Access control face recognition system and method based on three-dimensional technology

Abstract

The invention discloses an access control face recognition system and method based on a three-dimensional technology, and relates to the technical field of access control face recognition, and the recognition system comprises a face data acquisition module which is used for capturing three-dimensional face data of a target person by using a 3D camera; the multi-modal fusion module is used for fusing the three-dimensional face data and the traditional two-dimensional image data; the modeling module is used for generating an accurate three-dimensional face model; the dynamic adjustment module is used for creating a real-time dynamic adjustment model; the matching module is used for extracting facial features from the three-dimensional face model and matching the facial features with the three-dimensional face model in the database; and the identity verification and authorization module is used for carrying out identity verification according to the matching result. According to the method, three-dimensional face data and traditional two-dimensional image data are fused, the recognition accuracy and reliability can be improved, a real-time dynamic adjustment function is introduced, recognition parameters can be automatically optimized according to different environment light rays and angles, and better use prospects are brought.

Description

Technical Field

[0001] The present invention relates to the technical field of access control face recognition, and in particular to an access control face recognition system and method based on three-dimensional technology. Background Art

[0002] Access control face recognition is a biometric technology based on facial features. It is widely used in access control systems. Its main function is to verify the identity of an individual by identifying his or her facial features, thereby allowing or denying access to restricted areas. Access control face recognition systems use computer vision and image processing technology to analyze and identify individual facial features. Its core is to capture facial images or video streams, analyze facial features in these images, and compare them with archived data in the database to confirm identity. Application scenarios include office buildings and enterprises, controlling the access of employees and visitors to ensure safety; financial institutions: protecting the safety of bank branches and ATMs; public places such as airports, stations, etc., improve traffic efficiency through rapid identification; residential areas, enhance home security, and prevent unauthorized entry.

[0003] The existing access control face recognition system cannot combine two-dimensional information and three-dimensional information for comprehensive recognition, and cannot automatically optimize recognition parameters according to different ambient light and angles, and cannot achieve dynamic adjustment. Summary of the invention

[0004] In order to solve the above technical problems, a door access face recognition system and method based on three-dimensional technology are provided. This technical solution solves the above problems.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] Access control face recognition system based on 3D technology, the recognition system includes:

[0007] Face data acquisition module: used to capture the three-dimensional face data of the target person using a 3D camera to obtain detailed facial features;

[0008] Multimodal fusion module: The multimodal fusion module is electrically connected to the face data acquisition module, and the multimodal fusion module is used to fuse the three-dimensional face data with the traditional two-dimensional image data;

[0009] Modeling module: The modeling module is electrically connected to the multimodal fusion module and the face data acquisition module. The modeling module is used to pre-process the fused data, perform facial feature modeling, and generate an accurate three-dimensional face model;

[0010] Dynamic adjustment module: The dynamic adjustment module is electrically connected to the modeling module, the multimodal fusion module and the face data acquisition module. The dynamic adjustment module is used to create a real-time dynamic adjustment model, and based on the real-time dynamic adjustment model, outputs different ambient light and angles to automatically optimize recognition parameters;

[0011] Matching module: The matching module and the dynamic adjustment module are electrically connected to the modeling module, the multimodal fusion module and the face data acquisition module. The matching module is used to extract facial features from the three-dimensional face model, match it with the three-dimensional face model in the database, and identify the identity;

[0012] Authentication and authorization module: The authentication and authorization module is electrically connected to the feature extraction and matching module, the dynamic adjustment module and the modeling module, the multimodal fusion module and the face data acquisition module. The authentication and authorization module is used to perform identity authentication based on the matching results, decide whether to grant access control permissions, and record access logs.

