CN-122027341-A - Alarm operation control method and system based on face authentication

CN122027341ACN 122027341 ACN122027341 ACN 122027341ACN-122027341-A

Abstract

The application discloses an alarm operation control method and system based on face authentication, relating to the technical field of intelligent security, the method combines the facial features of an operator with the real-time emotion state to carry out identity verification and operation intention analysis through a high-precision face recognition technology and living body detection. And (3) evaluating the behavior mode of the operator and the state of the alarm host computer in real time through a deep learning model and a machine learning algorithm, and dynamically adjusting the authority level and the risk level of the operator. And generating an encryption control instruction by adopting a credibility-risk matching matrix and an intelligent contract technology, and authorizing an operator to execute high-risk operation. The contract triggering condition is based on the identity credibility and the operation risk of an operator, generates an encryption instruction after multi-node signature confirmation, and sends the encryption instruction to the alarm host executing module through double-layer binding encryption encapsulation. The method effectively avoids the limitation of a single verification mode, and improves the safety, the instantaneity and the tamper resistance of the system.

Inventors

FAN BING
JIN SHUGUANG
ZHANG XI
HAN XU

Assignees

北京天河地塬安防技术服务有限公司

Dates

Publication Date: 20260512
Application Date: 20260325

Claims (10)

1. The alarm operation control method based on the face authentication is characterized by comprising the following steps: acquiring facial images of an operator through a high-precision face recognition technology and performing living body detection; Collecting facial features and real-time emotional states of the operator if the operator is determined to be a living body; analyzing identity validity results and operation intention rationality results of the facial features and the real-time emotion states through a deep learning model; under the condition that the identity legitimacy result of the operator and the operational intention rationality result are qualified, combining the real-time state and the historical operation data of the alarm host, and carrying out real-time evaluation on the behavior mode of the operator based on a dynamic behavior analysis model of machine learning so as to adjust the operational authority and the risk level of the operator; And under the condition that the identity credibility of the operator is matched with the current operation risk of the alarm host, generating an encryption control instruction through an intelligent contract, and authorizing the operator to execute high-risk operation.
2. The alarm operation control method based on face authentication according to claim 1, wherein the capturing face images of an operator and performing living body detection by a high-precision face recognition technique comprises: Acquiring a face image of an operator through an integrated three-dimensional face recognition technology, and acquiring three-dimensional face data of the operator; Acquiring facial thermal imaging data of the operator by an infrared thermal imaging technique; And based on the three-dimensional face data and the face thermal imaging data, performing living body judgment processing by adopting a fusion algorithm, and outputting a living body judgment result.
3. The face authentication-based alarm operation control method according to claim 1, wherein the analyzing the identity validity result and the operation intention rationality result thereof by a deep learning model according to the facial features and the real-time emotional state comprises: Carrying out time sequence coding processing on the facial features and the real-time emotion states, constructing an emotion time sequence feature sequence, and constructing an emotion baseline model based on historical authentication data; Carrying out offset calculation on the emotion time sequence characteristic sequence and the emotion baseline model to obtain emotion offset parameters; Inputting the emotion time sequence characteristic sequence and the emotion offset parameter into a deep learning model for identity consistency judgment processing, and outputting an identity validity result; And constructing a layered risk mapping matrix based on the emotion time sequence characteristic sequence and the emotion offset parameter, comparing the layered risk mapping matrix with a preset operation risk threshold value, and outputting an operation intention rationality result.
4. The method for controlling alarm operation based on face authentication according to claim 3, wherein inputting the emotion timing feature sequence and the emotion offset parameter into a deep learning model for identity consistency determination processing, outputting an identity validity result, comprises: carrying out multi-scale time window segmentation processing on the emotion time sequence characteristic sequence to construct a segmented emotion characteristic subsequence; calculating an emotion state transition vector based on the segmented emotion feature subsequence; calculating individual behavior stability coefficients according to the emotion offset parameters; performing feature fusion processing on the segmented emotion feature subsequence, the emotion state transition vector and the individual behavior stability coefficient to form a joint feature vector; Inputting the combined feature vector into a dual-channel consistency judging network for identity consistency modeling processing, and outputting an identity matching score; And performing interval mapping processing on the identity matching score and a preset identity judgment threshold interval, and outputting the identity validity result.
5. The alarm operation control method based on face authentication according to claim 3, wherein the constructing a hierarchical risk mapping matrix based on the emotion timing sequence and the emotion offset parameter, comparing the hierarchical risk mapping matrix with a preset operation risk threshold, and outputting an operation intention rationality result includes: Calculating an emotion fluctuation gradient vector according to the emotion time sequence characteristic sequence; Calculating a risk offset grade parameter according to the emotion offset parameter; Carrying out dynamic weight distribution processing on the mood fluctuation gradient vector and the risk offset grade parameter to generate a risk weight vector; Constructing a multidimensional risk mapping matrix based on the risk weight vector; Carrying out hierarchical mapping treatment on the multidimensional risk mapping matrix to obtain risk grade parameters; And performing interval comparison processing on the risk level parameter and the preset operation risk threshold interval, and outputting the operation intention rationality result.
