Intelligent biological recognition method and system based on contact lenses
By integrating multiple sensors and networks on contact lenses, extracting and fusing eye biometrics, multiple challenges of existing biometric technologies are solved, and more efficient and accurate biometric effects are achieved.
Patent Information
- Application Number
- CN202510311617.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-17
- Publication Date
- 2025-06-06
AI Technical Summary
Existing biometric technologies face many challenges, including dirt and wear problems in fingerprint recognition, light and expression interference in facial recognition, high equipment costs for iris recognition, environmental noise interference in voiceprint recognition, and environmental sensitivity of venous recognition.
By integrating bioelectric, optical and pressure sensors on contact lenses, obtaining user eye information, pre-processing and extracting bioelectric, corneal and motor characteristics, using BiLSTM and textCNN networks for feature extraction and fusion, building a biometric database for matching, and outputting biometric results.
Improve the accuracy and reliability of biometrics, obtain rich information through multiple sensors, and extract different types of features from multiple networks, which can more comprehensively explore the biometric features of the eye, provide more distinctive feature representations, and achieve more accurate and efficient biometrics.
Smart Images
Figure CN120108052A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of biometric identification technology, and more particularly to an intelligent biometric identification method and system based on contact lenses. Background Art
[0002] Biometric technology refers to the technology of personal identification through the inherent physiological or behavioral characteristics of the human body. In recent years, with the rapid development of computer technology, image processing technology, sensor technology, etc., biometric technology has been widely used. Common biometric technologies include fingerprint recognition, facial recognition, iris recognition, voiceprint recognition, vein recognition, etc. These technologies play an important role in security authentication, personal device unlocking, payment verification, access control system, attendance system and other fields.
[0003] Although existing biometric technologies are widely used, they still face many challenges. Specifically, fingerprint recognition is easily affected by finger dirt and wear, and fingerprint data is easily collected and copied, posing a security risk. Facial recognition has a lower recognition rate under different lighting conditions and is easily interfered by expressions, makeup, and masks, with prominent privacy and security issues. Iris recognition requires active cooperation from users and has high equipment costs, which limits its large-scale application. Voiceprint recognition is easily affected by environmental noise, and sound collection and transmission are prone to distortion. Vein recognition requires specific infrared imaging equipment, is sensitive to the environment, and has low user acceptance.
[0004] Therefore, how to propose a more convenient, safe and efficient contact lens-based intelligent biometric identification method and system is an urgent problem to be solved by those skilled in the art. Summary of the invention
[0005] In view of this, the present invention provides an intelligent biometric identification method and system based on contact lenses.
[0006] In order to achieve the above object, the present invention adopts the following technical solution:
[0007] On the one hand, the present invention provides a contact lens-based intelligent biometric identification method, comprising the following steps:
[0008] Acquiring user eye information according to a sensor integrated on the contact lens, and preprocessing the user eye information to obtain preprocessed data;
[0009] extracting user biometrics based on the preprocessed data;
[0010] Constructing a biometric database containing feature information of authorized users, and matching the user's biometric features with the user's feature information in the biometric database;
[0011] Output the biometric result based on the matching result.
[0012] Preferably, the sensor includes a bioelectric sensor, an optical sensor and a pressure sensor;
[0013] Acquire eye bioelectric signals based on the bioelectric sensor; acquire corneal information based on the optical sensor; acquire eye movement information based on the pressure sensor;
[0014] The preprocessed data includes bioelectric preprocessed data, corneal preprocessed data and motion preprocessed data.
[0015] Preferably, extracting user biometrics based on the preprocessed data includes:
[0016] Dividing the preprocessed data into regions, extracting features from the preprocessed data in each region, and respectively obtaining local bioelectric time domain features, local corneal texture features, and local motion features;
[0017] Fusing the local bioelectric time domain features with the local corneal texture features and local motion features of the corresponding region to obtain regional features;
[0018] Integrate the regional features of adjacent regions to obtain comprehensive features;
[0019] The comprehensive features are input into a neural network algorithm to output the user's biometric features.
[0020] Preferably, the local bioelectric time domain features and the local corneal texture features and local motion features of the corresponding region are fused to obtain regional features, including:
[0021] Align the local bioelectric time domain features, local corneal texture features and local motion features of the same area;
[0022] The concatenate function is used to horizontally splice the aligned local bioelectric time domain features, local corneal texture features and local motion features of the same area to obtain the final regional features.
[0023] Preferably, a BiLSTM network is used to extract local bioelectric time domain features and local motion features; and a textCNN network is used to extract local corneal texture features.