[0013] Preferably, the face data collection module specifically includes:

[0014] 3D camera unit: uses a 3D sensor to capture the 3D facial data of the target person and measures the depth information of the face through laser and structured light;

[0015] Depth image sensor unit: obtains the depth image of the face;

[0016] 2D camera unit: captures a 2D image of the target person;

[0017] Synchronization unit: synchronizes data acquisition between 3D camera and 2D camera;

[0018] Data buffer unit: temporarily stores the collected 3D data to ensure the integrity of the data during transmission;

[0019] Real-time data monitoring unit: monitors the data quality during the collection process in real time and provides feedback.

[0020] Preferably, the multimodal fusion module specifically includes:

[0021] Data alignment unit: spatially aligns and registers data from different modalities;

[0022] Among them, the alignment formula is:

[0023]

[0024] In the formula, p i With q i is the corresponding point set, T is the transformation matrix, a T is the minimization objective function of T;

[0025] Fusion strategy unit: weights features from different modalities into a unified representation;

[0026] The fusion formula is:

[0027] F 融 =αF 2d +(1-α)F 3d

[0028] In the formula, F 融 is the fused feature, F 2d is a feature from two dimensions, F 3d is the feature from the three-dimensional modality, α is the fusion weight;

[0029] Data fusion output unit: outputs the fused features.

[0030] Preferably, the modeling module specifically includes:

[0031] Data preprocessing unit: cleans and corrects the fused 3D and 2D data, and uses filters to smooth the data;

[0032] The smoothing data formula is:

[0033]

[0034] In the formula, I 滤 (x, y) is the coordinate value of the filtered image at position (x, y), I(x`, y`) is the value of the original image at position (x`, y`), σ is the standard deviation of the Gaussian filter, 2πσ 2 is the normalization constant, is a Gaussian function, dx`dy` is the distance of position (x`, y`);

[0035] Feature extraction unit: uses Harris corner detection algorithm to extract facial features from processed data;

[0036] Model building unit: construct a 3D facial model using the extracted feature data, convert the fused data into a 3D point cloud, and generate a 3D mesh model from the point cloud;

[0037] Model optimization unit: Smooth the mesh and remove noise.

[0038] Preferably, the feature extraction unit specifically includes:

[0039] Among them, the Harris corner detection algorithm formula is:

[0040] R = det(M) - k(t(M)) 2

[0041] Where R is the Harris corner response function, M is the autocorrelation matrix, k is a constant, det(M) is the corner intensity of the local image area, and t(M) is the total intensity of the local image area.

[0042] Preferably, the dynamic adjustment module specifically includes:

[0043] Real-time data acquisition unit: obtains ambient light and angle data from sensors and cameras in real time;

[0044] Ambient light and angle analysis unit: analyzes the collected ambient light and angle data, evaluates their impact on the recognition effect, and obtains the ambient light adjustment coefficient and angle adjustment coefficient;

[0045] Dynamic adjustment model unit: based on the analysis results, dynamically adjust the recognition parameters, create and update the real-time dynamic adjustment model;

[0046] Among them, the dynamic adjustment model formula is:

[0047] L 新 =L+αA

[0048] Where, L 新 is the updated dynamic adjustment model, L is the previous dynamic adjustment model, α is the adjustment factor, and A is the adjustment amount calculated based on the current environmental data and target optimization requirements;

[0049] Information fusion unit: fuses information from different modalities;

[0050] Among them, the information fusion formula is:

[0051] J 融 =β1D+β2P

[0052] In the formula, J 融 is the fused data, D is the data from the visual modality, P is the data from the depth modality, β1 is the weighting coefficient of the visual data, and β2 is the weighting coefficient of the depth data;

[0053] Optimized output unit: Outputs optimized parameters based on dynamic adjustment model.

[0054] Preferably, the matching module comprises:

[0055] Feature extraction unit: extract facial features from the 3D face model;

[0056] Feature matching unit: matches the extracted features with the 3D face model in the database to identify the identity;

[0057] Data preprocessing unit: preprocess the input 3D face model;

[0058] Database update unit: Add new face models and their features to the database.

