Multi-mode recognition device for accurately tracking eye movement based on iris recognition

Abstract

The invention discloses a multi-modal recognition device for accurate eye movement tracking based on iris recognition, which relates to the technical field of identity recognition and comprises an equipment composition system and a function analysis system. The equipment composition system comprises a camera module, an infrared illumination module, a motor driving module, an image display module, a tracking and identification unit and a communication interface. The function analysis system comprises an environment adaptation module, a multi-modal fusion analysis module, a user behavior prediction module, a security encryption module and a cloud synchronization module; according to the device, eyeball movement information and iris characteristics are effectively combined, a highly safe and accurate identity verification scheme is provided for a user, and through precise coordination of the infrared illumination module and the high-resolution camera module, it is ensured that iris information can be accurately captured even under changeable light conditions. The eye movement tracking function increases an additional security level, so that the system is difficult to be cheated or counterfeited, and the identification difficulty and the tamper-proofing capability are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of identity recognition technology, and in particular to a multimodal recognition device for accurate eye movement tracking based on iris recognition. Background Art

[0002] In the electronic information age, people's emphasis on identity authentication is obvious to all. A major feature of the electronic information age is the digitization and hidden nature of identity.

[0003] At present, the main means of identity recognition are fingerprint recognition and face recognition. Fingerprint recognition has high accuracy, but it is easy to imitate; face recognition technology has the advantage of being easy to use, but the accuracy is low and often cannot meet the needs of actual applications. Therefore, in response to the above problems, the present invention provides an identification device based on iris recognition to achieve accurate eye tracking. By combining the information of eye movement and the characteristics of the iris, eye tracking and iris recognition are achieved at the same time to improve the accuracy, security and convenience of the existing identification system. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a multimodal recognition device for accurate eye tracking based on iris recognition to solve the problems raised in the above background technology.

[0005] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a multimodal recognition device for accurate eye tracking based on iris recognition, comprising a device component system and a function analysis system, wherein the device component system comprises a camera module, an infrared lighting module, a motor drive module, an image display module, a tracking and recognition unit, and a communication interface;

[0006] The function analysis system includes an environment adaptation module, a multimodal fusion analysis module, a user behavior prediction module, a security encryption module and a cloud synchronization module;

[0007] Each module in the functional analysis system is connected via a high-speed data bus to ensure high efficiency and stability of data transmission;

[0008] The camera module, infrared lighting module and motor drive module interact with the tracking and recognition unit through a real-time control protocol, while the image display module is directly connected to the processing chip of the device to achieve real-time image output, and the communication interface is responsible for data interaction with the external network or HMI interface.

[0009] Further optimizing the technical solution, the camera module includes a first camera module and a second camera module;

[0010] The first camera module is used to capture iris images and is equipped with a high-resolution camera and an infrared filter function to ensure that clear iris images can be obtained under different light conditions;

[0011] The second camera module is used to capture facial images, supports wide-angle shooting, and can perform facial feature analysis.

[0012] To further optimize the technical solution, the infrared illumination module emits infrared light to the eyes of the person being collected, ensuring that the iris image can be collected with high quality even in a dark environment;

[0013] Automatically adjust the intensity of infrared light according to the ambient light intensity to ensure the quality of iris images and prevent overexposure or underexposure;

[0014] A precisely directed light beam ensures that light reaches the user's eyes directly, reducing interference caused by astigmatism.

[0015] To further optimize the technical solution, the motor drive module controls the angle adjustment of the camera module, accurately positions the camera module to the user's eyes, adapts to the needs of different heights and standing positions, and uses an adaptive adjustment algorithm to automatically calculate and adjust the optimal position and angle of the camera based on the user's height and distance data;

[0016] The image display module uses a high-resolution display screen to display the captured iris and face images in real time, making it easy for operators to monitor and adjust;

[0017] The communication interface is configured with an Ethernet interface and supports TCP / IP protocol support, ensuring the connection between the device and the central management system, supporting large data transmission, and enabling the device to interact with data in a network environment; at the same time, built-in HMI integration is closely integrated with the human-machine interface, providing an intuitive operation and management interface, including iris entry, deletion and permission setting.

