A smart security access control system

Abstract

This invention discloses a smart security access control system, relating to the field of access control technology. Key technical points include: acquiring dynamic behavior feature clusters one, two, and three of a target object within the access control monitoring area in a baseline scenario, a disturbance scenario, and a strong disturbance scenario, respectively; judging dynamic behavior feature clusters one and two to obtain a first judgment result; correcting the dynamic behavior feature cluster two of the target object in the disturbance scenario based on the first judgment result to obtain a corrected dynamic behavior feature cluster; judging dynamic behavior feature clusters one, the corrected dynamic behavior feature cluster, and the third dynamic behavior feature cluster to obtain a second judgment result; if the second judgment result clearly indicates that the target object is in an authorized access state, detecting the first scenario transition time and the first behavior feature adaptation increment between the baseline scenario and the disturbance scenario, resulting in a continuous and stable access control function.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of access control technology, more particularly, it relates to a smart security access control system. BACKGROUND

[0002] In the field of smart security, traditional access control systems rely on identity recognition technology in a single scene, such as card swiping, fingerprint recognition, or simple face recognition. However, there are many interference factors in the real environment, such as changes in environmental light and shadow, sound wave interference, and the distribution of obstructions. These interferences can cause the recognition accuracy of traditional access control systems to drop significantly. For example, face recognition access control may not be able to accurately capture facial features under strong light; at the same time, traditional access control systems lack the ability to distinguish and adapt to different interference scenes, and cannot dynamically adjust the recognition strategy according to the changes in the intensity of the scene interference, thereby affecting the security and efficiency of the access control system. SUMMARY

[0003] In view of the deficiencies in the prior art, the purpose of the present application is to provide a smart security access control system.

[0004] To achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0005] A smart security access control system, comprising:

[0006] An acquisition module: acquiring dynamic behavior feature clusters one, two and three of a target object in an access control monitoring area in a reference scene, a disturbance scene and a strong disturbance scene, respectively;

[0007] A judgment and correction module: judging the dynamic behavior feature clusters one and two to obtain a first judgment result; and correcting the dynamic behavior feature cluster two of the target object in the disturbance scene according to the first judgment result to obtain a dynamic behavior correction feature cluster;

[0008] A processing module: judging the dynamic behavior feature clusters one, two and three to obtain a second judgment result;

[0009] A detection module: if the second judgment result clearly indicates that the target object is in an authorized access state, detecting a first scene transition duration and a first behavior feature adaptation increment between the reference scene and the disturbance scene, and detecting a second scene transition duration and a second behavior feature adaptation increment between the disturbance scene and the strong disturbance scene, and collecting behavior feature extraction parameters of the target object in the disturbance scene;

[0010] An output module: processing the behavior feature extraction parameters, the first scene transition duration, the first behavior feature adaptation increment, the second scene transition duration and the second behavior feature adaptation increment to obtain an access control result.

[0011] Preferably, the first dynamic behavior feature cluster and the second dynamic behavior feature cluster are judged to obtain a first judgment result; the second dynamic behavior feature cluster of the target object in the disturbed scene is corrected according to the first judgment result to obtain a dynamic behavior correction feature cluster, specifically including the following steps:

[0012] If the scene adaptability of both dynamic behavior feature cluster one and dynamic behavior feature cluster two is greater than or equal to the first preset adaptability threshold, then the target object is determined to be in the pass authorization state.

[0013] If the scene adaptability of dynamic behavior feature cluster one or / and dynamic behavior feature cluster two is less than the first preset adaptability threshold, then the target object is determined to be in a non-authorized state.

[0014] If the target object is in an unauthorized state, collect the behavioral association area between the target object and the access control execution unit in the disturbance scenario, and divide the behavioral association area into association partition sets according to the distribution of interference factors;

[0015] The interference influence coefficient between the behavioral feature error rate and the scene interference change value in the historical period is extracted. Based on the interference influence coefficient, the dynamic behavioral feature cluster 2 of the target object in the disturbed scene is corrected to obtain the dynamic behavioral correction feature cluster.

[0016] Preferably, the method further includes the following steps:

[0017] The baseline scenario is a scenario of normal operation without interference;

[0018] The system collects the disturbance intensity of the operating scenario and compares the disturbance intensity with a preset disturbance intensity threshold.

[0019] If the disturbance intensity is less than or equal to the preset disturbance intensity threshold, the running scenario will be marked as a disturbance scenario;

[0020] If the disturbance intensity is greater than the preset disturbance intensity threshold, the running scenario will be marked as a strong disturbance scenario.

[0021] Preferably, the behavioral association region is divided into association partition sets based on the distribution trend of interference factors, specifically including the following steps:

[0022] Collect interference elements of the disturbed scene; wherein, the interference elements include changes in ambient light and shadow, sound wave interference, and distribution of obstructions;

[0023] The system senses the source, propagation trajectory, and intensity of interference elements, and determines the distribution pattern of interference factors within the access control monitoring area based on these parameters.

[0024] The related regions are divided according to the distribution of interference factors to obtain the related partition set.

