers from the candidate associated object identifier set to obtain the associated object identifier.

[0085] The associated object information can also be understood as an end-side comparison instruction. This end-side comparison instruction takes the form of associated object pairs, which are the associated object pairs that exist between all candidate objects in the cloud storage. The candidate associated object identifier set can be understood as the set of object identifiers of cloud-side candidate objects that are associated with local candidate objects on the end-side from the full set of cloud-side candidate objects. For example, if the local candidate object identifier is A, and the associated object information indicates that the identifiers of A's associated objects are B and C, then the candidate associated object identifier set will include B and C.

[0086] The associated object identifier can be understood as the object identifier of a cloud-side candidate object that is associated with a local candidate object. It's important to note that a cloud-side candidate object is one that is associated with a local candidate object but is not stored locally. Based on the local object identifier, there are several ways to deduplicate object identifiers in the candidate associated object identifier set. For example, deleting object identifiers identical to local object identifiers from the candidate associated object identifier set will yield the associated object identifiers. It should be noted that the size of the edge feature library is much smaller than the cloud feature library; therefore, associated objects in the edge feature library may not be present there, requiring supplementary distribution. In this process, the first step is to retrieve all cloud-side candidate objects that are associated with local candidate objects. However, the retrieved cloud-side candidate objects may also exist in the local feature library. Therefore, deduplication using the local object identifier will then yield the candidate objects that are associated with local candidate objects but are not stored locally.

[0087] After identifying the associated object identifier, the cloud-side candidate object corresponding to the local candidate can be determined based on the associated object identifier. There are multiple ways to determine the cloud-side candidate object. For example, the processor 620 can obtain the cloud-side candidate object set corresponding to the preset biometric set, and filter out the candidate object corresponding to the associated object identifier in the cloud-side candidate object set to obtain the cloud-side candidate object corresponding to the local candidate object.

[0088] After identifying cloud-side candidate objects, the corresponding biometric features can be filtered from the preset biometric feature set. There are multiple ways to filter the biometric features corresponding to cloud-side candidate objects. For example, the processor 620 filters the corresponding biometric features of cloud-side candidate objects from the preset biometric feature set and merges the filtered biometric features to obtain a cloud-side associated biometric feature set.

[0089] Taking facial recognition as an example in biometrics, false recognition primarily stems from misidentification among users with similar appearances. The cloud-based facial feature database contains the facial features of all users in the business database, providing more comprehensive coverage for detecting similar faces within the full user pool. Therefore, the cloud-based associated facial feature set selected from the full facial feature database can provide the client-side with similar face information from a global business perspective, fully considering the high-risk of false recognition among users in the full database. This significantly reduces the probability of false recognition on the client-side, thereby improving the accuracy of facial recognition.

[0090] S230: Send the cloud-side associated biometric set as a cloud-side feature set to the local device.

[0091] For example, the processor 620 can directly send the cloud-related biometric set as a cloud-related feature set to the local device, or it can send the cloud-related biometric set to a third party so that the third party can send the cloud-related biometric set to the local device, or it can send the storage address of the cloud-related biometric set to the local device so that the local device can obtain the cloud-related biometric set based on the storage address and use the cloud-related biometric set as the cloud-related feature set.

[0092] In some embodiments, the cloud side can also send the behavioral data of the local candidate object on the cloud side to the device side as supplementary features for biometric identification. The cloud-side feature set can also include the behavioral data of the local candidate object on the cloud side. Therefore, after receiving the feature acquisition request from the local device, the processor 620 can also generate the historical behavioral features of the local candidate object based on the identification information, and send the historical behavioral features and the cloud-side associated biometric features as a cloud-side feature set to the local device.

[0093] Historical behavioral features can be the behavioral characteristics of local candidate objects within a historical period. There are various types of historical behavioral features; for example, in a face recognition scenario, they could include the user's historical attribute information, historical face-scanning behavior, spatial behavior, etc. Based on the identifier information, there are multiple ways to generate the historical behavioral features of local candidate objects. For example, the processor 620 can filter out the historical behavioral features corresponding to the identifier information from a preset set of historical behavioral features to obtain the historical behavioral features of the local candidate objects. Alternatively, it can obtain the historical behavioral information of the local candidate objects based on the identifier information, perform multi-dimensional feature extraction on the historical behavioral information, and then filter out at least one type of historical behavioral feature from the extracted initial historical behavioral features to obtain the historical behavioral features corresponding to the local candidate objects. Other methods can also be used to generate the historical behavioral features of local candidate objects.