[0018] To further optimize the technical solution, the tracking and identification unit has a built-in machine learning tracking algorithm and a machine learning identification algorithm;

[0019] Machine learning tracking algorithm, which uses machine vision and machine learning technology to track eye movements in real time, accurately capture the user's gaze point, and provide data support for iris capture;

[0020] Deep learning recognition algorithm uses deep neural network to analyze iris features and improves the accuracy and robustness of the recognition algorithm through a large amount of image training data.

[0021] To further optimize the technical solution, the machine learning tracking algorithm introduces time series analysis to capture the temporal variation characteristics of eye movements, including:

[0022] Temporal convolutional networks, used to process time series data, such as consecutive frames of eye movements;

[0023] Feature extraction, using multiple layers of convolution and pooling layers to extract spatiotemporal features from eye movement video sequences;

[0024] State prediction: using the fully connected layer and Softmax classifier to predict each state of the eyeball;

[0025] Input sequence:

[0026] (X t =[I t-k , ..., I t ])

[0027] Among them, I t is the eye image at time point t;

[0028] Convolutional Layer:

[0029] [C l =ReLU(W l *X t +b l )]

[0030] Among them, * represents the convolution operation, W l and b l are the convolution kernel and bias parameters;

[0031] Fully connected layer output:

[0032] [F=W f Flatten(C L )+b f ]

[0033] Status prediction:

[0034] [P(y t |X t )=Softmax(F)]

[0035] Among them, y t is the predicted eye movement state;

[0036] A grid search is performed on the depth and width of the convolutional layers to optimize the network architecture.

[0037] Use regularization techniques such as Dropout to reduce model overfitting.

[0038] To further optimize the technical solution, the machine learning recognition algorithm adopts a hybrid neural network model including local and global feature extraction to enhance the robustness and accuracy of recognition, including:

[0039] Local feature network, which uses a small convolution window to extract local features of the iris;

[0040] Global feature network, which uses a large convolution window to extract overall iris features;

[0041] Feature fusion and matching, combining local and global features, performing feature fusion through a fully connected layer, and calculating the similarity with the features in the database;

[0042] Local feature extraction:

[0043] [L = ReLU (W l *I+b l )]

[0044] Global feature extraction:

[0045] [G = ReLU (W g *I+b g )]

[0046] Feature Fusion:

[0047] [F=W f Concat(L, G)+b f ]

[0048] Similarity measure:

[0049]

[0050] Among them, F db is a feature in the database;

[0051] Use different types of convolution kernel sizes to experiment with optimal extraction of local and global features;

[0052] Apply Siamese network or triplet loss to enhance the learning of feature matching.

[0053] To further optimize the technical solution, in the functional analysis system:

[0054] Environmental adaptation module, which automatically adjusts the camera's exposure and focus parameters to adapt to different ambient light conditions, ensuring that image quality is always excellent;

[0055] The multimodal fusion analysis module combines iris recognition and facial recognition data to improve the accuracy and security of recognition. It can perform in-depth fusion analysis on the collected data and use composite algorithms to further improve the accuracy of recognition.

[0056] User behavior prediction module, which analyzes the user's eye movement patterns and frequencies, predicts the user's intentions and behaviors, and is used to improve the response speed and accuracy of the interactive system;

[0057] The security encryption module performs end-to-end encryption on all transmitted data to protect the security and privacy of user data;

[0058] The cloud synchronization module allows the device to synchronize iris and face data with the cloud server, facilitating large-scale data analysis and remote identity authentication.

[0059] To further optimize the technical solution, the specific fusion model of the composite algorithm in the multimodal fusion analysis module adopts a weighted average method:

[0060] [S=w iris ×P iris +w face ×P face ]

[0061] Among them, S is the final safety score, P iris and P face are the confidence of iris and face recognition, respectively, w iris and w face It is a weight, which is adjusted according to actual security needs.

[0062] To further optimize the technical solution, the user behavior prediction module predicts the user's intention and behavior by analyzing the user's eye movement pattern and frequency, and the prediction model uses a simplified time series analysis method:

[0063] [B t+1 =αB t +(1-α)X t ]

[0064] Among them, B t+1 is the behavior prediction for the next time step, B t is the behavior at the current time step, X t is the currently observed eye movement data, and α is a smoothing coefficient that is adjusted to optimize prediction accuracy.