[0025] Preferably, the dynamic behavior feature cluster of the target object in the disturbed scene is corrected according to the interference influence coefficient to obtain the dynamic behavior correction feature cluster, which specifically includes the following steps:

[0026] The association feature dataset is obtained by statistically analyzing the association range and behavioral interaction frequency of each association partition in the association partition set;

[0027] Based on the interference impact coefficient, the associated feature dataset, and the scene interference change values ​​of each associated partition in the associated partition set, the feature extraction degree of the associated partition set is determined to obtain the feature extraction degree value.

[0028] The dynamic behavior feature cluster 2 is modified based on the feature extraction degree value to obtain the dynamic behavior modified feature cluster.

[0029] Preferably, the second judgment result is obtained by judging the dynamic behavior feature cluster one, the dynamic behavior correction feature cluster, and the dynamic behavior feature cluster three, specifically including the following steps:

[0030] If the scene adaptability of dynamic behavior feature cluster 1, dynamic behavior correction feature cluster and dynamic behavior feature cluster 3 are all higher than the second preset adaptability threshold, then the target object is confirmed to be in the pass authorization state.

[0031] If any one of the scene adaptability of dynamic behavior feature cluster one, dynamic behavior correction feature cluster and dynamic behavior feature cluster three is not higher than the second preset adaptability threshold, then the target object is confirmed to be in a non-authorized state.

[0032] Preferably, the access control result is obtained by processing the behavior feature extraction parameters, the first scene transition time, the first behavior feature adaptation increment, the second scene transition time, and the second behavior feature adaptation increment, specifically including the following steps:

[0033] The first scene transition time and the first behavior feature are incrementally combined to form the first scene feature; the second scene transition time and the second behavior feature are incrementally combined to form the second scene feature.

[0034] The historical scene features and historical feature extraction accuracy of the target object in a historical period are obtained, and the influence between the historical scene features and historical feature extraction accuracy is evaluated to obtain the accuracy influence coefficient.

[0035] The first scene features, accuracy impact coefficient, and behavioral feature extraction parameters are processed to obtain the first dynamic behavioral data;

[0036] Collect the behavioral characteristic parameters of the target object in a strong disturbance scenario, and process the second scenario features, accuracy influence coefficient and behavioral characteristic parameters to obtain the second dynamic behavior data;

[0037] The access control result is obtained by confirming the access authorization status of the target object based on the first dynamic behavior data and the second dynamic behavior data.

[0038] Preferably, the first scene features, accuracy impact coefficient, and behavioral feature extraction parameters are processed to obtain the first dynamic behavioral data, specifically including the following steps:

[0039] The first influence coefficient is extracted from the accuracy influence coefficient based on the first scene features. The perturbation scene extraction parameters are obtained by calibrating the behavior feature extraction parameters based on the first scene features and the first influence coefficient.

[0040] The first dynamic behavior data of the target object in the disturbance scene is collected based on the parameters extracted from the disturbance scene.

[0041] Preferably, the behavioral characteristic parameters of the target object in a strongly disturbed scenario are collected, and the second scenario features, accuracy impact coefficient, and behavioral characteristic parameters are processed to obtain the second dynamic behavioral data. Specifically, this includes the following steps:

[0042] The second influence coefficient is extracted from the accuracy influence coefficient based on the second scene features. The behavioral feature status parameters are then calibrated based on the second scene features and the second influence coefficient to obtain the strong disturbance scene extraction parameters.

[0043] The second dynamic behavior data of the target object in the strong disturbance scene is collected based on the parameters extracted from the strong disturbance scene.

[0044] Preferably, the distribution pattern of the interference factors includes the spatial distribution and temporal variation of the interference elements within the access control monitoring area.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] This invention collects dynamic behavioral feature clusters in a baseline scenario, a disturbed scenario, and a strongly disturbed scenario, achieving comprehensive capture of the target object's behavioral patterns under different interference environments. By judging and correcting the first two types of feature clusters, the misleading effect of interference on behavioral features in the disturbed scenario is eliminated. Combining the feature clusters of the three scenarios for a second judgment improves the accuracy of access authorization status judgment. Furthermore, in-depth analysis and feature correction are performed on interference elements in the disturbed and strongly disturbed scenarios. The judgment and correction module divides the associated partitions according to the distribution of interference factors and corrects the feature clusters based on the interference influence coefficient, enabling the system to stably extract effective behavioral features even in complex interference environments with changes in light and shadow, sound wave interference, and obstruction distribution. By calibrating the feature extraction parameters of the current scenario using the accuracy influence coefficient of historical data, the system can adapt to changes in the behavioral patterns of different target objects and dynamic adjustments to the interference levels of different scenarios. Therefore, this application can maintain good adaptability through its own calibration mechanism and continuously and stably perform access control functions. Attached Figure Description

[0047] Fig. 1 This invention presents a schematic diagram of a smart security access control system module;

[0048] Fig. 2 This invention provides a schematic diagram illustrating the steps involved in obtaining dynamic behavior correction feature clusters in a smart security access control system. Detailed Implementation

[0049] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0050] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0051] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0052] Reference Figs. 1-2 As shown.

[0053] The embodiments further illustrate the intelligent security access control system proposed in this invention.