[0094] After generating the historical behavioral features of local candidate objects, the historical behavioral features and the cloud-related biometric feature set can be sent to the local device as a cloud-side feature set, thereby supplementing the problem of limited and missing edge-side data features and further improving the capabilities of the decision-making model.

[0095] It should also be noted that the cloud-side feature set may include historical behavioral features and cloud-side associated biometric features, or at least one of the historical behavioral features and cloud-side associated biometric features.

[0096] (2) The processor 620 indirectly obtains the cloud side feature set from the cloud side.

[0097] For example, the processor 620 can send the identification information of local candidate objects to a third party so that the third party can generate a feature acquisition request based on the identification information of local candidate objects. The feature acquisition request includes the address on the end side. The third party sends the feature acquisition request to the cloud server (server 300) and then receives the associated object information generated based on the identification information and the cloud-side feature set corresponding to the local feature set returned by the cloud server (server 300).

[0098] The method by which the cloud server (server 300) generates associated object information based on identification information and cloud-side feature sets corresponding to local feature sets is described above and will not be repeated here.

[0099] After obtaining the cloud-side feature set stored on the cloud side, such as Figure 3 As shown, method P100 may also include:

[0100] S120: Acquire a target biological image of the target object, and determine the target biological characteristics of the target object based on the target biological image.

[0101] Among them, the target biological image can be a biological image that represents the biological information of the target object. The so-called biological image can include images of the biological information of the object. There can be many types of biological images, such as facial images, body images, organ images or other biological information images, etc.

[0102] Among them, target biometrics can be biometric features that characterize the biological information of the target object. Biometrics can be understood as feature information that characterizes biological features. There are many types of biometrics, such as facial features, body features, organ features, or other biological information features, etc.

[0103] There are several ways to acquire images of the target organism, including the following:

[0104] For example, the processor 620 can directly acquire biological images of the target object through an image acquisition device to obtain the target biological image of the target object; or it can receive target biological images of the target object sent by a third party; or it can obtain target biological images of the target object from a network or image database; or it can crawl raw images from the network, filter out images containing biological information from the raw images, and thus obtain the target biological image of the target object, and so on.

[0105] The image acquisition device can be integrated into the client 200 or the server 300. Taking the target biological image as a face as an example, the client 200 can be of various types, such as a supermarket checkout machine, access control equipment, or the target user 100's mobile terminal. The image acquisition device can include at least one of two-dimensional image acquisition devices, depth image acquisition devices, or other types of image acquisition devices. A two-dimensional image acquisition device can be understood as a device that acquires two-dimensional images, such as an RGB camera, an infrared camera, a visible light camera, or a camera corresponding to other color spaces. A depth image acquisition device can be understood as an image device that acquires three-dimensional images, such as a 3D structured light camera, a laser detector, or a 3D scanner. The images acquired by the depth image acquisition device can include depth images, which can be understood as images with depth information. In some embodiments, a depth image can be a grayscale image with depth information. The grayscale level of a pixel reflects the depth and its changes. It should be noted that a grayscale image with depth information is only one form of depth image. In some embodiments, the depth image can also be presented directly using three-dimensional coordinates, which is more accurate. In some embodiments, a depth image can also be a stereoscopic image generated by combining a two-dimensional image with depth information, which can store and display to the user the shape of the target user's face in various directions in three-dimensional space. In this case, the depth image stores depth information on each surface. In some embodiments, the depth image can be in the form of a two-dimensional image carrying depth information. In this case, it can not only meet the recognition and calculation requirements of three-dimensional images, but also reduce the storage space and computing resources occupied.

[0106] After acquiring the target biological image of the target object, the target biological features of the target object can be determined based on the target biological image. Target biological features can be the biological characteristics of the target object extracted from the target biological image. Biological features can be characteristic information representing the biological information of an object, and there can be various types of biological features, such as at least one of facial features (human face features), body features, blood features, organ features, hair features, or skin color features. There are several ways to determine the target biological features of the target object. For example, the processor 620 can use a biological feature extraction model to extract the target biological features of the target object from the target biological image. Alternatively, the target biological image can be sent to a feature extraction server so that the feature extraction server can extract the biological features from the target biological image. Then, the processor can receive the biological features returned by the feature extraction server and use these biological features as the target biological features of the target object.

