Multi-modal biological characteristic adaptive identification method and system, intelligent door lock and storage medium

Abstract

The invention discloses a multi-modal biological characteristic self-adaptive identification method and system, an intelligent door lock and a storage medium, and the method comprises the steps: obtaining biological characteristics of a user, carrying out the confidence query in a pre-stored multi-modal biological characteristic library according to the biological characteristics, and obtaining an initial confidence score; acquiring environment parameters in the sensor, adjusting the initial confidence score according to the environment parameters to obtain a plurality of environment weights, and performing weight calculation according to all the environment weights to obtain a biological recognition weight; and acquiring a current timestamp, selecting a target recognition strategy from a modal biological feature library according to the timestamp, the biological recognition weight and the environmental parameters, performing identity authentication on a user according to the target recognition strategy, and unlocking after the authentication is passed. According to the invention, through collection of biological characteristics, environmental adjustment confidence, selection of strategy authentication and unlocking after successful identity verification, the recognition accuracy is improved, the failure rate is reduced, and the security and user experience are improved at the same time.

Description

Technical Field

[0001] The present invention relates to the technical field of smart door locks, and in particular to a multimodal biometric adaptive recognition method and system, a smart door lock, and a computer-readable storage medium. Background Art

[0002] With the development of smart security technology, biometrics have been widely used in identity authentication. Traditional single biometric technologies have certain limitations. For example, fingerprint recognition is easily affected by factors such as moisture and wear on the finger, facial recognition fails in low-light or obscured scenes, and iris recognition requires active user cooperation and is relatively expensive. To improve the accuracy and security of identity authentication, multimodal biometric technology has emerged. However, existing multimodal biometric systems typically use a fixed combination of biometric features for verification, lacking flexibility and environmental adaptability, resulting in a poor user experience. As a result, when smart door locks use a single recognition technology, they are unable to adaptively switch to the corresponding recognition method according to the environment, which has become a pressing problem that needs to be solved.

[0003] Therefore, the existing technology still needs to be improved and developed. Summary of the Invention

[0004] The main purpose of the present invention is to provide a multimodal biometric adaptive recognition method, system, smart door lock and computer-readable storage medium, aiming to solve the problem in the prior art that when smart door locks use a single technology for recognition, the smart door locks cannot adaptively switch the corresponding recognition method according to the environment, resulting in the smart door locks being unable to recognize the user and unlock the door under the interference of environmental factors.

[0005] To achieve the above object, the present invention provides a multimodal biometric adaptive recognition method, which comprises the following steps:

[0006] Obtaining a user's biometric features, and performing a confidence query on a pre-stored multimodal biometric feature database based on the biometric features to obtain an initial confidence score;

[0007] Acquiring environmental parameters from a sensor, adjusting the initial confidence score according to the environmental parameters to obtain a plurality of environmental weights, and performing weight calculation based on all the environmental weights to obtain a biometric recognition weight;

[0008] Obtain the current timestamp, select a target recognition strategy from the modal biometric library according to the timestamp, the biometric recognition weight and the environmental parameters, authenticate the user according to the target recognition strategy and unlock the door after the authentication is passed.

[0009] Optionally, in the multimodal biometric adaptive recognition method, the biometric feature includes any one of a fingerprint feature, a facial feature, a palm vein feature, a finger vein feature, and a voiceprint feature;

[0010] The step of obtaining the user's biometric features and querying the confidence of a pre-stored multimodal biometric feature library based on the biometric features to obtain an initial confidence score specifically includes:

[0011] Receiving a biometric feature of a user acquired by a sensor, and performing a confidence query in a pre-stored multimodal biometric feature library based on the fingerprint feature, the facial feature, the palm vein feature, the finger vein feature, or the voiceprint feature to obtain an initial confidence score;

[0012] The initial confidence score is used by the smart door lock to adjust according to different influencing information.

[0013] Optionally, in the multimodal biometric adaptive recognition method, the environmental parameters include any one or more of light intensity, humidity and temperature;

[0014] The step of acquiring environmental parameters from the sensor, adjusting the initial confidence score according to the environmental parameters to obtain a plurality of environmental weights, and performing weight calculation according to all the environmental weights to obtain a biometric recognition weight specifically includes:

[0015] Obtaining any one or more of the light intensity, the humidity, and the temperature collected by a sensor, and adjusting the initial confidence score according to any one or more of the light intensity, the humidity, and the temperature to obtain multiple environmental weights;

[0016] A weight calculation is performed based on all the environmental weights to obtain a plurality of weight scores, and all the weight scores are sorted in order from high to low to obtain a biometric weight.