[0059] The access control face recognition method based on three-dimensional technology is characterized in that the recognition method includes:

[0060] Use a 3D camera to capture the three-dimensional face data of the target person and obtain detailed facial features;

[0061] Fusion of 3D face data with traditional 2D image data;

[0062] Preprocess the fused data, perform facial feature modeling, and generate an accurate 3D face model;

[0063] Create a real-time dynamic adjustment model, based on which the recognition parameters are automatically optimized under different ambient light and angles;

[0064] Extract facial features from the 3D face model, match it with the 3D face model in the database, and identify the identity;

[0065] Authentication is performed based on the matching results, and a decision is made as to whether access control permissions are granted, while access logs are recorded.

[0066] The access control face recognition method based on three-dimensional technology includes:

[0067] Use a 3D camera to capture the three-dimensional face data of the target person and obtain detailed facial features;

[0068] Fusion of 3D face data with traditional 2D image data;

[0069] Preprocess the fused data, perform facial feature modeling, and generate an accurate 3D face model;

[0070] Create a real-time dynamic adjustment model, based on which the recognition parameters are automatically optimized under different ambient light and angles;

[0071] Extract facial features from the 3D face model, match it with the 3D face model in the database, and identify the identity;

[0072] Authentication is performed based on the matching results, and a decision is made as to whether access control permissions are granted, while access logs are recorded.

[0073] Preferably, fusing the three-dimensional face data with the traditional two-dimensional image data specifically includes:

[0074] Data alignment unit: spatially aligns and registers data from different modalities;

[0075] Fusion strategy unit: weights features from different modalities into a unified representation;

[0076] Data fusion output unit: outputs the fused features.

[0077] Based on the real-time dynamic adjustment model, the output of different ambient light and angles automatically optimizes the recognition parameters, including:

[0078] Get data about ambient light and angle from sensors and cameras in real time;

[0079] Analyze the collected ambient light and angle data, evaluate their impact on the recognition effect, and obtain the ambient light adjustment coefficient and angle adjustment coefficient;

[0080] Based on the analysis results, dynamically adjust the recognition parameters, create and update the real-time dynamic adjustment model;

[0081] Fuse information from different modalities;

[0082] Output optimized parameters based on dynamic adjustment model.

[0083] Compared with the prior art, the present invention has the following beneficial effects:

[0084] The present invention proposes to integrate three-dimensional face data with traditional two-dimensional image data, which can improve recognition accuracy and reliability, and improve the accuracy of three-dimensional face feature extraction and matching. The present invention also introduces a real-time dynamic adjustment function, which can automatically optimize recognition parameters according to different ambient light and angles, bringing better usage prospects and commercial value. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 It is the internal system framework diagram of the present invention;

[0086] Figure 2 It is a step flow framework diagram of the present invention. DETAILED DESCRIPTION

[0087] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0088] Reference Figure 1 As shown, the access control face recognition system based on three-dimensional technology includes:

[0089] Face data acquisition module: used to capture the three-dimensional face data of the target person using a 3D camera to obtain detailed facial features;

[0090] Multimodal fusion module: The multimodal fusion module is electrically connected to the face data acquisition module, and the multimodal fusion module is used to fuse the three-dimensional face data with the traditional two-dimensional image data;

[0091] Modeling module: The modeling module is electrically connected to the multimodal fusion module and the face data acquisition module. The modeling module is used to pre-process the fused data, perform facial feature modeling, and generate an accurate three-dimensional face model;

[0092] Dynamic adjustment module: The dynamic adjustment module is electrically connected to the modeling module, the multimodal fusion module and the face data acquisition module. The dynamic adjustment module is used to create a real-time dynamic adjustment model, and based on the real-time dynamic adjustment model, outputs different ambient light and angles to automatically optimize recognition parameters;

[0093] Matching module: The matching module and the dynamic adjustment module are electrically connected to the modeling module, the multimodal fusion module and the face data acquisition module. The matching module is used to extract facial features from the three-dimensional face model, match it with the three-dimensional face model in the database, and identify the identity;

[0094] Authentication and authorization module: The authentication and authorization module is electrically connected to the feature extraction and matching module, the dynamic adjustment module and the modeling module, the multimodal fusion module and the face data acquisition module. The authentication and authorization module is used to perform identity authentication based on the matching results, decide whether to grant access control permissions, and record access logs.