[0054] A smart security access control system includes:

[0055] Acquisition module: Acquires dynamic behavior feature cluster 1, dynamic behavior feature cluster 2, and dynamic behavior feature cluster 3 of the target object within the access control monitoring area in the baseline scenario, disturbance scenario, and strong disturbance scenario, respectively;

[0056] Dynamic behavioral feature clusters one, two, and three all include the target object's gait characteristics, limb movement characteristics, and facial features. Gait characteristics include stride length, stride frequency, and gait cycle; limb movement characteristics include arm swing amplitude and body rotation angle; and facial features include facial contour and the positional relationship of facial features.

[0057] Judgment and correction module: The dynamic behavior feature cluster one and dynamic behavior feature cluster two are judged to obtain the first judgment result; Based on the first judgment result, the dynamic behavior feature cluster two of the target object in the disturbance scene is corrected to obtain the dynamic behavior correction feature cluster;

[0058] Processing module: The second judgment result is obtained by judging the dynamic behavior feature cluster one, the dynamic behavior correction feature cluster, and the dynamic behavior feature cluster three;

[0059] Detection module: If the second judgment result clearly indicates that the target object is in the pass authorization state, detect the first scene transition time and the first behavior feature adaptation increment between the baseline scene and the disturbance scene, and detect the second scene transition time and the second behavior feature adaptation increment between the disturbance scene and the strong disturbance scene, and collect the behavior feature extraction parameters of the target object in the disturbance scene;

[0060] The detection measures the first scene transition time between the baseline scene and the disturbed scene. The first scene transition time refers to the time it takes for the target object to move from the baseline scene to the disturbed scene. For example, it is the time it takes for the target object to walk from an undisturbed indoor hall (baseline scene) to a corridor with slight changes in lighting (disturbed scene). Simultaneously, the first behavioral feature adaptation increment is detected. This first behavioral feature adaptation increment measures the change in the target object's behavioral features from dynamic behavioral feature cluster one to dynamic behavioral feature cluster two. Taking gait stride length as an example, if the stride length in the baseline scene is... In the perturbation scenario, the step size is Then the first row of feature adaptation increment (step size dimension) is represented as: Other behavioral characteristics, such as cadence and range of motion, will also be calculated accordingly. The increments of these different feature dimensions will be combined to form the first behavioral feature adaptation increment.

[0061] Next, the transition time between the disturbed scene and the strongly disturbed scene is detected, which is the time it takes for the target object to move from the disturbed scene to the strongly disturbed scene. For example, the time it takes to walk from a corridor with slight changes in lighting (disturbed scene) to a device room with strong sound interference (strong disturbed scene) is set as... Simultaneously, the second behavioral feature adaptation increment is detected. This second behavioral feature adaptation increment is the change in the target object's behavioral features when adapting from the dynamic behavioral correction feature cluster to the third dynamic behavioral feature cluster. Again, taking the step size as an example, if the step size in the dynamic behavioral correction feature cluster is... In strong disturbance scenarios, the step size is Then the second behavior feature adaptation increment (step size dimension) is expressed as: Other behavioral feature dimensions are also calculated in the same way to obtain the second behavioral feature adaptation increment.

[0062] The parameters for extracting behavioral features of the target object in a disturbed scene include settings and metrics for extracting behavioral features of the target object in a disturbed scene, such as the sampling frequency and the range of the feature extraction area. For example, when extracting gait features of the target object in a disturbed scene, the sampling frequency is set to n times per second, and the extraction area is the range of the limb from the ankle to the waist. These all fall under the category of behavioral feature extraction parameters.

[0063] Output module: Processes the behavior feature extraction parameters, the first scene transition time, the first behavior feature adaptation increment, the second scene transition time, and the second behavior feature adaptation increment to obtain the access control result.

[0064] The first judgment result is obtained by judging the dynamic behavior feature cluster 1 and the dynamic behavior feature cluster 2; the dynamic behavior feature cluster 2 of the target object in the disturbed scene is corrected according to the first judgment result to obtain the dynamic behavior correction feature cluster, which specifically includes the following steps:

[0065] If the scene adaptability of both dynamic behavior feature cluster one and dynamic behavior feature cluster two is greater than or equal to the first preset adaptability threshold, it indicates that the target object's behavior features match the scene well in both the baseline scene and the disturbed scene, and therefore the target object is determined to be in a pass-authorization state. For example, if the scene adaptability of the authorized person's gait stride length, stride frequency, and limb movement amplitude behavior features reaches or exceeds the first preset adaptability threshold in both the baseline scene without interference and the disturbed scene with slight interference from people walking, then the target object is determined to be in a pass-authorization state.

[0066] If the scene adaptability of dynamic behavior feature cluster one and / or dynamic behavior feature cluster two is less than the first preset adaptability threshold, the target object is determined to be in an unauthorized state. This indicates that the target object's behavior features do not match the scene well in the baseline or disturbed scene, thus determining the target object to be in an unauthorized state. For example, in the baseline scene, an unauthorized person may have a low scene adaptability of gait due to unfamiliarity with the environment, or in the disturbed scene, their body movements may have a scene adaptability lower than the threshold due to tension. In this case, they are determined to be in an unauthorized state.

[0067] If the target object is in an unauthorized state, collect the behavioral association area between the target object and the access control execution unit in a disturbance scenario.

[0068] When a target object is determined to be in an unauthorized state, it is necessary to collect the behavioral association area between the target object and the access control unit in a disturbed scenario. The behavioral association area is the area where the target object and the access control unit (such as an access control camera or card reader) interact, for example, the area where the target object attempts to swipe its card in front of the access control or lingers in front of the camera.