[0107] Among them, the biometric feature extraction model can be a model that extracts biometric information from biological images. The structure or type of the biometric feature extraction model can be various, such as convolutional networks, residual networks or various networks that can perform feature extraction, etc.

[0108] S130: Compare the target biometric feature with features in the preset feature set to obtain the feature comparison information corresponding to the target biometric feature.

[0109] The preset feature set includes a local feature set and a cloud-side feature set.

[0110] Feature comparison information can be understood as the information composed of the comparison results between the target biofeature and each biofeature in the preset feature set. The comparison result can be the similarity between biofeatures, or it can be understood as the feature distance or angle between biofeatures, etc.

[0111] There are several ways to compare the target biometric features with features in a preset feature set, including the following:

[0112] For example, the processor 620 can filter out the biometric features corresponding to local candidate objects from the local feature set to obtain a local candidate biometric feature set, filter out the biometric features corresponding to cloud candidate objects from the cloud feature set to obtain a cloud-related biometric feature set, compare the target biometric feature with the biometric features in the target biometric feature set to obtain the feature comparison information corresponding to the target biometric feature. The target biometric feature set includes the local candidate biometric feature set and the cloud-related biometric feature set.

[0113] The local candidate biometric set can be understood as an edge feature library stored locally, which includes the biometric features corresponding to local candidate objects.

[0114] Among these, cloud-side candidate objects and local candidate objects are associated. The so-called association relationship is used to characterize one or more feature relationships between local candidate objects and cloud-side candidate objects. The types of association relationships can be various, such as at least one of the following: appearance similarity, blood relationship, kinship, social relationship, social connection, and pre-defined responsibility and right relationships. There are multiple ways to filter out the biometric features corresponding to cloud-side candidate objects from the cloud-side feature set. For example, the processor 620 can acquire associated object information, determine the cloud-side candidate objects corresponding to local candidate objects based on the associated object information, and filter out the biometric features corresponding to cloud-side candidate objects from the cloud-side feature set to obtain a cloud-side associated biometric feature set.

[0115] There are several ways to determine the cloud-side candidate object corresponding to the local candidate object based on the associated object information. For example, the processor 620 can extract at least one pair of associated objects from the associated object information. The pair of associated objects includes any two associated objects that have an association relationship. The associated object corresponding to the local candidate object is selected from at least one pair of associated objects to obtain the cloud-side candidate object.

[0116] After selecting cloud-side candidate objects, the corresponding biometric features of the cloud-side candidate objects can be directly selected from the cloud-side feature set to obtain the cloud-side associated biometric feature set.

[0117] After selecting the cloud-related biometric feature set, the local candidate biometric features and the cloud-related biometric feature set are used as the target biometric feature set. The target biometric feature can then be compared with the biometric features in the target biometric feature set. There are several ways to compare the target biometric feature with the biometric features in the target biometric feature set. For example, the processor 620 can compare the target biometric feature with the biometric features in the local candidate biometric feature set to obtain local comparison information, compare the target biometric feature with the biometric features in the cloud-related biometric feature set to obtain cloud-side comparison information, or fuse the local comparison information and the cloud-side comparison information to obtain the feature comparison information corresponding to the target biometric feature.

[0118] The local comparison information can be understood as the result of comparing the target biometric feature with the biometric features in the local candidate biometric feature set on the device. There are various ways to compare the target biometric feature with the biometric features in the local candidate biometric feature set. For example, the processor 620 retrieves any biometric feature from the local candidate biometric feature set and compares the retrieved biometric feature with the target biometric feature until all biometric features in the local candidate biometric feature set have been retrieved. This yields the feature comparison result between each biometric feature in the local candidate biometric feature set and the target biometric feature, and these feature comparison results are used as the local feature comparison information.

[0119] There are several ways to compare the retrieved biometric features with the target biometric features. For example, the processor 620 can calculate the feature similarity between the retrieved biometric features and the target biometric features, or it can calculate the feature distance or angle between the retrieved biometric features and the target biometric features, etc., so as to obtain the feature comparison results between the retrieved biometric features and the target biometric features.