[0095] The face data collection module specifically includes:

[0096] 3D camera unit: uses a 3D sensor to capture the 3D facial data of the target person and measures the depth information of the face through laser and structured light;

[0097] Depth image sensor unit: obtains the depth image of the face;

[0098] 2D camera unit: captures a 2D image of the target person;

[0099] Synchronization unit: synchronizes data acquisition between 3D camera and 2D camera;

[0100] Data buffer unit: temporarily stores the collected 3D data to ensure the integrity of the data during transmission;

[0101] Real-time data monitoring unit: monitors the data quality during the collection process in real time and provides feedback.

[0102] The multimodal fusion module specifically includes:

[0103] Data alignment unit: spatially aligns and registers data from different modalities;

[0104] Among them, the alignment formula is:

[0105]

[0106] In the formula, p i With q i is the corresponding point set, T is the transformation matrix, a T is the minimization objective function of T;

[0107] Fusion strategy unit: weights features from different modalities into a unified representation;

[0108] The fusion formula is:

[0109] F 融 =αF 2d +(1-α)F 3d

[0110] In the formula, F 融 is the fused feature, F 2d is a feature from two dimensions, F 3d is the feature from the three-dimensional modality, α is the fusion weight;

[0111] Data fusion output unit: outputs the fused features.

[0112] The modeling module specifically includes:

[0113] Data preprocessing unit: cleans and corrects the fused 3D and 2D data, and uses filters to smooth the data;

[0114] The smoothing data formula is:

[0115] In the formula, I 滤 (x, y) is the coordinate value of the filtered image at position (x, y), I(x`, y`) is the value of the original image at position (x`, y`), σ is the standard deviation of the Gaussian filter, 2πσ 2 is the normalization constant, is a Gaussian function, dx`dy` is the distance of position (x`, y`);

[0116] Feature extraction unit: uses Harris corner detection algorithm to extract facial features from processed data;

[0117] Model building unit: construct a 3D facial model using the extracted feature data, convert the fused data into a 3D point cloud, and generate a 3D mesh model from the point cloud;

[0118] Model optimization unit: Smooth the mesh and remove noise.

[0119] The feature extraction unit specifically includes:

[0120] Among them, the Harris corner detection algorithm formula is:

[0121] R = det(M) - k(t(M)) 2

[0122] Where R is the Harris corner response function, M is the autocorrelation matrix, k is a constant, det(M) is the corner intensity of the local image area, and t(M) is the total intensity of the local image area.

[0123] The dynamic adjustment module specifically includes:

[0124] Real-time data acquisition unit: obtains ambient light and angle data from sensors and cameras in real time;

[0125] Ambient light and angle analysis unit: analyzes the collected ambient light and angle data, evaluates their impact on the recognition effect, and obtains the ambient light adjustment coefficient and angle adjustment coefficient;

[0126] Dynamic adjustment model unit: based on the analysis results, dynamically adjust the recognition parameters, create and update the real-time dynamic adjustment model;

[0127] Among them, the dynamic adjustment model formula is:

[0128] L 新 =L+αA

[0129] Where, L 新 is the updated dynamic adjustment model, L is the previous dynamic adjustment model, α is the adjustment factor, and A is the adjustment amount calculated based on the current environmental data and target optimization requirements;

[0130] Information fusion unit: fuses information from different modalities;

[0131] Among them, the information fusion formula is:

[0132] J 融 =β1D+β2P

[0133] In the formula, J 融 is the fused data, D is the data from the visual modality, P is the data from the depth modality, β1 is the weighting coefficient of the visual data, and β2 is the weighting coefficient of the depth data;

[0134] Optimized output unit: Outputs optimized parameters based on dynamic adjustment model.