[0069] The related regions are divided according to the distribution of interference factors to obtain a set of related partitions;

[0070] The interference influence coefficient between the behavioral feature error rate and the scene interference change value in the historical period is extracted. Based on the interference influence coefficient, the dynamic behavioral feature cluster 2 of the target object in the disturbed scene is corrected to obtain the dynamic behavioral correction feature cluster.

[0071] It also includes the following steps:

[0072] The baseline scenario is a scenario of normal operation without interference;

[0073] The system collects the disturbance intensity of the operating scenario and compares the disturbance intensity with a preset disturbance intensity threshold.

[0074] If the disturbance intensity is less than or equal to the preset disturbance intensity threshold, the running scenario will be marked as a disturbance scenario;

[0075] If the disturbance intensity is greater than the preset disturbance intensity threshold, the running scenario will be marked as a strong disturbance scenario.

[0076] The baseline scenario is a non-interference normal operation scenario, such as a quiet office area entrance with no changes in light and shadow, no obstructions and no additional sound wave interference. In this scenario, the environment within the access control monitoring area is stable, making it easy to obtain the most original dynamic behavior characteristics of the target object.

[0077] The data acquisition process assesses the intensity of disturbances occurring in the scene. Disturbance intensity is a comprehensive indicator of the degree of interference in the scene. Interference factors include changes in ambient light and shadow (such as fluctuations in light intensity, the speed and range of light and shadow movement), acoustic interference (such as the decibel level of noise and the frequency range of sound waves), and the distribution of obstructions (such as the number, size, and movement of obstructions). Assuming the disturbance intensity is... The preset disturbance intensity threshold is .when At this time, the operating scenario is marked as a disturbed scenario. For example, if there is a slight swaying of tree shadows outside the window at the entrance of the office area, or low-decibel background conversation, the disturbance intensity is at a low level and it is considered a disturbed scenario.

[0078] when When a scenario is marked as a strong disturbance, such as a brightly flashing light at the entrance of an office area, high-decibel equipment noise, or a large amount of temporary storage obstructing the access control monitoring area, the disturbance intensity exceeds the preset threshold and is considered a strong disturbance scenario.

[0079] This scenario segmentation method can clearly identify scenario types with different levels of interference, thereby providing an accurate scenario background for subsequent acquisition of dynamic behavioral feature clusters of target objects in different scenarios, feature judgment and correction operations, and ensuring that the access control system can accurately identify the access authorization status of target objects in various interference environments.

[0080] Based on the distribution pattern of interference factors, the behavioral association regions are divided into association partition sets, which specifically include the following steps:

[0081] Collect interference elements of the disturbed scene; among them, interference elements include changes in ambient light and shadow, sound wave interference, and the distribution of obstructions;

[0082] The system senses the source, propagation trajectory, and intensity of interference elements. Based on these factors, it determines the distribution pattern of interference factors within the access control monitoring area. This distribution pattern includes the spatial distribution and temporal variation of interference elements within the access control monitoring area.

[0083] The related regions are divided according to the distribution of interference factors to obtain the related partition set.

[0084] First, the interference elements of the disturbed scene are collected. These elements include changes in ambient light and shadow, sound interference, and the distribution of obstructions. Changes in ambient light and shadow can be fluctuations in brightness and the speed and range of light and shadow movement, such as the flickering light in a corridor caused by cloud movement. Sound interference includes the decibel level of noise and the frequency range of sound waves, such as the low noise of equipment operating outside the office. The distribution of obstructions involves the number, size, and movement of obstructions, such as the temporary obstruction caused by occasional people passing by near the access control monitoring area.

[0085] Next, we perceive the source, propagation trajectory, and intensity of these interfering elements. The source refers to where the interference originates; for example, the source of light and shadow changes is the sun and clouds, the source of sound wave interference is equipment, and the source of obstructions is passing personnel. The propagation trajectory describes how the interference travels through space, such as the path of light entering the monitoring area from outside the window, the direction of sound wave propagation in the air, and the movement route of obstructions within the monitoring area. The intensity is the degree of impact of the interference on the monitoring area, such as the difference in brightness due to light changes, the decibel value of sound waves, and the proportion of the monitoring area obstructed by obstructions. Let's assume the intensity of the interfering element is... By generating the source, propagation trajectory, and intensity of interference, the distribution pattern of interference factors within the access control monitoring area is determined. This distribution pattern includes the spatial distribution and temporal variation of interference elements within the monitoring area. For example, at a certain time, light moves from the left side of the monitoring area, a sound wave of 30 decibels comes from the right side, and an obstruction stays in the middle area for 5 seconds. These pieces of information together constitute the spatial distribution and temporal variation pattern of the interference factors.

[0086] The behaviorally related regions are divided into related partition sets based on the distribution of interference factors. The behaviorally related region is the area where the target object and the access control unit interact. This region is further divided into different partitions based on the distribution of interference factors. For example, the left-hand area with strong light changes, the right-hand area with significant sound interference, and the middle area with frequent obstructions within the monitoring area are each divided into different related partitions. Each partition has a different degree and type of interference, which helps in subsequent feature correction based on the interference situation in different partitions.