[0017] Optionally, the multimodal biometric adaptive recognition method, wherein the weight calculation is performed based on all the environmental weights to obtain multiple weight scores, and all the weight scores are sequentially sorted to obtain the biometric recognition weight, specifically includes:

[0018] Performing matching calculation on the biometric features based on a biometric recognition algorithm to obtain similarity;

[0019] Performing weight calculation based on the environmental adaptability coefficient in the sensor, all the environmental weights, and the similarity to obtain a plurality of weight scores, and sequentially sorting all the weight scores to obtain a target weight;

[0020] A weighted calculation is performed according to the environmental adaptation coefficient, the target weight and the similarity to obtain a biometric recognition weight.

[0021] Optionally, in the multimodal biometric adaptive recognition method, the modal biometric library includes fingerprint recognition, face recognition, biomarker pattern recognition, palm vein recognition, and finger vein recognition;

[0022] The obtaining of the current timestamp, selecting a target recognition strategy from the modal biometric library according to the timestamp, the biometric recognition weight, and the environmental parameters, and authenticating the user according to the target recognition strategy and unlocking the door after the authentication is passed, specifically includes:

[0023] Obtaining a current timestamp, and dynamically selecting among fingerprint recognition, face recognition, biometric recognition, palm vein recognition, and finger vein recognition based on the timestamp and the biometric weight, as well as the light intensity, humidity, or temperature, to obtain multiple recognition strategies;

[0024] Screening the plurality of identification strategies according to a preset threshold trigger strategy to obtain a target identification strategy;

[0025] Authenticate the user according to the target identification strategy, and unlock the door when the authentication is passed;

[0026] The target recognition strategy includes any one or more of target fingerprint recognition, target face recognition, target biomarker recognition, target palm vein recognition and target finger vein recognition.

[0027] Optionally, the multimodal biometric adaptive recognition method, wherein a current timestamp is obtained, a target recognition strategy is selected from the modal biometric library according to the timestamp, the biometric recognition weight, and the environmental parameters, and the user is authenticated according to the target recognition strategy and unlocked after the authentication is passed, further comprises:

[0028] Acquire a data set of a smart door lock, preprocess the data set to obtain a target data set, analyze the target recognition strategy based on the target data set and a preset time series model, and obtain an optimal recognition strategy;

[0029] The data set includes the number of unlocking times, biometric identification method and unlocking duration.

[0030] Optionally, the multimodal biometric adaptive recognition method, wherein the step of acquiring a data set of a smart door lock, preprocessing the data set to obtain a target data set, and analyzing the target recognition strategy based on the target data set and the preset time series model to obtain an optimal recognition strategy, specifically includes:

[0031] Obtaining the number of unlocks, the biometric identification method, and the unlocking duration of the smart door lock, preprocessing the number of unlocks, the biometric identification method, and the unlocking duration to obtain a target data set, and training a preset time series model based on the target data set to obtain a neural network model;

[0032] Analyzing the target recognition strategy through the neural network model to obtain an optimal recognition strategy;

[0033] Wherein, the preprocessing includes outlier processing and missing value filling processing;

[0034] The preset timing model includes TCN or LSTM.

[0035] In addition, to achieve the above-mentioned purpose, the present invention further provides a multimodal biometric adaptive recognition system, wherein the multimodal biometric adaptive recognition system:

[0036] An initial confidence acquisition module is used to obtain the user's biometric features, perform a confidence query on the pre-stored multimodal biometric feature library based on the biometric features, and obtain an initial confidence score;

[0037] a biometric weight calculation module, configured to obtain environmental parameters from a sensor, adjust the initial confidence score according to the environmental parameters to obtain a plurality of environmental weights, and perform weight calculation based on all the environmental weights to obtain a biometric weight;

[0038] The target recognition strategy identification module is used to obtain the current timestamp, select the target recognition strategy from the modal biometric library according to the timestamp, the biometric weight and the environmental parameters, and authenticate the user according to the target recognition strategy and unlock the door after the authentication is passed.

[0039] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a multimodal biometric adaptive recognition program, and when the multimodal biometric adaptive recognition program is executed by a processor, it implements the steps of the multimodal biometric adaptive recognition method described above.