6. The method for controlling alarm operation based on face authentication according to claim 1, wherein said combining real-time status and historical operation data of said alarm host machine, based on a machine-learned dynamic behavior analysis model, evaluates the behavior pattern of said operator in real time to adjust the operation authority and risk level thereof, comprises: acquiring real-time state parameters and historical operation data of the alarm host, and constructing a host state feature vector; collecting current operation behavior data of the operator, and constructing a behavior track feature sequence; performing space-time alignment processing on the host state feature vector and the behavior track feature sequence to generate a behavior state joint feature matrix; constructing a behavior offset vector based on the behavior state joint feature matrix, and calculating a behavior anomaly gradient parameter; Inputting the behavior state joint feature matrix, the behavior offset vector and the behavior anomaly gradient parameters into a dynamic behavior analysis model for risk scoring processing, and outputting behavior risk scores; and generating a risk level parameter according to the behavior risk score, and adjusting the operation authority level of the operator based on the risk level parameter.
7. The method for controlling alarm operation based on face authentication according to claim 6, wherein inputting the behavior state joint feature matrix, the behavior shift vector and the behavior anomaly gradient parameters into a dynamic behavior analysis model for risk scoring processing, outputting a behavior risk score, comprises: Weighting the time sequence behavior patterns in the behavior state joint feature matrix, and dynamically adjusting the weight coefficient of the time sequence behavior patterns based on an adaptive weight distribution mechanism, wherein the adaptive weight distribution mechanism dynamically adjusts feature weights by calculating the change rate and importance evaluation of each feature so as to give priority to the behavior patterns related to high-risk operation; and inputting the weighted behavior state joint feature matrix, the behavior offset vector and the behavior anomaly gradient parameter into a deep reinforcement learning model for risk assessment processing, and outputting a behavior risk score, wherein the deep reinforcement learning model performs reinforcement learning on historical behavior data, optimizes weight distribution in real time, and adjusts a behavior assessment strategy based on a reward function in a self-adaptive manner.
8. The face authentication-based alarm operation control method according to claim 6, wherein the generating a risk level parameter according to the behavioral risk score and adjusting the operation authority level of the operator based on the risk level parameter comprises: Performing interval mapping processing according to the behavior risk score and a preset operation risk threshold interval to generate a risk level parameter and a corresponding risk weight; Fusing the risk weight with the operation history data and the identity authentication result of the operator to generate a permission adjustment factor; invoking a permission mapping table according to the risk level parameter to generate a target permission level; Performing difference calculation on the target authority level and the current authority level to generate an authority transition instruction; And executing authority freezing processing when the authority transition instruction is a degradation instruction, executing delay confirmation processing when the authority transition instruction is an upgrading instruction, and updating the operation authority level of the operator according to the authority transition instruction.
9. The face authentication-based alarm operation control method according to claim 1, wherein the generating an encryption control instruction by an intelligent contract in a case where the identity credibility of the operator matches the current operation risk of the alarm host, authorizing the operator to perform a high risk operation, comprises: constructing a credibility-risk matching matrix according to the identity credibility and the current operation risk, and generating contract trigger condition parameters; writing the contract trigger condition parameters into an intelligent contract template to generate a contract execution request; Performing multi-node signature processing on the contract execution request to generate a contract confirmation identifier; generating an encryption control instruction structure body according to the contract confirmation mark, wherein the encryption control instruction structure body comprises an operation type parameter, a permission level parameter and an aging control parameter; And carrying out encryption packaging processing on the encryption control instruction structure body, and sending the encryption control instruction structure body to the alarm host executing module.
10. The alarm operation control method based on face authentication, which is used for the alarm operation control method based on face authentication as claimed in any one of claims 1 to 9, is characterized by comprising the following steps: a living body detection unit for acquiring a face image of an operator through a high-precision face recognition technique and performing living body detection; a data acquisition unit configured to acquire facial features and a real-time emotional state of the operator in a case where the operator is determined to be a living body; A result analysis unit for analyzing an identity validity result and an operation intention rationality result thereof through a deep learning model according to the facial features and the real-time emotional state; The evaluation and adjustment unit is used for carrying out real-time evaluation on the behavior mode of the operator based on a dynamic behavior analysis model of machine learning by combining the real-time state and historical operation data of the alarm host under the condition that the identity legitimacy result of the operator and the rationality result of the operation intention are qualified so as to adjust the operation authority and the risk level of the operator; and the encryption authorization unit is used for generating an encryption control instruction through an intelligent contract under the condition that the identity credibility of the operator is matched with the current operation risk of the alarm host, and authorizing the operator to execute high-risk operation.