[0065] Compared with the prior art, the present invention provides a multimodal recognition device for accurate eye tracking based on iris recognition, which has the following beneficial effects:

[0066] This multimodal recognition device based on iris recognition for precise eye tracking effectively combines eye movement information with iris features, providing users with a highly secure and accurate identity authentication solution. Through the precise coordination of the infrared lighting module and the high-resolution camera module, it ensures that iris information can be accurately captured even under changing light conditions. In addition, the integrated eye tracking function adds an extra layer of security, making the system difficult to be deceived or counterfeited, significantly improving the difficulty of identification and anti-tampering capabilities. By using deep learning algorithms to extract and fuse features, the device not only improves the speed and efficiency of identity authentication, but also enhances the interactivity and convenience of the user interface through real-time eye tracking, making it an ideal choice in a variety of environments with high security requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 A schematic diagram of the system structure of a multimodal recognition device for accurate eye tracking based on iris recognition proposed by the present invention;

[0068] Figure 2 This is a workflow diagram of the multimodal recognition device for accurate eye tracking based on iris recognition proposed by the present invention when in use. DETAILED DESCRIPTION

[0069] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] Embodiment 1:

[0071] See also Figure 1 A multimodal recognition device for accurate eye tracking based on iris recognition includes a device composition system and a function analysis system. The device composition system includes a camera module, an infrared lighting module, a motor drive module, an image display module, a tracking and recognition unit, and a communication interface.

[0072] The camera module includes a first camera module and a second camera module.

[0073] The first camera module is used to capture iris images and is equipped with a high-resolution camera and an infrared filter function to ensure that clear iris images can be obtained under different lighting conditions.

[0074] The second camera module is used to capture facial images, supports wide-angle shooting, and can perform facial feature analysis.

[0075] In this embodiment, the first camera module includes:

[0076] High-resolution camera: Equipped with a high-pixel camera, it can capture subtle iris texture, which is crucial to improving recognition accuracy.

[0077] Infrared filter function: Use infrared filter lenses to filter out non-infrared light to ensure the clarity and contrast of infrared images, so that iris images can be effectively collected under various lighting conditions.

[0078] Automatic focus adjustment: Intelligently adjust the focus to quickly and accurately focus on the user's eyes, shorten recognition time and improve user experience.

[0079] In this embodiment, the second camera module includes:

[0080] Wide-angle lens: supports wider shooting angles, making it easier to capture complete facial images even when the distance to the user is close.

[0081] Facial feature analysis technology: Integrates advanced image processing algorithms to analyze and identify facial features in real time, which can be used to assist in verification and enhance security.

[0082] Infrared lighting module

[0083] Automatic light intensity adjustment: Automatically adjust the intensity of infrared light according to the ambient light intensity to ensure the quality of iris images and prevent overexposure or underexposure.

[0084] Directional beam technology: A precisely directed beam ensures that light reaches the user's eyes directly, reducing interference caused by astigmatism.

[0085] The infrared illumination module emits infrared light to the eyes of the person being collected, ensuring that the iris image can be collected with high quality even in a dark environment.

[0086] The intensity of infrared light is automatically adjusted according to the ambient light intensity to ensure the quality of iris images and prevent overexposure or underexposure.

[0087] A precisely directed light beam ensures that light reaches the user's eyes directly, reducing interference caused by astigmatism.

[0088] The motor drive module controls the angle adjustment of the camera module, accurately positions the camera module at the user's eyes, and adapts to the needs of different heights and standing positions. It also uses an adaptive adjustment algorithm to automatically calculate and adjust the optimal position and angle of the camera based on the user's height and distance data.

[0089] The image display module uses a high-resolution display screen to display the captured iris and face images in real time, which is convenient for operators to monitor and adjust.

[0090] The communication interface is configured with an Ethernet interface and supports TCP / IP protocol support, ensuring the connection between the device and the central management system, supporting large data transmission, and enabling the device to interact with data in a network environment; at the same time, built-in HMI integration is closely integrated with the human-machine interface to provide an intuitive operation and management interface, including iris entry, deletion and permission setting.

[0091] The tracking and identification unit has a built-in machine learning tracking algorithm and a machine learning identification algorithm;

[0092] Machine learning tracking algorithm, which uses machine vision and machine learning technology to track eye movements in real time, accurately capture the user's gaze point, and provide data support for iris capture;

[0093] Deep learning recognition algorithm uses deep neural network to analyze iris features and improves the accuracy and robustness of the recognition algorithm through a large amount of image training data.