[0120] The cloud-side comparison information can be understood as the result of comparing the target biometric feature with the biometric features in the cloud-related biometric feature set sent from the cloud to the device. It's important to note that the cloud-side comparison information is not performed on the cloud itself, but rather on the device; the comparison is done using a cloud-related biometric feature set that comes from pre-delivery or real-time delivery from the cloud. The method for comparing the target biometric feature with the biometric features in the cloud-related biometric feature set is the same as the method for comparing the target biometric feature with the biometric features in the local candidate biometric feature set, as detailed above, and will not be repeated here.

[0121] It should be noted that the processor 620 can compare the target biometrics simultaneously with biometrics in both the local candidate biometrics set and the cloud-based associated biometrics set, or it can perform feature comparisons separately. There are no timing requirements during the separate comparison process; the target biometrics can be compared first with biometrics in the local candidate biometrics set, or first with biometrics in the cloud-based associated biometrics set.

[0122] After obtaining local and cloud-based comparison information, the two sets of information can be fused to obtain the feature comparison information corresponding to the target biometric feature. There are several ways to fuse the local and cloud-based comparison information. For example, the processor 620 can concatenate the local and cloud-based comparison information to obtain the feature comparison information corresponding to the target biometric feature. Alternatively, it can combine or fuse the local and cloud-based comparison information according to a preset combination rule. Another option is to obtain the weighting coefficients for the local and cloud-based comparison information separately, weight the local and cloud-based comparison information based on these coefficients, and then fuse the weighted local and cloud-based comparison information to obtain the feature comparison information corresponding to the target biometric feature.

[0123] In the process of feature comparison, it can be found that in addition to comparing features with the biometrics in the local candidate biometrics set on the client side, it can also compare features with the biometrics in the cloud-associated biometrics set issued by the cloud side. Moreover, the biometrics in the cloud-associated biometrics set are all biometrics of candidate objects that are associated with the local candidate objects on the client side. Thus, from a global business perspective, through the fusion retrieval on the client side, the high risk of misidentification of the target object in the full database can be fully considered.

[0124] S140: Identify the target object based on feature comparison information, local feature set, and cloud-side feature set.

[0125] For example, the processor 620 can determine local target features based on feature comparison information and local feature sets, and identify target objects based on local target features and cloud-side feature sets.

[0126] Local target features can be understood as the feature information generated locally for target object recognition, or as the real-time features on the edge for target object recognition. Based on feature comparison information and the local feature set, there are multiple ways to determine local target features. For example, the processor 620 can extract the current local features from the local feature set and fuse the current local features with the feature comparison information to obtain the local target features.

[0127] Local current features refer to other features that can be obtained in real time on the device side without relying on retrieval (feature comparison). They can also be called local real-time features. There are various types of local current features, such as the real-time geographical location of the local device (equipment), the acquisition time of the target biological image, or the size of the local candidate biological feature set (library), etc. After extracting the local current features, they can be fused with the feature comparison information. There are several ways to fuse them. For example, the processor 620 can concatenate the feature comparison information with the local current features to obtain the local target features. Alternatively, it can combine or fuse the local current features with the feature comparison information according to a preset combination rule to obtain the local target features. Or, it can obtain the weighting coefficients corresponding to the local current features and the feature comparison information respectively, and then weight the local current features and the feature comparison information based on the weighting coefficients. Finally, it can fuse the weighted local current features and the weighted feature comparison information to obtain the local target features.

[0128] After determining the local target features, the target object can be identified based on the local target features and the cloud-side feature set. There are multiple ways to identify the target object. For example, the processor 620 can extract the historical behavior features of the local candidate object from the cloud-side feature set, fuse the historical behavior features with the local target features to obtain the edge decision features corresponding to the target object, and determine the biometric result of the target object based on the edge decision features.

[0129] Among them, historical behavioral features can be the behavioral characteristics of local candidate objects within a historical period. The types of historical behavioral features can be varied; for example, in a facial recognition scenario, they could include the user's historical attribute information, historical facial recognition behavior, spatial behavior, etc. These historical behavioral features can be features pre-deployed from the cloud side. The difference between these historical behavioral features and the cloud-associated biometric feature set is that these historical behavioral features are generated by the cloud side based on local candidate objects from the edge device, and are pre-processed on the cloud side. In contrast, the biometric features in the cloud-associated biometric feature set are the biometric features of candidate objects that are associated with local candidate objects from the full set of cloud-side biometric features.