[0040] In the present invention, a user's biometrics are acquired, and a confidence query is performed on a pre-stored multimodal biometric library based on the biometrics to obtain an initial confidence score. Environmental parameters are acquired from the sensor, and the initial confidence score is adjusted based on the environmental parameters to obtain multiple environmental weights. A weight calculation is performed based on all the environmental weights to obtain a biometric weight. The current timestamp is acquired, and a target recognition strategy is selected from the modal biometric library based on the timestamp, the biometric weights, and the environmental parameters. The user is authenticated according to the target recognition strategy, and the door is unlocked after successful authentication. By acquiring biometrics, adjusting confidence based on the environment, selecting a strategy for authentication, and unlocking the door after successful authentication, the present invention improves recognition accuracy, reduces failure rate, and enhances security and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a flow chart of a preferred embodiment of the multimodal biometric adaptive recognition method of the present invention;

[0042] Figure 2 is a flow chart of step S20 of a preferred embodiment of the multimodal biometric adaptive recognition method of the present invention;

[0043] Figure 3 is a flow chart of S22 of a preferred embodiment of the multimodal biometric adaptive recognition method of the present invention;

[0044] Figure 4 is a flow chart of S30 of a preferred embodiment of the multimodal biometric adaptive recognition method of the present invention;

[0045] Figure 5 1 is a structural diagram of a preferred embodiment of the multimodal biometric adaptive recognition system of the present invention;

[0046] Figure 6 It is a structural diagram of a preferred embodiment of the intelligent door lock of the device of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of the present invention more clear and distinct, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0048] With the development of intelligent security technology, biometrics have been widely used in identity authentication. Traditional single-mode biometrics have limitations. For example, fingerprint recognition is susceptible to moisture and wear, facial recognition fails in low-light or obscured conditions, and iris recognition requires active user cooperation and is costly. To improve the accuracy and security of identity authentication, multimodal biometrics have emerged. However, existing multimodal biometric systems typically use a fixed combination of biometric features for verification, lacking flexibility and environmental adaptability, resulting in a poor user experience. Furthermore, existing technologies fail to dynamically adjust recognition methods based on environmental conditions, resulting in increased recognition failure rates. They also lack the ability to learn from user habits and are unable to optimize the verification process. Multimodal biometrics simply superimpose, without enhancing security through weight allocation and confidence fusion. Therefore, a multimodal biometric adaptive recognition method is needed that addresses the limitations of single-mode biometrics through dynamic weight allocation, environmental adaptation strategies, and AI learning mechanisms, while improving security and user experience.

[0049] The multimodal biometric adaptive recognition method described in the preferred embodiment of the present invention is as follows: Figure 1 As shown, the multimodal biometric adaptive recognition method includes the following steps:

[0050] Step S10: Obtain the user's biometric features, perform a confidence query in a pre-stored multimodal biometric feature database based on the biometric features, and obtain an initial confidence score.

[0051] Specifically, the biometric features include; any one of fingerprint features, facial features, palm vein features, finger vein features and voiceprint features (first, the biometric data (biometric features) are input, and the corresponding biometric information is collected through different biometric sensors / modules, such as fingerprints, faces, palm veins / finger veins, voiceprints, etc. Further, liveness detection is enhanced, such as combining 3D structured light / 3DTOF stereo environment detection during binocular face verification to prevent forgery attacks such as photos and masks); the biometric features of the user acquired by the sensor are received, and a confidence query is performed in a pre-stored multimodal biometric library based on the fingerprint features, the facial features, the palm vein features, the finger vein features or the voiceprint features to obtain an initial confidence score (the smart lock system assigns a confidence score to each biometric method). An initial confidence score (0% to 100%), and adjusted during use according to different dimensional information); wherein, the initial confidence score is used for smart door locks to adjust according to different influencing information, such as: environmental factor assessment, lighting conditions, the confidence of face recognition under low light or high light conditions is reduced from 90% to 70%, while the confidence of recognition methods that do not rely on light, such as fingerprints, is improved, noise level, under high noise conditions, the confidence of voiceprint recognition is reduced, for example, from 80% to 50%, while the confidence of recognition methods such as face and vein is improved, ambient humidity, when the ambient humidity is too high, the confidence of fingerprint recognition affected by hand humidity is reduced, and the confidence of face recognition is improved. The principle is to avoid the weaknesses of the sensor according to the environmental conditions and realize the complementarity of information of different dimensions.

[0052] Step S20: Acquire environmental parameters from the sensor, adjust the initial confidence score according to the environmental parameters to obtain multiple environmental weights, perform weight calculation based on all the environmental weights to obtain a biometric weight.

[0053] like Figure 2 As shown, step S20 includes:

[0054] Step S21: obtaining any one or more of the light intensity, the humidity, and the temperature collected by the sensor, and adjusting the initial confidence score according to any one or more of the light intensity, the humidity, and the temperature to obtain multiple environmental weights;

[0055] Step S22: performing weight calculation based on all the environmental weights to obtain multiple weight scores, and sorting all the weight scores in order from high to low to obtain the biometric weight.