[0094] In this embodiment, the machine learning tracking algorithm introduces time series analysis to capture the temporal variation characteristics of eye movements, including:

[0095] Temporal convolutional networks, used to process time series data, such as consecutive frames of eye movements;

[0096] Feature extraction, using multiple layers of convolution and pooling layers to extract spatiotemporal features from eye movement video sequences;

[0097] State prediction: using the fully connected layer and Softmax classifier to predict each state of the eyeball;

[0098] Input sequence:

[0099] (X t =[I t-k , ..., I t ])

[0100] Among them, I t is the eye image at time point t;

[0101] Convolutional Layer:

[0102] [C l =ReLU(W l *X t +b l )]

[0103] Among them, * represents the convolution operation, W l and b l are the convolution kernel and bias parameters;

[0104] Fully connected layer output:

[0105] [F=Wf Flatten(C L )+b f ]

[0106] Status prediction:

[0107] [P(y t |X t )=Softmax(F)]

[0108] Among them, y t is the predicted eye movement state;

[0109] A grid search is performed on the depth and width of the convolutional layers to optimize the network architecture.

[0110] Use regularization techniques such as Dropout to reduce model overfitting.

[0111] In this embodiment, the machine learning recognition algorithm uses a hybrid neural network model that includes local and global feature extraction to enhance the robustness and accuracy of recognition, including:

[0112] Local feature network, which uses a small convolution window to extract local features of the iris;

[0113] Global feature network, which uses a large convolution window to extract overall iris features;

[0114] Feature fusion and matching, combining local and global features, performing feature fusion through a fully connected layer, and calculating the similarity with the features in the database;

[0115] Local feature extraction:

[0116] [L = ReLU (W l *I+b l )]

[0117] Global feature extraction:

[0118] [G = ReLU (W g *I+b g )]

[0119] Feature Fusion:

[0120] [F=W f Concat(L, G)+b f ]

[0121] Similarity measure:

[0122]

[0123] Among them, F db is a feature in the database;

[0124] Use different types of convolution kernel sizes to experiment with optimal extraction of local and global features;

[0125] Apply Siamese network or triplet loss to enhance the learning of feature matching.

[0126] The function analysis system includes an environment adaptation module, a multimodal fusion analysis module, a user behavior prediction module, a security encryption module and a cloud synchronization module.

[0127] The modules in the functional analysis system are connected via a high-speed data bus to ensure the high efficiency and stability of data transmission.

[0128] In the functional analysis system:

[0129] Environmental adaptation module, which automatically adjusts the camera's exposure and focus parameters to adapt to different ambient light conditions, ensuring that image quality is always excellent;

[0130] The multimodal fusion analysis module combines iris recognition and facial recognition data to improve the accuracy and security of recognition. It can perform in-depth fusion analysis on the collected data and use composite algorithms to further improve the accuracy of recognition.

[0131] User behavior prediction module, which analyzes the user's eye movement patterns and frequencies, predicts the user's intentions and behaviors, and is used to improve the response speed and accuracy of the interactive system;

[0132] The security encryption module performs end-to-end encryption on all transmitted data to protect the security and privacy of user data;

[0133] The cloud synchronization module allows the device to synchronize iris and face data with the cloud server, facilitating large-scale data analysis and remote identity authentication.

[0134] In this embodiment, the specific fusion model of the composite algorithm in the multimodal fusion analysis module adopts a weighted average method:

[0135] [S=w iris ×P iris +w face ×P face ]

[0136] Among them, S is the final safety score, P iris and P face are the confidence of iris and face recognition, respectively, w iris and w face It is a weight, which is adjusted according to actual security needs.

[0137] In this embodiment, the user behavior prediction module predicts the user's intention and behavior by analyzing the user's eye movement pattern and frequency. The prediction model uses a simplified time series analysis method:

[0138] [B t+1 =αB t +(1-α)X t ]

[0139] Among them, B t+1 is the behavior prediction for the next time step, B t is the behavior at the current time step, X t is the currently observed eye movement data, and α is a smoothing coefficient that is adjusted to optimize prediction accuracy.