[0024] On the other hand, the present invention also proposes a contact lens-based intelligent biometric identification system, which is used to implement the above-mentioned contact lens-based intelligent biometric identification method, comprising:
[0025] A data processing module, used to obtain user eye information according to a sensor integrated on the contact lens, and pre-process the user eye information to obtain pre-processed data;
[0026] A feature extraction module, used to extract user biometric features based on the preprocessed data;
[0027] A matching module, used to construct a biometric database containing feature information of authorized users, and match the user's biometric features with the user's feature information in the biometric database;
[0028] The biometric recognition module is used to output a biometric recognition result according to the matching result.
[0029] Preferably, the feature extraction module includes:
[0030] A local feature extraction unit, used to divide the preprocessed data into regions, extract features from the preprocessed data in each region, and respectively obtain local bioelectric time domain features, local corneal texture features, and local motion features;
[0031] A feature fusion unit, used for fusing the local bioelectric time domain feature with the local corneal texture feature and the local motion feature of the corresponding area to obtain a regional feature;
[0032] A feature integration unit, used for integrating the regional features of adjacent regions to obtain comprehensive features;
[0033] The biometric feature output unit is used to input the comprehensive feature into the neural network algorithm and output the user's biometric feature.
[0034] Preferably, the feature fusion unit includes:
[0035] A feature alignment subunit, used to align local bioelectric time domain features, local corneal texture features, and local motion features of the same area;
[0036] The feature stitching subunit is used to use the concatenate function to horizontally stitch the aligned local bioelectric time domain features, local corneal texture features and local motion features of the same area to obtain the final regional features.
[0037] It can be seen from the above technical solution that compared with the prior art, the present invention discloses an intelligent biometric identification method and system based on contact lenses, which obtains user eye information through bioelectric, optical and pressure sensors on contact lenses, obtains bioelectric, corneal and motion preprocessing data through preprocessing, divides the preprocessed data by region and extracts local bioelectric time domain features (using BiLSTM network), local corneal texture features (using textCNN network) and local motion features (using BiLSTM network) respectively, aligns the above features of the same region and horizontally splices them through the concatenate function to obtain regional features, integrates the regional features of adjacent regions into comprehensive features, and then inputs the comprehensive features into the neural network algorithm to output user biometric features, and at the same time constructs a biometric feature database containing authorized user feature information, matches the extracted user biometric features with it, and outputs biometric identification results according to the matching results. The present invention makes full use of multiple sensors to obtain rich eye information, extracts different types of local features through different networks, and can more comprehensively explore the eye biometrics, thereby improving the accuracy and reliability of biometrics. The method of integrating multiple features not only takes into account information at different levels of the eye, but also the region division and feature splicing help capture the relationship between regions and the overall eye features, providing a more distinctive feature representation for user identity recognition, thereby achieving more accurate and efficient biometrics. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0039] Figure 1 A flow chart of the method provided by the present invention;
[0040] Figure 2 This is a system architecture diagram provided by the present invention. DETAILED DESCRIPTION
[0041] The following will be combined with the drawings in 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.
[0042] On the one hand, the embodiment of the present invention discloses an intelligent biometric identification method based on contact lenses, such as Figure 1As shown, the method comprises the following steps:
[0043] S1. Obtain user eye information according to a sensor integrated on the contact lens, pre-process the user eye information, and obtain pre-processed data.
[0044] In this embodiment, the sensors include a bioelectric sensor, an optical sensor and a pressure sensor.
[0045] The bioelectric sensor is integrated into the contact lens using micro-nano manufacturing technology, and its electrodes are distributed in the area of the contact lens close to the eye muscles to ensure accurate detection of eye bioelectric signals, including but not limited to retinal potential, eye muscle electrical signals, etc.
[0046] Optical sensors use micro-optical coherence tomography (OCT) or a high-resolution micro-camera to obtain corneal information.
[0047] The pressure sensors use flexible micro-electromechanical systems (MEMS) technology and are distributed on the surface and edges of the contact lenses to sense pressure changes caused by eye movement.
[0048] The preprocessed data includes bioelectric preprocessed data, corneal preprocessed data and motion preprocessed data, and the processing process is as follows:
[0049] 1. Bioelectric preprocessing data:
[0050] First, the collected bioelectric signals are filtered using a bandpass filter to remove high-frequency noise and low-frequency drift, while retaining the signal frequency band related to eye physiological activities.
[0051] The filtered signal is de-averaged to eliminate the DC component of the signal.
[0052] The de-averaged signal is normalized to map the signal amplitude to the range of [-1, 1].
[0053] To handle outliers in the signal, a statistical method was used to consider data points beyond 3 times the standard deviation as outliers and replace them with the average of the adjacent data points.
[0054] 2. Corneal preprocessing data:
[0055] For corneal images, denoising is first performed, and Gaussian filtering or bilateral filtering can be used to remove salt and pepper noise and speckle noise in the image while retaining the corneal texture details.
[0056] Image enhancement is performed by using histogram equalization or adaptive histogram equalization to enhance the contrast between different corneal tissues and make the corneal texture more clearly discernible.