[0056] Specifically, any one or more of the light intensity, the humidity and the temperature collected by the sensor are obtained, and the initial confidence score is adjusted according to any one or more of the light intensity, the humidity and the temperature (weight adjustment is performed according to user usage habits, for example, if the user is accustomed to unlocking at night, the weight of fingerprint recognition is increased by 10% at night, and when a certain verification method fails repeatedly, the weight of this recognition method is reduced) to obtain multiple environmental weights; weight calculation is performed according to all the environmental weights to obtain multiple weight scores, and all weight scores are sorted in order from high to low to obtain biometric weight (A: algorithm matching score, B: environmental adaptability coefficient (environmental confidence obtained by detecting environmental data by environmental sensors), C: user behavior weighting a, b, c: are learnable parameters, and dynamic optimization is performed through reinforcement learning. The final biometric weight is assumed to be W: W = a*A+b*B+c*C).

[0057] For example, on rainy days, the environment is humid and fingerprints are easily affected, so the environmental coefficient B fingerprint decreases by 30%. At the same time, the environmental coefficient B face of the face method frequently used by users increases by 20%. The door lock system automatically increases the weight of face recognition and reduces the weight of fingerprint recognition.

[0058] like Figure 3 As shown, step S22 includes:

[0059] Step S221: performing a matching calculation on the biometric features based on a biometric recognition algorithm to obtain a similarity;

[0060] Step S222: performing weight calculation based on the environmental adaptability coefficient in the sensor, all the environmental weights, and the similarity to obtain a plurality of weight scores, and sequentially sorting all the weight scores to obtain a target weight;

[0061] Step S223: Perform weighted calculation based on the environmental adaptation coefficient, the target weight, and the similarity to obtain a biometric recognition weight.

[0062] Specifically, a matching calculation is performed on the biometric feature based on a biometric recognition algorithm to obtain a similarity (0% to 100% similarity output by the biometric recognition algorithm (matching rate from the recognition model, such as the similarity between the fingerprint image, face image verification pattern and the comparison library pattern)), a weight calculation is performed based on the environmental adaptability coefficient in the sensor, all the environmental weights and the similarity to obtain multiple weight scores (dynamic weight allocation strategy, dynamically adjust the weight according to the security level of the biometric feature (environmental adaptability coefficient), environmental applicability (environmental adaptability coefficient) and real-time confidence (the similarity), and calculate a comprehensive security score (multiple weight scores)), all weight scores are sorted in sequence to obtain a target weight (the recognition results of each module are merged according to the weight (such as when the fingerprint passes, the face fails, and the iris passes), the module with a high weight dominates the final decision, that is, the module with a high weight is set as the target weight), and a weighted calculation is performed based on the environmental adaptability coefficient, the target weight and the similarity to obtain a biometric weight.

[0063] Furthermore, each verification result (success / failure) is fed back to the door lock main system, and the weight parameters a, b, and c are updated through a lightweight neural network (such as a TinyML model). Simulated attack data (such as forged fingerprints and imitation 3D masks) are injected to reduce the long-term weight of vulnerable modules. According to individual usage habits and individual characteristics, a unique weight baseline is generated (for example, the fingerprint quality of children and the elderly is poor, so the weights of face and vein recognition are automatically increased). Secondly, when there is a contradiction in the weight distribution (such as the face passes with a high weight but the iris detects a live attack), secondary authentication of other high-weight recognition methods is triggered. When a biometric sensor fails, the recognition method is turned off and automatically switched to other recognition methods (for example, when a fingerprint sensor abnormality is detected, the recognition method is turned off and other methods are automatically turned on, and the weights are readjusted).

[0064] In this embodiment, the environmental perception module includes a light sensor, a temperature and humidity sensor, and a 3D time-of-flight (TOF) module. The light sensor detects ambient light intensity, the temperature and humidity sensor detects ambient temperature and humidity, and the 3D time-of-flight (TOF) module is used for liveness detection. For example, the biometric fusion module fuses biometric features from various modalities using a weighted fusion strategy. Initial weights are assigned based on the biometric's security level and environmental suitability, for example, a weight of 0.6 for face, 0.4 for finger vein, 0.3 for fingerprint, and 0.1 for palm vein. Real-time weights are dynamically adjusted based on the matching confidence of each modality.

[0065] Step S30: Obtain the current timestamp, select a target recognition strategy from the modal biometric library according to the timestamp, the biometric recognition weight and the environmental parameters, authenticate the user according to the target recognition strategy and unlock the door after the authentication is passed.

[0066] like Figure 4 As shown, step S30 includes:

[0067] Step S31: obtaining a current timestamp, and dynamically selecting among fingerprint recognition, face recognition, biometric recognition, palm vein recognition, and finger vein recognition based on the timestamp and the biometric recognition weight, as well as the light intensity, humidity, or temperature, to obtain multiple recognition strategies;

[0068] Step S32: screening the plurality of identification strategies according to a preset threshold trigger strategy to obtain a target identification strategy;

[0069] Step S33: Authenticate the user according to the target identification strategy, and unlock the door when the authentication is successful.