[0130] Among them, the edge-side decision features are the feature information used to determine the final identification result of the target object based on the retrieval results. These edge-side decision features can also be understood as the feature information input to the biometric model or the identification decision model. There are several ways to fuse historical behavioral features with local target features. For example, the processor 620 can concatenate historical behavioral features with local target features to obtain the edge-side decision features corresponding to the target object; or, it can combine or fuse local target features with historical behavioral features according to preset combination rules to obtain the edge-side decision features corresponding to the target object; or, it can obtain the weighting coefficients corresponding to local target features and historical behavioral features respectively, and then weight the local target features and historical behavioral features based on these weighting coefficients, and finally fuse the weighted local target features and weighted historical behavioral features to obtain the edge-side decision features corresponding to the target object.

[0131] After obtaining the edge decision features corresponding to the target object, the biometric result of the target object can be determined based on these features. The biometric result can be understood as the recognition result generated by the target object in this biometric identification process. This result can be of various types; for example, it can directly identify the target object's identity information or facial information. Or, taking a face recognition scenario as an example, it can also include whether the face recognition was successful, and so on. There are several ways to determine the biometric result of the target object. For example, the processor 620 can use edge decision features to model and obtain an edge decision model, input the edge decision features into the edge decision model, and then output the biometric result of the target object through the edge decision model. Alternatively, the edge decision features can be input into a pre-trained biometric model, and the trained biometric model can make biometric decisions based on the edge decision model, thereby outputting the biometric result of the target object. The decision-making process here can take several forms. For example, the processor 620 can determine candidate biometric results for the target object based on feature comparison information in the edge-side decision features using an edge-side decision model or a trained biometric model. Based on other features in the edge-side decision features, it can then filter out the target object's biometric result from the candidate results. Alternatively, it can directly map the candidate probability of each candidate biometric result corresponding to the target object based on the edge-side decision model, and then decide on the biometric result corresponding to the target object from the candidate results based on this probability. Taking face recognition as an example, when the target object is a locally stored candidate object or a candidate object in cloud storage that is related to a locally stored candidate object, the processor can decide that the target object's face recognition passes; otherwise, it can decide that the target object's face recognition fails, thus obtaining the target object's face recognition result.

[0132] The edge-side decision-making model or the trained biometric model can have various structure, such as CNN networks, GNN networks, residual networks, or any other network structure capable of biometric (face recognition). This solution includes an independent decision-making model stage that considers retrieval data and behavioral data in addition to biometric features. Making decisions through the model further improves decision accuracy, thereby enhancing the accuracy of biometric recognition.

[0133] It should be noted that when performing biometric identification on a target object, cloud-side features can be used or not. Therefore, the types of features included in the edge-side decision features can be varied. For example, it can include local target features and historical behavior features, or it can include only local target features, or it can include only local target features composed of local comparison information and local current features, and so on.

[0134] Taking facial recognition as an example in the context of biometrics, the facial recognition process in this solution can be described as follows: Figure 5 As shown, the face recognition process mainly includes two parts: the edge and the cloud. On the cloud side, supplementary features of user behavior data within the edge feature library can be generated, thus obtaining the historical behavioral features of local candidate objects. On the cloud side, within the full face database, full or partial face feature comparisons can be performed to calculate pairwise similarity. User relationships reaching a certain threshold are defined as associated users, thus obtaining associated user information. This associated user information is sent to the edge side as an edge-side comparison command. Furthermore, based on this associated user information, the biometric features of associated accounts corresponding to local candidate objects can be filtered from the full face database, thus obtaining a cloud-side associated biometric feature set. The cloud-side associated biometric features, associated user information, and pre-processed historical behavioral features of local candidate objects are pre-deployed to the edge side. When a user triggers facial recognition on the device, the device can capture the user's facial image and extract biometric features from it. These extracted biometric features are then fused and retrieved against both the local biometric feature set and a cloud-based pre-deployed biometric feature set. This fusion retrieval can be understood as searching and comparing the local and cloud-based biometric feature sets separately, and then fusing the comparison information to obtain feature comparison information (fused retrieval features). This feature comparison information is then fused with other local real-time features to obtain local target features (device-based real-time features). These local target features are then fused with the historical behavioral features of local candidate objects pre-deployed by the cloud to obtain device-based decision features. Finally, these device-based decision features are input into the device-based decision model (the trained biometric model) for decision-making, outputting the facial recognition result of the target object. In the face recognition process, this solution adopts an architecture where the face recognition decision is made locally on the edge device. It supports local retrieval, and during the recognition process, the cloud-side data delivery capability supports the supplementary delivery of historical experience data from the cloud to the edge device, supporting edge-side decision-making and further improving the accuracy of edge-side decision-making. In addition, it introduces the related objects of local candidate objects, fully mining the high-risk false recognition risk of users in the edge-side database in the full face feature database on the cloud, and delivering it to the edge device through local comparison instructions. The edge device adopts a fusion retrieval method. In addition to the local feature database retrieval, each comparison in the retrieval process is compared with the biometric features of the related user information, transforming the high-risk false recognition risk into fusion retrieval features, which can further improve the decision-making accuracy. Finally, when making decisions, an independent decision-making model (post-trained biometric model) is used, considering retrieval data and behavioral data information in addition to face features. Decision-making through the model can further improve the decision-making accuracy, thereby greatly improving the accuracy of face recognition.