[0135] The matching module comprises:

[0136] Feature extraction unit: extract facial features from the 3D face model;

[0137] Feature matching unit: matches the extracted features with the 3D face model in the database to identify the identity;

[0138] Data preprocessing unit: preprocess the input 3D face model;

[0139] Database update unit: Add new face models and their features to the database.

[0140] The access control face recognition method based on three-dimensional technology includes:

[0141] Use a 3D camera to capture the three-dimensional face data of the target person and obtain detailed facial features;

[0142] Fusion of 3D face data with traditional 2D image data;

[0143] Preprocess the fused data, perform facial feature modeling, and generate an accurate 3D face model;

[0144] Create a real-time dynamic adjustment model, based on which the recognition parameters are automatically optimized under different ambient light and angles;

[0145] Extract facial features from the 3D face model, match it with the 3D face model in the database, and identify the identity;

[0146] Authentication is performed based on the matching results, and a decision is made as to whether access control permissions are granted, while access logs are recorded.

[0147] Reference Figure 2 As shown, the access control face recognition method based on three-dimensional technology includes:

[0148] Use a 3D camera to capture the three-dimensional face data of the target person and obtain detailed facial features;

[0149] Fusion of 3D face data with traditional 2D image data;

[0150] Preprocess the fused data, perform facial feature modeling, and generate an accurate 3D face model;

[0151] Create a real-time dynamic adjustment model, based on which the recognition parameters are automatically optimized under different ambient light and angles;

[0152] Extract facial features from the 3D face model, match it with the 3D face model in the database, and identify the identity;

[0153] Authentication is performed based on the matching results, and a decision is made as to whether access control permissions are granted, while access logs are recorded.

[0154] The fusion of 3D face data with traditional 2D image data specifically includes:

[0155] Data alignment unit: spatially aligns and registers data from different modalities;

[0156] Fusion strategy unit: weights features from different modalities into a unified representation;

[0157] Data fusion output unit: outputs the fused features.

[0158] Based on the real-time dynamic adjustment model, the output of different ambient light and angles automatically optimizes the recognition parameters, including:

[0159] Get data about ambient light and angle from sensors and cameras in real time;

[0160] Analyze the collected ambient light and angle data, evaluate their impact on the recognition effect, and obtain the ambient light adjustment coefficient and angle adjustment coefficient;

[0161] Based on the analysis results, dynamically adjust the recognition parameters, create and update the real-time dynamic adjustment model;

[0162] Fuse information from different modalities;

[0163] Output optimized parameters based on dynamic adjustment model.

[0164] The use process of the present invention is:

[0165] Step 1: Use a 3D sensor to capture the target person’s 3D facial data and measure the facial depth information through laser and structured light;

[0166] Step 2: Obtain a depth image of the face;

[0167] Step 3: Capture a two-dimensional image of the target person;

[0168] Step 4: Synchronize data collection of 3D camera and 2D camera;

[0169] Step 5: Temporarily store the collected 3D data to ensure the integrity of the data during transmission;

[0170] Step 6: Monitor the data quality during the collection process in real time and provide feedback;

[0171] Step 7: spatially align and register the data from different modalities;

[0172] Step 8: Weight the features from different modalities into a unified representation;

[0173] Step 9: Output the fused features;

[0174] Step 10: Clean and correct the fused 3D and 2D data, and use filters to smooth the data;

[0175] Step 11: Use Harris corner detection algorithm to extract facial features from the processed data;

[0176] Step 12: Use the extracted feature data to build a 3D facial model, convert the fused data into a 3D point cloud, and generate a 3D mesh model from the point cloud;

[0177] Step 13: Smooth the mesh and remove noise.