[0070] Specifically, the current timestamp is obtained, and the fingerprint recognition, the face recognition, the iris recognition, the palm vein recognition and the finger vein recognition are dynamically selected according to the timestamp and the biometric weight, as well as the light intensity or the humidity or the temperature to obtain multiple recognition strategies (the adaptive decision module dynamically selects the optimal recognition method according to environmental parameters, time and other factors. For example, finger vein recognition is used first when the light is insufficient, and palm vein recognition is switched to when the temperature and humidity are too high. In abnormal cases, such as fingerprint failure for three consecutive times, it automatically switches to face recognition and records a log. In the middle of the night or at unusual times, finger vein + face dual verification is triggered. For example, in abnormal cases, such as iris recognition fails for three consecutive times, it automatically switches to palm print recognition and records a log. In the middle of the night or at unusual times, iris + face + voice triple verification is triggered). The strategy is triggered for multiple of the above according to the preset threshold. The recognition strategy is screened to obtain the target recognition strategy (the verification order is automatically optimized, and the user's most commonly used biometric verification method is prioritized by default (target recognition strategy). If the confidence level is lower than the threshold (preset threshold trigger strategy), secondary verification is automatically triggered. For example, when a certain recognition method is used more than 60% of the time in the past 7 days, the recognition method is set as the default recognition method. When a certain recognition method is used less than 10% of the time in the past seven days, the recognition method is turned off and set as the secondary recognition method). The user is authenticated according to the target recognition strategy, and the door is unlocked after the authentication is passed. Secondly, the verification order is automatically optimized, and the user's most commonly used biometric verification method is prioritized by default. If the confidence level is lower than the threshold, secondary verification is automatically triggered (secondary verification refers to other recognition methods that are not commonly used. For example, if the user often uses fingerprints and faces, palm vein, iris, etc. will be automatically downgraded to secondary verification methods).

[0071] Furthermore, after step S30, the step also includes obtaining a data set of the smart door lock, preprocessing the data set to obtain a target data set, analyzing the target recognition strategy according to the target data set and a preset timing model to obtain an optimal recognition strategy; wherein, the data set includes the number of unlocking times, biometric recognition method and unlocking time (timestamp (accurate to minutes), i.e., unlocking time, biometric recognition method used (fingerprint / face / iris, etc.), unlocking time, number of attempts, and obtaining user behavior information, such as the user actively switching the recognition method from fingerprint recognition to face recognition).

[0072] In this embodiment, the number of unlocking times, biometric identification method and unlocking time of the smart door lock are obtained, the number of unlocking times, the biometric identification method and the unlocking time are preprocessed to obtain a target data set, and a preset time series model is trained according to the target data set to obtain a neural network model; the target recognition strategy is analyzed by the neural network model to obtain an optimal recognition strategy; wherein, the preprocessing includes outlier processing and missing value filling processing (real-time collection of data from the device end, building a sample library with the collected data, preprocessing the data in the sample library, outlier processing, missing value filling, the processed data is put into the model for training, and the output of the model is saved for decision analysis); wherein, the preset time series model includes TCN or LSTM (select a suitable model according to needs, for example: the time series model selects a lightweight time series model such as TCN or LSTM, micro Transformer: uses a single-head attention mechanism for long sequence analysis, such as analysis of a week's behavior pattern).

[0073] Furthermore, if Figure 5 As shown, based on the above multimodal biometric adaptive recognition method, the present invention also provides a multimodal biometric adaptive recognition system, wherein the multimodal biometric adaptive recognition system includes:

[0074] An initial confidence acquisition module 51 is used to obtain a user's biometric features, perform a confidence query on a pre-stored multimodal biometric feature database based on the biometric features, and obtain an initial confidence score;

[0075] a biometric weight calculation module 52 for obtaining environmental parameters from a sensor, adjusting the initial confidence score according to the environmental parameters to obtain a plurality of environmental weights, and performing weight calculation based on all the environmental weights to obtain a biometric weight;

[0076] The target recognition strategy identification module 53 is used to obtain the current timestamp, select a target recognition strategy from the modal biometric library according to the timestamp, the biometric weight and the environmental parameters, and authenticate the user according to the target recognition strategy and unlock the door after the authentication is passed.

[0077] Furthermore, if Figure 6 As shown, based on the above-mentioned multimodal biometric adaptive recognition method and system, the present invention also provides a smart door lock, which includes a processor 10, a memory 20 and a display 30. Figure 6 Only some components of the smart door lock are shown, but it should be understood that it is not required to implement all the shown components, and more or fewer components may be implemented instead.