[0087] The dynamic behavior feature cluster is obtained by correcting the target object's dynamic behavior feature cluster two in the disturbed scene based on the interference influence coefficient. The specific steps include:

[0088] The association feature dataset is obtained by statistically analyzing the association range and behavioral interaction frequency of each association partition in the association partition set;

[0089] Based on the interference impact coefficient, the associated feature dataset, and the scene interference change values ​​of each associated partition in the associated partition set, the feature extraction degree of the associated partition set is determined to obtain the feature extraction degree value.

[0090] The dynamic behavior feature cluster 2 is modified based on the feature extraction degree value to obtain the dynamic behavior modified feature cluster.

[0091] First, the association range and behavioral interaction frequency of each associated partition in the associated partition set are statistically analyzed to form an association feature dataset. The association range refers to the spatial coverage of each associated partition within the access control monitoring area; for example, an associated partition may cover an area 1 to 3 meters in front of the access control unit. The behavioral interaction frequency is the number of times a target object interacts with the access control unit within that partition, such as the number of times authorized personnel swipe their cards or linger. These data collectively constitute the association feature dataset, used to reflect the behavioral association attributes of each partition.

[0092] Based on the interference impact coefficient, the associated feature dataset, and the scene interference variation values ​​of each associated partition in the associated partition set, the feature extraction level of the associated partition set is determined. The interference impact coefficient is the correlation coefficient between the behavioral feature error rate and the scene interference variation values ​​over historical periods. The scene interference change value is the amount of change in the interference level within each associated partition. The association range in the associated feature dataset is set to The frequency of behavioral interaction is set to The feature extraction level value E can then be calculated by combining these parameters, for example... The feature extraction degree value is used to measure the depth and breadth of behavioral features that need to be extracted for each associated partition. The larger the interference influence coefficient, the larger the scene interference change value, the wider the association range, or the higher the frequency of behavioral interaction, the higher the feature extraction degree value, which means that more detailed behavioral features need to be extracted for that partition.

[0093] The dynamic behavior feature cluster 2 is modified based on the feature extraction level values ​​to obtain the modified dynamic behavior feature cluster. For example, if the feature extraction level value of a certain associated partition is high, it indicates that interference in that partition has a significant impact on the behavioral features, and the dynamic behavioral features (such as gait and limb movements) collected in that partition need to be modified by adjusting the feature parameters to offset the impact of interference. Partitions with lower feature extraction level values ​​only require simple modification. By modifying the features of each associated partition in the dynamic behavior feature cluster 2 in this way, the features are finally integrated to obtain the modified dynamic behavior feature cluster, which can more accurately reflect the real behavioral patterns of the target object in a disturbed scenario, providing a reliable basis for subsequent authorization judgments.

[0094] When the feature extraction level value of a certain associated partition is high, it indicates that the interference in that partition has a significant impact on behavioral features, and the dynamic behavioral features (such as stride length, stride frequency, and amplitude of limb movements) within that partition need to be corrected. For example, if the gait feature acquisition in that partition is deviated due to light and shadow interference, the gait parameters are adjusted according to the feature extraction level value. For instance, if the stride length was originally misjudged as 50 cm due to interference, it is corrected to the actual 60 cm based on the level value.

[0095] The second judgment result is obtained by judging the dynamic behavior feature cluster one, the dynamic behavior modification feature cluster, and the dynamic behavior feature cluster three. The specific steps include:

[0096] If the scene adaptability of dynamic behavior feature cluster 1, dynamic behavior correction feature cluster, and dynamic behavior feature cluster 3 are all higher than the second preset adaptability threshold, it indicates that the target object's behavior features match well with their respective scenes in the baseline scene, the corrected disturbance scene, and the strong disturbance scene. Therefore, the target object is confirmed to be in the authorized access state. For example, if the authorized person's gait stride length, stride frequency, and amplitude of limb movements all have a scene adaptability higher than the second preset adaptability threshold in the following scenarios: a baseline scene without interference (such as a quiet office building lobby entrance), a disturbance scene with slight light and shadow interference that has been corrected (such as tree shadows swaying at the lobby entrance and features corrected), and a strong disturbance scene with strong sound wave interference (such as equipment operating at high decibels near the lobby), then the target object is confirmed to be in the authorized access state.

[0097] If any one of the scene adaptability values ​​of dynamic behavior feature cluster one, dynamic behavior correction feature cluster, and dynamic behavior feature cluster three is not higher than the second preset adaptability threshold, that is, if the scene adaptability value of any one of the three types of behavior feature clusters is not higher than the second preset adaptability threshold, it indicates that the target object's behavior features do not match the scene in at least one scenario. In this case, the target object is confirmed to be in an unauthorized state. For example, in the baseline scenario, an unauthorized person may have a low scene adaptability value for their gait due to unfamiliarity with the environment; or in the corrected perturbation scenario, residual interference may still cause the scene adaptability value of the behavior features to be lower than the threshold; or in the strong perturbation scenario, excessive interference may cause the behavior features to be completely distorted, resulting in a scene adaptability value lower than the threshold. In this case, the target object is confirmed to be in an unauthorized state.