[0140] The camera module, infrared lighting module and motor drive module interact with the tracking and recognition unit through a real-time control protocol, while the image display module is directly connected to the processing chip of the device to achieve real-time image output, and the communication interface is responsible for data interaction with the external network or HMI interface.

[0141] Embodiment 2:

[0142] See also Figure 2 , based on the multimodal recognition device for accurate eye tracking based on iris recognition as described in the first embodiment, the device pre-sets configuration information, the camera module is electrically connected to the tracking and recognition unit, the user's eye movement is tracked in real time, and the user's iris features are captured and photographed at the same time, and the photographed iris features are transmitted to the tracking and recognition unit, and the tracking and recognition unit determines whether the photographed iris features match its preset iris features. If the match is successful, the operation command corresponding to the iris information is executed. If the match fails, the iris information needs to be captured and photographed again to verify.

[0143] By offsetting the infrared lighting module, when performing iris recognition on people of different heights, the infrared lighting module can be deflected to the corresponding angle according to the height of the person to be recognized, so that the infrared light emitted by the infrared lighting module can accurately illuminate the eyes of the person being recognized, thereby facilitating the camera module to accurately capture the iris image of the human eye, obtain highly accurate recognition results, and improve the accuracy of identity verification.

[0144] The beneficial effects of the present invention are:

[0145] This multimodal recognition device based on iris recognition for precise eye tracking effectively combines eye movement information with iris features, providing users with a highly secure and accurate identity authentication solution. Through the precise coordination of the infrared lighting module and the high-resolution camera module, it ensures that iris information can be accurately captured even under changing light conditions. In addition, the integrated eye tracking function adds an extra layer of security, making the system difficult to be deceived or counterfeited, significantly improving the difficulty of identification and anti-tampering capabilities. By using deep learning algorithms to extract and fuse features, the device not only improves the speed and efficiency of identity authentication, but also enhances the interactivity and convenience of the user interface through real-time eye tracking, making it an ideal choice in a variety of environments with high security requirements.

[0146] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0147] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multimodal recognition device for accurate eye tracking based on iris recognition, characterized in that: It includes a device component system and a function analysis system, wherein the device component system includes a camera module, an infrared lighting module, a motor drive module, an image display module, a tracking and recognition unit, and a communication interface; The function analysis system includes an environment adaptation module, a multimodal fusion analysis module, a user behavior prediction module, a security encryption module and a cloud synchronization module; Each module in the functional analysis system is connected via a high-speed data bus to ensure high efficiency and stability of data transmission; The camera module, infrared lighting module and motor drive module interact with the tracking and recognition unit through a real-time control protocol, while the image display module is directly connected to the processing chip of the device to achieve real-time image output, and the communication interface is responsible for data interaction with the external network or HMI interface.

2. The multimodal recognition device for accurate eye tracking based on iris recognition according to claim 1, characterized in that: The camera module includes a first camera module and a second camera module; The first camera module is used to capture iris images and is equipped with a high-resolution camera and an infrared filter function to ensure that clear iris images can be obtained under different light conditions; The second camera module is used to capture facial images, supports wide-angle shooting, and can perform facial feature analysis.

3. The multimodal recognition device for accurate eye tracking based on iris recognition according to claim 1, characterized in that: The infrared illumination module emits infrared light to the eyes of the person being collected, ensuring that the iris image can be collected with high quality even in a dark environment; Automatically adjust the intensity of infrared light according to the ambient light intensity to ensure the quality of iris images and prevent overexposure or underexposure; A precisely directed light beam ensures that light reaches the user's eyes directly, reducing interference caused by astigmatism.

4. The multimodal recognition device for accurate eye tracking based on iris recognition according to claim 1, characterized in that: The motor drive module controls the angle adjustment of the camera module, accurately positions the camera module to the user's eyes, adapts to the needs of different heights and standing positions, and uses an adaptive adjustment algorithm to automatically calculate and adjust the optimal position and angle of the camera based on the user's height and distance data; The image display module uses a high-resolution display screen to display the captured iris and face images in real time, making it easy for operators to monitor and adjust; The communication interface is configured with an Ethernet interface and supports TCP / IP protocol support, ensuring the connection between the device and the central management system, supporting large data transmission, and enabling the device to interact with data in a network environment; at the same time, built-in HMI integration is closely integrated with the human-machine interface to provide an intuitive operation and management interface, including iris entry, deletion and permission setting.