[0078] In some embodiments, the memory 20 may be an internal storage unit of the smart door lock, such as a hard disk or memory of the smart door lock. In other embodiments, the memory 20 may also be an external storage device of the smart door lock, such as a plug-in hard disk equipped on the smart door lock, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Furthermore, the memory 20 may also include both an internal storage unit of the smart door lock and an external storage device. The memory 20 is used to store application software and various types of data installed in the smart door lock, such as the program code for installing the smart door lock. The memory 20 may also be used to temporarily store data that has been output or is to be output. In one embodiment, a multimodal biometric adaptive recognition program 40 is stored on the memory 20, and the multimodal biometric adaptive recognition program 40 can be executed by the processor 10, thereby implementing the multimodal biometric adaptive recognition method in this application.

[0079] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other data processing chip, configured to execute program codes or process data stored in the memory 20, such as executing the multimodal biometric adaptive recognition method.

[0080] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 30 is used to display information on the smart door lock and to display a visual user interface. The smart door locks communicate with each other via a system bus.

[0081] In one embodiment, when the processor 10 executes the multimodal biometric adaptive recognition program 40 in the memory 20, the following steps are implemented:

[0082] Obtaining a user's biometric features, and performing a confidence query on a pre-stored multimodal biometric feature database based on the biometric features to obtain an initial confidence score;

[0083] Acquiring environmental parameters from a sensor, adjusting the initial confidence score according to the environmental parameters to obtain a plurality of environmental weights, and performing weight calculation based on all the environmental weights to obtain a biometric recognition weight;

[0084] Obtain the current timestamp, select a target recognition strategy from the modal biometric library according to the timestamp, the biometric recognition weight and the environmental parameters, authenticate the user according to the target recognition strategy and unlock the door after the authentication is passed.

[0085] The biometric feature includes any one of fingerprint feature, facial feature, palm vein feature, finger vein feature and voiceprint feature;

[0086] The step of obtaining the user's biometric features and querying the confidence of a pre-stored multimodal biometric feature library based on the biometric features to obtain an initial confidence score specifically includes:

[0087] Receiving a biometric feature of a user acquired by a sensor, and performing a confidence query in a pre-stored multimodal biometric feature library based on the fingerprint feature, the facial feature, the palm vein feature, the finger vein feature, or the voiceprint feature to obtain an initial confidence score;

[0088] The initial confidence score is used by the smart door lock to adjust according to different influencing information.

[0089] Wherein, the environmental parameters include any one or more of light intensity, humidity and temperature;

[0090] The step of acquiring environmental parameters from the sensor, adjusting the initial confidence score according to the environmental parameters to obtain a plurality of environmental weights, and performing weight calculation according to all the environmental weights to obtain a biometric recognition weight specifically includes:

[0091] Obtaining any one or more of the light intensity, the humidity, and the temperature collected by a sensor, and adjusting the initial confidence score according to any one or more of the light intensity, the humidity, and the temperature to obtain multiple environmental weights;

[0092] A weight calculation is performed based on all the environmental weights to obtain a plurality of weight scores, and all the weight scores are sorted in order from high to low to obtain a biometric weight.

[0093] The weight calculation is performed based on all the environmental weights to obtain multiple weight scores, and all the weight scores are sorted in sequence to obtain the biometric weight, which specifically includes:

[0094] Performing matching calculation on the biometric features based on a biometric recognition algorithm to obtain similarity;

[0095] Performing weight calculation based on the environmental adaptability coefficient in the sensor, all the environmental weights, and the similarity to obtain a plurality of weight scores, and sequentially sorting all the weight scores to obtain a target weight;

[0096] A weighted calculation is performed according to the environmental adaptation coefficient, the target weight and the similarity to obtain a biometric recognition weight.

[0097] The modal biometric library includes fingerprint recognition, face recognition, biomarker recognition, palm vein recognition and finger vein recognition;

[0098] The obtaining of the current timestamp, selecting a target recognition strategy from the modal biometric library according to the timestamp, the biometric recognition weight, and the environmental parameters, and authenticating the user according to the target recognition strategy and unlocking the door after the authentication is passed, specifically includes:

[0099] Obtaining a current timestamp, and dynamically selecting among fingerprint recognition, face recognition, biometric recognition, palm vein recognition, and finger vein recognition based on the timestamp and the biometric weight, as well as the light intensity, humidity, or temperature, to obtain multiple recognition strategies;

[0100] Screening the plurality of identification strategies according to a preset threshold trigger strategy to obtain a target identification strategy;

[0101] Authenticate the user according to the target identification strategy, and unlock the door when the authentication is passed;

[0102] The target recognition strategy includes any one or more of target fingerprint recognition, target face recognition, target biomarker recognition, target palm vein recognition and target finger vein recognition.