[0098] The access control result is obtained by processing the behavioral feature extraction parameters, the first scene transition time, the first behavioral feature adaptation increment, the second scene transition time, and the second behavioral feature adaptation increment. Specifically, this includes the following steps:

[0099] The first scene transition time and the first behavior feature are incrementally combined to form the first scene feature; the second scene transition time and the second behavior feature are incrementally combined to form the second scene feature.

[0100] The historical scene features and historical feature extraction accuracy of the target object in a historical period are obtained, and the influence between the historical scene features and historical feature extraction accuracy is evaluated to obtain the accuracy influence coefficient.

[0101] The first scene features, accuracy impact coefficient, and behavioral feature extraction parameters are processed to obtain the first dynamic behavioral data;

[0102] Collect the behavioral characteristic parameters of the target object in a strong disturbance scenario, and process the second scenario features, accuracy influence coefficient and behavioral characteristic parameters to obtain the second dynamic behavior data;

[0103] The access control result is obtained by confirming the access authorization status of the target object based on the first dynamic behavior data and the second dynamic behavior data.

[0104] We obtain historical scene features of the target object (adapted incremental data on historical scene transition duration and behavioral features similar to the first and second scene features) and historical feature extraction accuracy (the proportion of accurate behavioral features extracted from the object historically). We then assess the degree of influence between the two to obtain the accuracy influence coefficient. .

[0105] Assume that a certain parameter in the historical scene features is The corresponding historical feature extraction accuracy is ,Establish and The mathematical relationship between them is used to derive the accuracy impact coefficient. . Where a and b are fitting coefficients, then different models are determined based on this model. corresponding This coefficient can reflect the degree of influence of historical scene features on the accuracy of historical feature extraction, and provide a basis for subsequent calibration of behavioral feature parameters based on current scene features.

[0106] Based on the characteristics of the first scenario, the accuracy impact coefficient Extract the corresponding first influence coefficient Combining the characteristics of the first scenario and the first influence coefficient The behavioral feature extraction parameters (such as sampling frequency and region range for extracting target behavioral features in a disturbed scene, for example, sampling frequency of n times per second and extraction region from ankle to waist) are calibrated to obtain the disturbed scene extraction parameters. Then, the first dynamic behavioral data of the target object in the disturbed scene are collected based on these parameters. For example, after calibration, the sampling frequency is adjusted to n′ times per second to collect gait and limb movement feature data.

[0107] Collect behavioral characteristic parameters of the target object in a strongly disturbed scenario (raw data of the target's behavioral characteristics under strongly disturbed scenarios, such as stride length fluctuations and unstable amplitude of limb movements), and determine the accuracy impact coefficient based on the characteristics of the second scenario. Extract the corresponding second influence coefficient Combining the characteristics of the second scenario and the second influence coefficient The behavioral feature parameters are calibrated to obtain the strong disturbance scene extraction parameters. Then, based on these parameters, the second dynamic behavior data of the target object in the strong disturbance scene is collected, such as the feature data after noise reduction of the step size fluctuation data after calibration.

[0108] Finally, the access authorization status of the target object is determined by combining the first dynamic behavior data and the second dynamic behavior data: if both types of dynamic behavior data match the behavior pattern of the authorized object (such as gait parameters matching historical authorization data after correction, and limb movement characteristics maintaining the stability of authorization characteristics under strong perturbation), then access is confirmed; if either data does not match, then access is confirmed to be denied, thus obtaining the final access control result.

[0109] The first scene features, accuracy impact coefficient, and behavioral feature extraction parameters are processed to obtain the first dynamic behavioral data, specifically including the following steps:

[0110] The first influence coefficient is extracted from the accuracy influence coefficient based on the first scene features. The perturbation scene extraction parameters are obtained by calibrating the behavior feature extraction parameters based on the first scene features and the first influence coefficient.

[0111] The first dynamic behavior data of the target object in the disturbance scene is collected based on the parameters extracted from the disturbance scene.

[0112] The first influence coefficient is extracted from the accuracy influence coefficient based on the first scene features. The first scene features consist of the transition time from the baseline scene to the perturbation scene and the behavioral feature adaptation increment; for example, the transition time is 5 seconds and the gait stride adaptation increment is 10 centimeters. The accuracy influence coefficient is the correlation coefficient between scene features and feature extraction accuracy obtained from historical data. Based on the specific values ​​of the first scene features, the corresponding first influence coefficient is matched from this coefficient set. For example, in historical data, a conversion time of 5 seconds and a step size increment of 10 centimeters correspond to an influence coefficient of 0.8. Therefore, the first influence coefficient... =0.8.

[0113] The behavioral feature extraction parameters are calibrated based on the first scene features and the first influence coefficient to obtain the perturbed scene extraction parameters. These parameters include the sampling frequency and region range for extracting target behavioral features in the perturbed scene; for example, the original sampling frequency is 5 times per second, and the extraction region is from the shoulder to the foot. These parameters are combined with the transformation time, step size increment, and the first influence coefficient of the first scene features. Adjust these parameters. Assume the calibration formula is... ,in It is the calibrated sampling frequency. It is the first influence coefficient. It is the conversion time. It is the step size increment, if the original sampling frequency =5, =5, =10, =0.8, then the calibrated sampling frequency =9 times / second.