Description

Alarm operation control method and system based on face authentication Technical Field The application relates to the technical field of intelligent security, in particular to an alarm operation control method and system based on face authentication. Background With the wide application of security systems in residential communities, financial institutions, industrial parks and smart cities, an alarm host serves as a core control device and plays key roles in arming, disarming, bypass setting, alarm confirmation and the like. The operation generally has higher safety sensitivity, and once the operation is mishandled or maliciously operated by unauthorized personnel, the protection failure, alarm false alarm or safety loophole expansion can be caused, so that serious property loss or safety risk is brought. The existing alarm operation control mode depends on password input, card swiping verification or a simple face recognition authentication mechanism, but the traditional password mode has leakage risk, the card swiping mode has hidden danger of losing or copying, and the single face recognition mode can be bypassed when facing photo, video or deep counterfeiting attack. In addition, most of the existing systems only perform one-time judgment on the identity verification level, lack comprehensive analysis on the emotional state, the behavior mode and the current operation risk of an operator, and fail to establish a dynamic matching mechanism between identity credibility and operation risk. Meanwhile, in the aspect of high-risk operation authorization, a centralized control strategy is generally adopted in the prior art, an execution instruction is directly generated by a single control module, a hierarchical authorization, distributed confirmation and structured encryption control mechanism is lacked, and traceable, auditable and verifiable operation closed loop is difficult to realize. Therefore, how to construct a control method combining multidimensional identity verification, behavior dynamic evaluation and intelligent contract authorization in the high-risk operation process of an alarm host so as to improve the system safety and operation controllability becomes a problem to be solved urgently. Disclosure of Invention In the summary, a series of concepts in a simplified form are introduced, which will be further described in detail in the detailed description. The summary of the application is not intended to define the key features and essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter. In a first aspect, the present application provides a method for controlling an alarm operation based on face authentication, including: acquiring facial images of an operator through a high-precision face recognition technology and performing living body detection; Collecting facial features and real-time emotional states of the operator when the operator is determined to be a living body; analyzing identity validity results and operation intention rationality results of the facial features and the real-time emotion states through a deep learning model; Under the condition that the identity legitimacy result of the operator and the rationality result of the operation intention are qualified, the real-time evaluation is carried out on the behavior mode of the operator based on a dynamic behavior analysis model of machine learning by combining the real-time state and historical operation data of the alarm host so as to adjust the operation authority and risk level of the operator; And under the condition that the identity credibility of the operator is matched with the current operation risk of the alarm host, generating an encryption control instruction through an intelligent contract, and authorizing the operator to execute high-risk operation. In one possible embodiment, the capturing the face image of the operator and performing the living body detection by the high-precision face recognition technology includes: acquiring face images of operators through an integrated three-dimensional face recognition technology, and acquiring three-dimensional face data of the operators; Acquiring facial thermal imaging data of the operator through an infrared thermal imaging technology; based on the three-dimensional face data and the face thermal imaging data, a fusion algorithm is adopted to perform living body judgment processing, and a living body judgment result is output. In a possible embodiment, the analyzing the identity validity result and the operation intention rationality result according to the facial features and the real-time emotion state through a deep learning model includes: carrying out time sequence coding processing on the facial features and the real-time emotion states, constructing an emotion time sequence feature sequence, and constructing an emotion baseline model based on historical authentication data; C