[0178] Step 14: Obtain data on ambient light and angle from the sensor and camera in real time;

[0179] Step 15: Analyze the collected ambient light and angle data, evaluate their impact on the recognition effect, and obtain the ambient light adjustment coefficient and the angle adjustment coefficient;

[0180] Step 16: Based on the analysis results, dynamically adjust the recognition parameters, create and update the real-time dynamic adjustment model;

[0181] Step 17: Fuse information from different modalities;

[0182] Step 18: Output optimized parameters according to the dynamic adjustment model;

[0183] Step 19: Extract facial features from the 3D face model;

[0184] Step 20: Match the extracted features with the 3D face model in the database to identify the identity;

[0185] Step 21: preprocessing the input 3D face model;

[0186] Step 22: Add the new face model and its features to the database.

[0187] To sum up, the advantages of the present invention are: the present invention proposes to integrate three-dimensional face data with traditional two-dimensional image data, which can improve recognition accuracy and reliability, and improve the accuracy of three-dimensional face feature extraction and matching. The present invention introduces a real-time dynamic adjustment function, which can automatically optimize recognition parameters according to different ambient light and angles, bringing better usage prospects and commercial value.

[0188] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. The access control face recognition system based on three-dimensional technology is characterized by: The identification system includes: Face data acquisition module: used to capture the three-dimensional face data of the target person using a 3D camera to obtain detailed facial features; Multimodal fusion module: The multimodal fusion module is electrically connected to the face data acquisition module, and the multimodal fusion module is used to fuse the three-dimensional face data with the traditional two-dimensional image data; Modeling module: The modeling module is electrically connected to the multimodal fusion module and the face data acquisition module. The modeling module is used to pre-process the fused data, perform facial feature modeling, and generate an accurate three-dimensional face model; Dynamic adjustment module: The dynamic adjustment module is electrically connected to the modeling module, the multimodal fusion module and the face data acquisition module. The dynamic adjustment module is used to create a real-time dynamic adjustment model, and based on the real-time dynamic adjustment model, outputs different ambient light and angles to automatically optimize recognition parameters; Matching module: The matching module and the dynamic adjustment module are electrically connected to the modeling module, the multimodal fusion module and the face data acquisition module. The matching module is used to extract facial features from the three-dimensional face model, match it with the three-dimensional face model in the database, and identify the identity; Authentication and authorization module: The authentication and authorization module is electrically connected to the feature extraction and matching module, the dynamic adjustment module and the modeling module, the multimodal fusion module and the face data acquisition module. The authentication and authorization module is used to perform identity authentication based on the matching results, decide whether to grant access control permissions, and record access logs.

2. The access control face recognition system and method based on three-dimensional technology according to claim 1 is characterized in that: The face data collection module specifically includes: 3D camera unit: uses a 3D sensor to capture the 3D facial data of the target person and measures the depth information of the face through laser and structured light; Depth image sensor unit: obtains the depth image of the face; 2D camera unit: captures a 2D image of the target person; Synchronization unit: synchronizes data acquisition between 3D camera and 2D camera; Data buffer unit: temporarily stores the collected 3D data to ensure the integrity of the data during transmission; Real-time data monitoring unit: monitors the data quality during the collection process in real time and provides feedback.

3. The access control face recognition system based on three-dimensional technology according to claim 2 is characterized in that: The multimodal fusion module specifically includes: Data alignment unit: spatially aligns and registers data from different modalities; Among them, the alignment formula is: In the formula, p i With q i is the corresponding point set, T is the transformation matrix, a T is the minimization objective function of T; Fusion strategy unit: weights features from different modalities into a unified representation; The fusion formula is: F 融 =αF 2d +(1-α)F 3d In the formula, F 融 is the fused feature, F 2d is a feature from two dimensions, F 3d is the feature from the three-dimensional modality, α is the fusion weight; Data fusion output unit: outputs the fused features.