[0114] Finally, the first dynamic behavior data of the target object in the disturbed scene is collected based on the parameters extracted from the disturbed scene. Following a calibrated sampling frequency of 9 times / second and an adjusted extraction area, dynamic behavioral characteristics of the target object in the disturbed scene, such as gait and limb movements, are collected. For example, data such as stride length variation sequences and changes in arm swing angles are collected. This data constitutes the first dynamic behavior data, which is used for subsequent confirmation of the target object's access authorization status.

[0115] Collect behavioral characteristic parameters of the target object in a strongly disturbed scenario, and process the second scenario features, accuracy impact coefficient, and behavioral characteristic parameters to obtain the second dynamic behavioral data. This process includes the following steps:

[0116] The second influence coefficient is extracted from the accuracy influence coefficient based on the second scene features. The behavioral feature status parameters are then calibrated based on the second scene features and the second influence coefficient to obtain the strong disturbance scene extraction parameters.

[0117] The second dynamic behavior data of the target object in the strong disturbance scene is collected based on the parameters extracted from the strong disturbance scene.

[0118] The second scene feature is used to extract the second influence coefficient from the accuracy influence coefficient. The second scene feature consists of the transition time from a perturbed scene to a strongly perturbed scene and the behavioral feature adaptation increment; for example, the transition time is 8 seconds and the gait stride adaptation increment is 15 centimeters. The accuracy influence coefficient is the correlation coefficient between scene features and feature extraction accuracy obtained from historical data. Based on the specific value of the second scene feature, the corresponding second influence coefficient is matched from this coefficient set. For example, in historical data, the influence coefficient corresponding to a conversion time of 8 seconds and a step size increment of 15 centimeters is 0.7. Therefore, the second influence coefficient... It is 0.7.

[0119] The behavioral feature parameters are calibrated based on the second scene features and the second influence coefficient to obtain the extracted parameters for the perturbed scene. The behavioral feature parameters are the raw collected data of the target's behavioral characteristics under the perturbed scene, such as the stride length fluctuation amplitude and the instability of limb movements. These parameters are then combined with the second scene features, transition time, stride length increment, and the second influence coefficient. Adjust these parameters. Assume the calibration formula is... ,in It is the step size fluctuation amplitude after calibration. It is the original step size fluctuation range. It is the conversion time. It is the step size increment, if the original step size fluctuation range =6 cm, =8, =15, =0.7, then the step size fluctuation amplitude after calibration The value is 22.2 cm, and parameters such as the degree of instability of limb movements are also calibrated in the same way to obtain the extraction parameters for strong disturbance scenes.

[0120] Finally, based on the parameters extracted from the strong disturbance scenario, the second dynamic behavior data of the target object in the strong disturbance scenario is collected. In accordance with the calibrated step length fluctuation amplitude collection standard and the requirements for collecting limb movement instability, the dynamic behavior characteristics of the target object under the strong disturbance scenario, such as gait and limb movements, are collected. For example, data such as step length fluctuation sequence and the range of changes in arm swing angle are collected. These data constitute the second dynamic behavior data, which is used to confirm the access authorization status of the target object in the subsequent process.

[0121] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0122] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smart security access control system, characterized in that, Comprise: Acquisition module: in the benchmark scene, disturbance scene and strong disturbance scene respectively acquire the dynamic behavior characteristic cluster one, dynamic behavior characteristic cluster two and dynamic behavior characteristic cluster three of the target object in the access control monitoring area; Judgment correction module: the first judgment result is obtained by judging the dynamic behavior characteristic cluster one and the dynamic behavior characteristic cluster two;The dynamic behavior correction characteristic cluster is obtained by correcting the dynamic behavior characteristic cluster two of the target object in the disturbance scene according to the first judgment result; Processing module: the second judgment result is obtained by judging the dynamic behavior characteristic cluster one, dynamic behavior correction characteristic cluster and dynamic behavior characteristic cluster three; Detection module: if the second judgment result is clear that the target object is in the authorized state of passing through, the first scene conversion time length and the first behavior characteristic adaptation increment between the benchmark scene and the disturbance scene are detected, and the second scene conversion time length and the second behavior characteristic adaptation increment between the disturbance scene and the strong disturbance scene are detected, and the behavior characteristic extraction parameter of the target object in the disturbance scene is collected; Output module: the behavior characteristic extraction parameter, the first scene conversion time length, the first behavior characteristic adaptation increment, the second scene conversion time length and the second behavior characteristic adaptation increment are processed to obtain the access control result.

2. The intelligent security access control system according to claim 1, wherein, The first judgment result is obtained by judging the dynamic behavior characteristic cluster one and the dynamic behavior characteristic cluster two;The dynamic behavior correction characteristic cluster is obtained by correcting the dynamic behavior characteristic cluster two of the target object in the disturbance scene according to the first judgment result, which comprises the following steps: If the scene adaptation degree of the dynamic behavior characteristic cluster one and the dynamic behavior characteristic cluster two is greater than or equal to the first preset adaptation threshold, it is judged that the target object is in the authorized state of passing through; If the scene adaptation degree of the dynamic behavior characteristic cluster one or / and the dynamic behavior characteristic cluster two is less than the first preset adaptation threshold, it is judged that the target object is in the unauthorized state of passing through; If the target object is in the unauthorized state of passing through, the behavior association area between the target object and the access control execution unit in the disturbance scene is collected, and the behavior association area is divided into association partition set according to the interference factor distribution situation; The interference influence coefficient between the behavior characteristic error rate and the scene interference change value in the historical period is extracted, and the dynamic behavior correction characteristic cluster is obtained by correcting the dynamic behavior characteristic cluster two of the target object in the disturbance scene according to the interference influence coefficient.