[0135] As can be seen from the above technical solutions, the biometric identification method and system provided in this specification acquire a cloud-side feature set stored on the cloud side, which corresponds to a local feature set stored locally. Then, a target biometric image of the target object is acquired, and based on the target biometric image, the target biometric features of the target object are determined. Then, the target biometric features are compared with features in a preset feature set to obtain feature comparison information corresponding to the target biometric features. The preset feature set includes the local feature set and the cloud-side feature set. Based on the feature comparison information, the local feature set, and the cloud-side feature set, the target object is identified. Since this solution can acquire the cloud-side feature set from the cloud side, and the cloud-side feature set corresponds to the local feature set, the cloud-side feature set can supplement the deficiencies of the local feature set. In addition, by sending cloud-side data to support end-side decision-making, the accuracy of end-side decision-making is further improved. Therefore, the accuracy of biometric identification can be improved.

[0136] This specification, in another aspect, provides a non-transitory storage medium storing at least one set of executable instructions for performing biometric identification. When the executable instructions are executed by a processor, they instruct the processor to implement the steps of the biometric identification method P100 described herein. In some possible embodiments, various aspects of this specification can also be implemented as a program product comprising program code. When the program product is run on a computing device 600, the program code causes the computing device 600 to perform the steps of the biometric identification method P100 described herein. The program product for implementing the above method may employ a portable compact disc read-only memory (CD-ROM) containing program code and may run on the computing device 600. However, the program product of this specification is not limited thereto. In this specification, a readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system. The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. The computer-readable storage medium may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium may also be any readable medium other than a readable storage medium that can send, propagate, or transmit programs for use by or in connection with an instruction execution system, apparatus, or device. Program code contained on a readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof. Program code for performing the operations described herein can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on computing device 600, partially on computing device 600, as a standalone software package, partially on computing device 600 and partially on a remote computing device, or entirely on a remote computing device.

[0137] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0138] In summary, after reading this detailed disclosure, those skilled in the art will understand that the foregoing detailed disclosure is presented by way of example only and is not restrictive. Although not explicitly stated herein, those skilled in the art will understand that this specification requires various reasonable changes, improvements, and modifications to the embodiments. These changes, improvements, and modifications are intended to be made by this specification and are within the spirit and scope of the exemplary embodiments described herein.

[0139] Furthermore, certain terms in this specification have been used to describe embodiments of this specification. For example, "an embodiment," "an embodiment," and / or "some embodiments" mean that a particular feature, structure, or characteristic described in connection with that embodiment may be included in at least one embodiment of this specification. Therefore, it is to be emphasized and understood that two or more references to "an embodiment" or "an embodiment" or "alternative embodiment" in various parts of this specification do not necessarily refer to the same embodiment. Moreover, specific features, structures, or characteristics may be suitably combined in one or more embodiments of this specification.