[0103] The method further includes obtaining a current timestamp, selecting a target recognition strategy from the modal biometric library based on the timestamp, the biometric recognition weight, and the environmental parameters, performing identity authentication on the user based on the target recognition strategy, and unlocking the door after the authentication is passed. The method further includes:

[0104] Acquire a data set of a smart door lock, preprocess the data set to obtain a target data set, analyze the target recognition strategy based on the target data set and a preset time series model, and obtain an optimal recognition strategy;

[0105] The data set includes the number of unlocking times, biometric identification method and unlocking duration.

[0106] The step of obtaining a data set of a smart door lock, preprocessing the data set to obtain a target data set, and analyzing the target recognition strategy based on the target data set and the preset time series model to obtain an optimal recognition strategy specifically includes:

[0107] Obtaining the number of unlocks, the biometric identification method, and the unlocking duration of the smart door lock, preprocessing the number of unlocks, the biometric identification method, and the unlocking duration to obtain a target data set, and training a preset time series model based on the target data set to obtain a neural network model;

[0108] Analyzing the target recognition strategy through the neural network model to obtain an optimal recognition strategy;

[0109] Wherein, the preprocessing includes outlier processing and missing value filling processing;

[0110] The preset timing model includes TCN or LSTM.

[0111] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a multimodal biometric adaptive recognition program, and when the multimodal biometric adaptive recognition program is executed by a processor, the steps of the multimodal biometric adaptive recognition method described above are implemented.

[0112] In summary, the present invention provides a multimodal biometric adaptive recognition method, system, smart door lock, and storage medium. The method includes: obtaining a user's biometrics, performing a confidence query in a pre-stored multimodal biometric library based on the biometrics to obtain an initial confidence score; obtaining environmental parameters from a sensor, adjusting the initial confidence score based on the environmental parameters to obtain multiple environmental weights, performing weight calculation based on all environmental weights to obtain a biometric weight; obtaining a current timestamp, selecting a target recognition strategy from the modal biometric library based on the timestamp, the biometric weight, and the environmental parameters, and authenticating the user according to the target recognition strategy, and unlocking the door after the authentication is successful. The present invention improves recognition accuracy, reduces failure rate, and enhances security and user experience by collecting biometrics, adjusting confidence based on the environment, selecting a strategy for authentication, and unlocking the door after successful authentication.

[0113] It should be noted that, in this document, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or smart door lock system that includes a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or smart door lock system. In the absence of further restrictions, an element defined by the sentence "comprises a..." does not exclude the presence of other identical elements in the process, method, article or smart door lock system that includes the element.

[0114] Of course, those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing related hardware (such as a processor, controller, etc.) through a computer program. The program can be stored in a computer-readable storage medium that can be read by a computer. When the program is executed, it can include the processes in the above-described method embodiments. The computer-readable storage medium can be a memory, a magnetic disk, an optical disk, etc.

[0115] It should be understood that the application of the present invention is not limited to the above examples. For those skilled in the art, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A multimodal biometric adaptive recognition method, characterized in that: The multimodal biometric adaptive recognition method comprises: Obtaining a user's biometric features, and performing a confidence query on a pre-stored multimodal biometric feature database based on the biometric features to obtain an initial confidence score; Acquiring environmental parameters from a sensor, adjusting the initial confidence score according to the environmental parameters to obtain a plurality of environmental weights, and performing weight calculation based on all the environmental weights to obtain a biometric recognition weight; Obtain the current timestamp, select a target recognition strategy from the modal biometric library according to the timestamp, the biometric recognition weight and the environmental parameters, authenticate the user according to the target recognition strategy and unlock the door after the authentication is passed.

2. The multimodal biometric adaptive recognition method according to claim 1, characterized in that: The biometric feature includes any one of fingerprint feature, facial feature, palm vein feature, finger vein feature and voiceprint feature; The step of obtaining the user's biometric features and querying the confidence of a pre-stored multimodal biometric feature library based on the biometric features to obtain an initial confidence score specifically includes: Receiving a biometric feature of a user acquired by a sensor, and performing a confidence query in a pre-stored multimodal biometric feature library based on the fingerprint feature, the facial feature, the palm vein feature, the finger vein feature, or the voiceprint feature to obtain an initial confidence score; The initial confidence score is used by the smart door lock to adjust according to different influencing information.