4. The access control face recognition system based on three-dimensional technology according to claim 3 is characterized in that: The modeling module specifically includes: Data preprocessing unit: cleans and corrects the fused 3D and 2D data, and uses filters to smooth the data; The smoothing data formula is: In the formula, I 滤 (x, y) is the coordinate value of the filtered image at position (x, y), I(x`, y`) is the value of the original image at position (x`, y`), σ is the standard deviation of the Gaussian filter, 2πσ 2 is the normalization constant, is a Gaussian function, dx`dy` is the distance of position (x`, y`); Feature extraction unit: uses Harris corner detection algorithm to extract facial features from processed data; Model building unit: construct a 3D facial model using the extracted feature data, convert the fused data into a 3D point cloud, and generate a 3D mesh model from the point cloud; Model optimization unit: Smooth the mesh and remove noise.

5. The access control face recognition system based on three-dimensional technology according to claim 4 is characterized in that: The feature extraction unit specifically include: Among them, the Harris corner detection algorithm formula is: R=det(M)-k(t(M)) 2 Where R is the Harris corner response function, M is the autocorrelation matrix, k is a constant, det9M) is the corner strength of the local image area, and t(M) is the total intensity of the local image area.

6. The access control face recognition system based on three-dimensional technology according to claim 5 is characterized in that: The dynamic adjustment module specifically includes: Real-time data acquisition unit: obtains ambient light and angle data from sensors and cameras in real time; Ambient light and angle analysis unit: analyzes the collected ambient light and angle data, evaluates their impact on the recognition effect, and obtains the ambient light adjustment coefficient and angle adjustment coefficient; Dynamic adjustment model unit: based on the analysis results, dynamically adjust the recognition parameters, create and update the real-time dynamic adjustment model; Among them, the dynamic adjustment model formula is: L 新 =L+αA Where, L 新 is the updated dynamic adjustment model, L is the previous dynamic adjustment model, α is the adjustment factor, and A is the adjustment amount calculated based on the current environmental data and target optimization requirements; Information fusion unit: fuses information from different modalities; Among them, the information fusion formula is: J 融 =β1D+β2P In the formula, J 融 is the fused data, D is the data from the visual modality, P is the data from the depth modality, β1 is the weighting coefficient of the visual data, and β2 is the weighting coefficient of the depth data; Optimized output unit: Outputs optimized parameters based on dynamic adjustment model.

7. The access control face recognition system based on three-dimensional technology according to claim 6 is characterized in that: The matching module comprises: Feature extraction unit: extract facial features from the 3D face model; Feature matching unit: matches the extracted features with the 3D face model in the database to identify the identity; Data preprocessing unit: preprocess the input 3D face model; Database update unit: Add new face models and their features to the database.

8. The access control face recognition method based on three-dimensional technology is characterized by: Identification methods include: Use a 3D camera to capture the three-dimensional face data of the target person and obtain detailed facial features; Fusion of 3D face data with traditional 2D image data; Preprocess the fused data, perform facial feature modeling, and generate an accurate 3D face model; Create a real-time dynamic adjustment model, based on which the recognition parameters are automatically optimized under different ambient light and angles; Extract facial features from the 3D face model, match it with the 3D face model in the database, and identify the identity; Authentication is performed based on the matching results, and a decision is made as to whether access control permissions are granted, while access logs are recorded.

9. The access control face recognition method based on three-dimensional technology according to claim 8 is characterized in that: The fusion of 3D face data with traditional 2D image data specifically includes: Data alignment unit: spatially aligns and registers data from different modalities; Fusion strategy unit: weights features from different modalities into a unified representation; Data fusion output unit: outputs the fused features.

10. The access control face recognition method based on three-dimensional technology according to claim 9, characterized in that: Create a real-time dynamic adjustment model. Based on the real-time dynamic adjustment model, output different ambient light and angles to automatically optimize the recognition parameters, including: Get data about ambient light and angle from sensors and cameras in real time; Analyze the collected ambient light and angle data, evaluate their impact on the recognition effect, and obtain the ambient light adjustment coefficient and angle adjustment coefficient; Based on the analysis results, dynamically adjust the recognition parameters, create and update the real-time dynamic adjustment model; Fuse information from different modalities; Output optimized parameters based on dynamic adjustment model.

Images