[0140] It should be understood that in the foregoing description of the embodiments in this specification, various features are combined in a single embodiment, drawing, or description for the purpose of simplifying the description and aiding in the understanding of a feature. However, this does not mean that the combination of these features is necessary, and those skilled in the art may readily identify some of the devices as separate embodiments when reading this specification. That is, the embodiments in this specification can also be understood as an integration of multiple secondary embodiments. It is also valid when each secondary embodiment contains fewer than all the features of a single foregoing disclosed embodiment.

[0141] Each patent, patent application, publication of the patent application, and other materials such as articles, books, specifications, publications, documents, articles, etc., cited herein may be incorporated by reference. All contents used for all purposes, except for any history of prosecution documents relating to it, that may be inconsistent with or conflict with this document, or any such history of prosecution documents that may have a limiting effect on the widest extent of the claims, are now or hereafter associated with this document. For example, in the event of any inconsistency or conflict between the description, definition, and / or use of terms associated with any of the included materials and the terms, description, definition, and / or used in connection with this document, the terms used herein shall prevail.

[0142] Finally, it should be understood that the embodiments disclosed herein are illustrative of the principles of the embodiments described in this specification. Other modified embodiments are also within the scope of this specification. Therefore, the embodiments disclosed in this specification are merely examples and not limitations. Those skilled in the art can implement the applications described in this specification using alternative configurations based on the embodiments in this specification. Therefore, the embodiments in this specification are not limited to the embodiments precisely described in the applications.

Claims

1. A biometric identification method, comprising: obtaining a cloud-side feature set stored at a cloud side, the cloud-side feature set corresponding to a local feature set stored at a local side; obtaining a target biometric image of a target object, and determining a target biometric feature of the target object based on the target biometric image; performing feature comparison between the target biometric feature and features in a preset feature set to obtain feature comparison information corresponding to the target biometric feature, the preset feature set comprising the local feature set and the cloud-side feature set, the local feature set comprising biometric features corresponding to local candidate objects, the cloud-side feature set comprising biometric features corresponding to cloud-side candidate objects, the cloud-side candidate objects having an association relationship with the local candidate objects; and identifying the target object based on the feature comparison information, the local feature set and the cloud-side feature set.

2. The method of biometric identification according to claim 1, wherein, The performing feature comparison between the target biometric feature and features in a preset feature set to obtain feature comparison information corresponding to the target biometric feature comprises: filtering the biometric features corresponding to the local candidate objects from the local feature set to obtain a local candidate biometric feature set; filtering the biometric features corresponding to the cloud-side candidate objects from the cloud-side feature set to obtain a cloud-side associated biometric feature set; and performing comparison between the target biometric feature and biometric features in a target biometric feature set to obtain feature comparison information corresponding to the target biometric feature, the target biometric feature set comprising the local candidate biometric feature set and the cloud-side associated biometric feature set.

3. The method of biometric identification according to claim 2, wherein, The association relationship comprises at least one of a facial similarity, a blood relationship, a kinship, a social relationship and a preset responsibility relationship.

4. The method of biometric identification according to claim 2, wherein, The filtering the biometric features corresponding to the cloud-side candidate objects from the cloud-side feature set to obtain a cloud-side associated biometric feature set comprises: obtaining association object information; determining the cloud-side candidate objects corresponding to the local candidate objects based on the association object information; and filtering the biometric features corresponding to the cloud-side candidate objects from the cloud-side feature set to obtain a cloud-side associated biometric feature set.

5. The method of biometric identification according to claim 4, wherein, The determining the cloud-side candidate objects corresponding to the local candidate objects based on the association object information comprises: extracting at least one association object pair from the association object information, the association object pair comprising any two association objects having an association relationship; and filtering the association objects corresponding to the local candidate objects from the at least one association object pair to obtain the cloud-side candidate objects.

6. The method of biometric identification according to claim 2, wherein, The performing comparison between the target biometric feature and biometric features in a target biometric feature set to obtain feature comparison information corresponding to the target biometric feature comprises: performing comparison between the target biometric feature and biometric features in the local candidate biometric feature set to obtain local comparison information; performing comparison between the target biometric feature and biometric features in the cloud-side associated biometric feature set to obtain cloud-side comparison information; and fusing the local comparison information and the cloud-side comparison information to obtain the feature comparison information corresponding to the target biometric feature.

7. The method of biorecognition according to claim 1, wherein, The target object is identified based on the feature comparison information, the local feature set, and the cloud-side feature set, including: determining a local target feature based on the feature comparison information and the local feature set; and identifying the target object based on the local target feature and the cloud-side feature set.