3. The intelligent security access control system of claim 1, wherein, Further comprise the following steps: The benchmark scene is the normal operation scene without interference; The disturbance intensity of the running scene is collected, and the disturbance intensity is compared with the preset disturbance intensity threshold; If the disturbance intensity is less than or equal to the preset disturbance intensity threshold, the running scene is marked as the disturbance scene; If the disturbance intensity is greater than the preset disturbance intensity threshold, the running scene is marked as the strong disturbance scene.

4. The intelligent security access control system according to claim 2, wherein, The behavior association area is divided into association partition set according to the interference factor distribution situation, which comprises the following steps: Collect the interference elements of the disturbance scene;The interference elements include environmental light and shadow change, sound wave interference and shelter distribution; The generation source, propagation track and action intensity of the interference elements are perceived, and the interference factor distribution situation in the access control monitoring area is determined according to the generation source, propagation track and action intensity of the interference elements; According to the interference factor distribution situation, the behavior correlation region is divided to obtain a correlation partition set.

5. The intelligent security access control system of claim 2, wherein, According to the interference influence coefficient, the dynamic behavior characteristic cluster two of the target object in the disturbance scene is corrected to obtain a dynamic behavior corrected characteristic cluster, specifically including the following steps: The correlation range and behavior interaction frequency of each correlation partition in the correlation partition set are counted to obtain a correlation characteristic data set; According to the interference influence coefficient, the correlation characteristic data set, and the scene interference change value of each correlation partition in the correlation partition set, the feature extraction degree of the correlation partition set is determined to obtain a feature extraction degree value; According to the feature extraction degree value, the dynamic behavior characteristic cluster two is corrected to obtain a dynamic behavior corrected characteristic cluster.

6. The intelligent security access control system of claim 5, wherein, The dynamic behavior characteristic cluster one, the dynamic behavior corrected characteristic cluster, and the dynamic behavior characteristic cluster three are judged to obtain a second judgment result, specifically including the following steps: If the scene adaptation degrees of the dynamic behavior characteristic cluster one, the dynamic behavior corrected characteristic cluster, and the dynamic behavior characteristic cluster three are all higher than the second preset adaptation threshold, it is confirmed that the target object is in a traffic authorization state; If any one of the scene adaptation degrees of the dynamic behavior characteristic cluster one, the dynamic behavior corrected characteristic cluster, and the dynamic behavior characteristic cluster three is not higher than the second preset adaptation threshold, it is confirmed that the target object is in a non-traffic authorization state.

7. The intelligent security access control system of claim 1, wherein, The behavior feature extraction parameter, the first scene conversion time length, the first behavior feature adaptation increment, the second scene conversion time length, and the second behavior feature adaptation increment are processed to obtain an access control result, specifically including the following steps: The first scene conversion time length and the first behavior feature adaptation increment are combined as a first scene feature; the second scene conversion time length and the second behavior feature adaptation increment are combined as a second scene feature; The historical scene feature and the historical feature extraction accuracy rate of the target object in a historical period are obtained, the influence degree between the historical scene feature and the historical feature extraction accuracy rate is evaluated to obtain an accuracy rate influence coefficient; The first scene feature, the accuracy rate influence coefficient, and the behavior feature extraction parameter are processed to obtain first dynamic behavior data; The behavior feature status parameter of the target object in a strong disturbance scene is collected, and the second scene feature, the accuracy rate influence coefficient, and the behavior feature status parameter are processed to obtain second dynamic behavior data; The traffic authorization state of the target object is confirmed according to the first dynamic behavior data and the second dynamic behavior data to obtain an access control result.

8. The intelligent security access control system of claim 7, wherein, The first scene feature, the accuracy rate influence coefficient, and the behavior feature extraction parameter are processed to obtain first dynamic behavior data, specifically including the following steps: According to the first scene feature, a first influence coefficient is extracted from the accuracy rate influence coefficient, and the behavior feature extraction parameter is calibrated according to the first scene feature and the first influence coefficient to obtain a disturbance scene extraction parameter; The first dynamic behavior data of the target object in the disturbance scene is collected according to the disturbance scene extraction parameter.

9. The intelligent security access control system of claim 7, wherein, The behavior feature status parameter of the target object in a strong disturbance scene is collected, and the second scene feature, the accuracy rate influence coefficient, and the behavior feature status parameter are processed to obtain second dynamic behavior data, specifically including the following steps: The second influence coefficient is extracted from the accuracy influence coefficient according to the second scene feature, and the behavior feature condition parameter is calibrated according to the second scene feature and the second influence coefficient to obtain a strong interference scene extraction parameter; The second dynamic behavior data of the target object in the strong interference scene is collected according to the strong interference scene extraction parameter.

10. The intelligent security access control system of claim 4, wherein, The interference factor distribution situation includes a spatial distribution state and a time variation law of the interference elements in the access control monitoring area.

Images