5. The multimodal recognition device for accurate eye tracking based on iris recognition according to claim 1, characterized in that: The tracking and identification unit has a built-in machine learning tracking algorithm and a machine learning identification algorithm; Machine learning tracking algorithm, which uses machine vision and machine learning technology to track eye movements in real time, accurately capture the user's gaze point, and provide data support for iris capture; Deep learning recognition algorithm uses deep neural network to analyze iris features and improves the accuracy and robustness of the recognition algorithm through a large amount of image training data.

6. The multimodal recognition device for accurate eye tracking based on iris recognition according to claim 5, characterized in that: The machine learning tracking algorithm introduces time series analysis to capture the temporal variation characteristics of eye movements, including: Temporal convolutional networks, used to process time series data, such as consecutive frames of eye movements; Feature extraction, using multiple layers of convolution and pooling layers to extract spatiotemporal features from eye movement video sequences; State prediction: using the fully connected layer and Softmax classifier to predict each state of the eyeball; Input sequence: (X t =[I t-k ,...,I t ]) Among them, I t is the eye image at time point t; Convolutional Layer: [C l =ReLU(W l *X t +b l )] Among them, * represents the convolution operation, W l and b l are the convolution kernel and bias parameters; Fully connected layer output: [F=W f ·Flatten(C L )+b f ] Status prediction: [P(y t |X t )Softmax(F)] Among them, y t is the predicted eye movement state; A grid search is performed on the depth and width of the convolutional layers to optimize the network architecture. Use regularization techniques such as Dropout to reduce model overfitting.

7. The multimodal recognition device for accurate eye tracking based on iris recognition according to claim 5, characterized in that: The machine learning recognition algorithm uses a hybrid neural network model that includes local and global feature extraction to enhance the robustness and accuracy of recognition, including: Local feature network, which uses a small convolution window to extract local features of the iris; Global feature network, which uses a large convolution window to extract overall iris features; Feature fusion and matching, combining local and global features, performing feature fusion through a fully connected layer, and calculating the similarity with the features in the database; Local feature extraction: [L=ReLU(W l *I+b l )] Global feature extraction: [G=ReLU(W g *I+b g )] Feature Fusion: [F=W f ·Concat(L,G)+b f ] Similarity measure: Among them, F db is a feature in the database; Use different types of convolution kernel sizes to experiment with optimal extraction of local and global features; Apply Siamese networks or triplet loss to enhance the learning of feature matching.

8. The multimodal recognition device for accurate eye tracking based on iris recognition according to claim 1, characterized in that: In the functional analysis system: Environmental adaptation module, which automatically adjusts the camera's exposure and focus parameters to adapt to different ambient light conditions, ensuring that image quality is always excellent; The multimodal fusion analysis module combines iris recognition and facial recognition data to improve the accuracy and security of recognition. It can perform in-depth fusion analysis on the collected data and use composite algorithms to further improve the accuracy of recognition. User behavior prediction module, which analyzes the user's eye movement patterns and frequencies, predicts the user's intentions and behaviors, and is used to improve the response speed and accuracy of the interactive system; The security encryption module performs end-to-end encryption on all transmitted data to protect the security and privacy of user data; The cloud synchronization module allows the device to synchronize iris and face data with the cloud server, facilitating large-scale data analysis and remote identity authentication.

9. The multimodal recognition device for accurate eye tracking based on iris recognition according to claim 8, characterized in that: The specific fusion model of the composite algorithm in the multimodal fusion analysis module adopts the weighted average method: [S=w iris ×P iris +w face ×P face ] Among them, S is the final safety score, P iris and P face are the confidence of iris and face recognition, respectively, and w iris and w face It is a weight, which is adjusted according to actual security needs.

10. The multimodal recognition device for accurate eye tracking based on iris recognition according to claim 8, characterized in that: The user behavior prediction module predicts the user's intention and behavior by analyzing the user's eye movement pattern and frequency. The prediction model uses a simplified time series analysis method: [B t+1 =αB t +(1-α)X t ] Among them, B t+1 is the behavior prediction for the next time step, B t is the behavior at the current time step, X t is the currently observed eye movement data, and α is a smoothing coefficient that is adjusted to optimize prediction accuracy.

Images