8. The method of biometric identification according to claim 7, wherein, The local target feature is determined based on the feature comparison information and the local feature set, including: extracting a local current feature from the local feature set; and fusing the local current feature with the feature comparison information to obtain a local target feature.

9. The method of biometric identification according to claim 7, wherein, The target object is identified based on the local target feature and the cloud-side feature set, including: extracting a historical behavior feature of the local candidate object from the cloud-side feature set; fusing the historical behavior feature with the local target feature to obtain an end-side decision feature corresponding to the target object; and determining a biological recognition result of the target object based on the end-side decision feature.

10. The method of biorecognition according to claim 1, wherein, The cloud-side feature set stored on the cloud side is obtained, including: sending a feature acquisition request to a cloud server, the feature acquisition request including identification information of a local candidate object; receiving, from the cloud server, associated object information generated based on the identification information and a cloud-side feature set corresponding to the local feature set.

11. A biological recognition method applied to a distance server in communication connection with a local device, including: receiving a feature acquisition request sent by the local device, the feature acquisition request including identification information of a local candidate object, the local device storing biological feature information of the local candidate object and the identification information; filtering out a cloud-side associated biological feature set from a preset biological feature set based on the identification information, the cloud-side associated biological feature set including biological features of objects associated with the local candidate object; and sending the cloud-side associated biological feature set to the local device as a cloud-side feature set, for the local device to perform the biological recognition method of any one of claims 1 to 10, and identify the target object based on the cloud-side feature set.

12. The method of biometric identification according to claim 11, wherein, The cloud-side associated biological feature set is filtered out from the preset biological feature set based on the identification information, including: determining associated object information based on the preset biological feature set; determining a cloud-side candidate object corresponding to the local candidate object based on the identification information and the associated object information; and filtering out biological features corresponding to the cloud-side candidate object from the preset biological feature set to obtain the cloud-side associated biological feature set.

13. The method of biometric identification according to claim 12, wherein, The associated object information is determined based on the preset biological feature set, including: obtaining feature similarity between biological features in the preset biological feature set; filtering out at least one biological feature pair from the preset biological feature set, the biological feature pair being any two biological features with a similarity exceeding a preset similarity threshold; and determining associated object information based on the at least one biological feature pair.

14. The method of biometric identification according to claim 13, wherein, The determining the associated object information based on the at least one biometric feature pair comprises: obtaining a target object identifier of at least one candidate object corresponding to the at least one biometric feature pair; determining at least one associated object pair based on the target object identifier, the at least one associated object pair corresponding to the at least one biometric feature pair; and fusing the at least one associated object pair to obtain the associated object information.

15. The method of biorecognition according to claim 12, wherein, The determining the cloud-side candidate object corresponding to the local candidate object based on the identifier information and the associated object information comprises: obtaining a set of object identifier pairs corresponding to the associated object information; comparing the object identifier in the identifier information with the object identifiers in the set of object identifier pairs to obtain an associated object identifier; and determining the cloud-side candidate object corresponding to the local candidate object based on the associated object identifier.

16. The method of biometric identification according to claim 15, wherein, The comparing the object identifier in the identifier information with the object identifiers in the set of object identifier pairs to obtain an associated object identifier comprises: extracting a local object identifier of the local candidate object from the identifier information; screening at least one object identifier corresponding to the local object identifier from the set of object identifier pairs to obtain a set of candidate associated object identifiers; and based on the local object identifier, deduplicating the object identifiers in the set of candidate associated object identifiers to obtain the associated object identifier.

17. The method of biorecognition of claim 11, wherein, The receiving the feature acquisition request sent by the local device further comprises: generating a historical behavior feature of the local candidate object based on the identifier information; and The sending the cloud-side associated biometric feature set to the local device as a cloud-side feature set comprises sending the historical behavior feature and the cloud-side associated biometric feature set to the local device as the cloud-side feature set.

18. A biometric identification system comprising: at least one storage medium storing at least one instruction set for biometric identification; and at least one processor communicatively connected to the at least one storage medium, wherein when the biometric identification system is running, the at least one processor reads the at least one instruction set and executes the method of biometric identification according to the instructions of the at least one instruction set as claimed in any one of claims 1-17. ​

Images