3. The multimodal biometric adaptive recognition method according to claim 1, characterized in that: The environmental parameters include any one or more of light intensity, humidity and temperature; The step of acquiring environmental parameters from the sensor, adjusting the initial confidence score according to the environmental parameters to obtain a plurality of environmental weights, and performing weight calculation according to all the environmental weights to obtain a biometric recognition weight specifically includes: Obtaining any one or more of the light intensity, the humidity, and the temperature collected by a sensor, and adjusting the initial confidence score according to any one or more of the light intensity, the humidity, and the temperature to obtain multiple environmental weights; A weight calculation is performed based on all the environmental weights to obtain a plurality of weight scores, and all the weight scores are sorted in order from high to low to obtain a biometric weight.

4. The multimodal biometric adaptive recognition method according to claim 3, characterized in that: The weight calculation is performed based on all the environmental weights to obtain multiple weight scores, and all the weight scores are sorted in sequence to obtain the biometric weight, specifically including: Performing matching calculation on the biometric features based on a biometric recognition algorithm to obtain similarity; Performing weight calculation based on the environmental adaptability coefficient in the sensor, all the environmental weights, and the similarity to obtain a plurality of weight scores, and sequentially sorting all the weight scores to obtain a target weight; A weighted calculation is performed according to the environmental adaptation coefficient, the target weight and the similarity to obtain a biometric recognition weight.

5. The multimodal biometric adaptive recognition method according to claim 3, characterized in that: The modal biometric library includes fingerprint recognition, face recognition, biomarker recognition, palm vein recognition and finger vein recognition; The obtaining of the current timestamp, selecting a target recognition strategy from the modal biometric library according to the timestamp, the biometric recognition weight, and the environmental parameters, and authenticating the user according to the target recognition strategy and unlocking the door after the authentication is passed, specifically includes: Obtaining a current timestamp, and dynamically selecting among fingerprint recognition, face recognition, biometric recognition, palm vein recognition, and finger vein recognition based on the timestamp and the biometric weight, as well as the light intensity, humidity, or temperature, to obtain multiple recognition strategies; Screening the plurality of identification strategies according to a preset threshold trigger strategy to obtain a target identification strategy; Authenticate the user according to the target identification strategy, and unlock the door when the authentication is passed; The target recognition strategy includes any one or more of target fingerprint recognition, target face recognition, target biomarker recognition, target palm vein recognition and target finger vein recognition.

6. The multimodal biometric adaptive recognition method according to claim 1, characterized in that: Obtaining a current timestamp, selecting a target recognition strategy from the modal biometric library based on the timestamp, the biometric recognition weight, and the environmental parameters, performing identity authentication on the user based on the target recognition strategy, and unlocking the door after the authentication is passed, and then further comprising: Acquire a data set of a smart door lock, preprocess the data set to obtain a target data set, analyze the target recognition strategy based on the target data set and a preset time series model, and obtain an optimal recognition strategy; The data set includes the number of unlocking times, biometric identification method and unlocking duration.

7. The multimodal biometric adaptive recognition method according to claim 6, characterized in that: The step of obtaining a data set of the smart door lock, preprocessing the data set to obtain a target data set, and analyzing the target recognition strategy based on the target data set and the preset time series model to obtain an optimal recognition strategy specifically includes: Obtaining the number of unlocks, the biometric identification method, and the unlocking duration of the smart door lock, preprocessing the number of unlocks, the biometric identification method, and the unlocking duration to obtain a target data set, and training a preset time series model based on the target data set to obtain a neural network model; Analyzing the target recognition strategy through the neural network model to obtain an optimal recognition strategy; Wherein, the preprocessing includes outlier processing and missing value filling processing; The preset timing model includes TCN or LSTM.

8. A multimodal biometric adaptive recognition system, characterized in that: The multimodal biometric adaptive recognition system comprises: An initial confidence acquisition module is used to obtain the user's biometric features, perform a confidence query on the pre-stored multimodal biometric feature library based on the biometric features, and obtain an initial confidence score; a biometric weight calculation module, configured to obtain environmental parameters from a sensor, adjust the initial confidence score according to the environmental parameters to obtain a plurality of environmental weights, and perform weight calculation based on all the environmental weights to obtain a biometric weight; The target recognition strategy identification module is used to obtain the current timestamp, select the target recognition strategy from the modal biometric library according to the timestamp, the biometric weight and the environmental parameters, and authenticate the user according to the target recognition strategy and unlock the door after the authentication is passed.

9. A smart door lock, characterized in that: The smart door lock includes: a memory, a processor, and a multimodal biometric adaptive recognition program stored in the memory and runnable on the processor. When the multimodal biometric adaptive recognition program is executed by the processor, the steps of the multimodal biometric adaptive recognition method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a multimodal biometric adaptive recognition program, which, when executed by a processor, implements the steps of the multimodal biometric adaptive recognition method according to any one of claims 1 to 7.

Images