[0057] In order to eliminate the influence of illumination inhomogeneity, an illumination correction algorithm based on local areas is used to ensure uniform grayscale distribution of corneal images.
[0058] 3. Motion preprocessing data:
[0059] The data obtained by the pressure sensor is filtered by sliding average to smooth the pressure signal and reduce the noise caused by sensor jitter or instantaneous interference.
[0060] The first and second derivatives of the pressure signal are calculated to obtain the rate of change of pressure and the acceleration information of the change, thereby reflecting the speed and acceleration of eye movement.
[0061] S2. Extracting user biometrics based on preprocessed data, including:
[0062] S21. Divide the preprocessed data into regions, perform feature extraction on the preprocessed data in each region, and obtain local bioelectric time domain features, local corneal texture features, and local motion features, respectively.
[0063] Specifically, the steps for dividing the bioelectrical and motion data regions are as follows:
[0064] The bioelectric signals and motion data are divided into multiple regions according to the physiological structure of the eye. In this embodiment, the eye area is divided into the upper eyelid region, the lower eyelid region, the medial rectus region, the lateral rectus region, the superior oblique muscle region, and the inferior oblique muscle region according to the anatomical position of the eye muscles.
[0065] The pressure sensor data on the surface of the eyeball is divided into a central area, an upper area, a lower area, a left area and a right area. Each area can be further subdivided according to the geometric characteristics and movement characteristics of the eyeball to better reflect the movement information of different positions.
[0066] The corneal image area is divided as follows:
[0067] The corneal image is divided into concentric ring areas, such as 3 to 5 concentric rings with different radii, which represent different depths and position information of the cornea from the inside to the outside.
[0068] Alternatively, the cornea is divided into four quadrants, and the texture features of each quadrant are extracted respectively to reflect the local differences in corneal texture.
[0069] The BiLSTM network is used to extract local bioelectric time domain features and local motion features. For each bioelectric and motion data area, the time series data is input into the BiLSTM network. The input sequence length of the BiLSTM network is determined according to the data length and feature complexity. The number of hidden layers of the network can be set to 2 to 3 layers, and the number of units in each layer can be adjusted according to the amount of data and task complexity. BiLSTM can effectively capture the front-to-back dependencies of time series data and extract features such as long-term dependent periodic features and trend features. For bioelectric signals, features such as the periodicity of muscle contraction, the frequency and amplitude of action potentials can be extracted; for motion data, the periodic pattern of eye movement, the start and end features of movement, etc. can be extracted.
[0070] The textCNN network is used to extract local corneal texture features. The corneal image area is used as input, and convolution operations are performed using convolution kernels of different sizes (such as 3x3, 5x5, 7x7) and different numbers of convolution layers. A pooling layer (such as maximum pooling or average pooling) is used after each convolution layer to reduce the feature dimension and extract texture features of different scales. The number of convolution kernels and pooling parameters can be adjusted according to the experiment to obtain the best texture feature representation, such as local texture direction, texture density, and texture complexity.
[0071] S22. Fusing the local bioelectric time domain features and the local corneal texture features and local motion features of the corresponding region to obtain regional features, including:
[0072] S231. Align the local bioelectric time domain features, local corneal texture features and local motion features of the same area.
[0073] For each divided area, the temporal and spatial consistency of bioelectric time domain features, corneal texture features and motion features is ensured.
[0074] In terms of time, the features are aligned by timestamp or time index to ensure that they are extracted in the same or similar time window. If the time resolution of different features is different, interpolation methods such as linear interpolation or spline interpolation are used for time synchronization.
[0075] In space, a mapping relationship between different regions is established so that the bioelectric and motion feature regions correspond to the corneal image region. For irregular regions, geometric transformation and coordinate mapping algorithms can be used to unify regions with different features in space.
[0076] S232. Use the concatenate function to horizontally splice the aligned local bioelectric time domain features, local corneal texture features and local motion features of the same area to obtain the final regional features.
[0077] First, adjust different feature vectors to the same dimension. For feature vectors with shorter dimensions, use zero padding or replication to expand the dimension so that its length is the same as the longest feature vector.
[0078] The concatenate function is then used to sequentially concatenate the adjusted feature vectors together to form a longer regional feature vector to fully represent the multimodal biometric features of the region.
[0079] S24. Integrate regional features of adjacent regions to obtain comprehensive features.
[0080] According to the importance of different regions or the relevance of biological features, a deep learning algorithm is used to assign different weights to the features of different regions. The area close to the pupil may be given a higher weight and multiplied by the corresponding weight coefficient during stitching to emphasize the feature contribution of the key area.
[0081] S25. Input the comprehensive features into the neural network algorithm and output the user's biometric features.
[0082] In this embodiment, the neural network algorithm adopts a multi-layer perceptron (MLP). For MLP, the number of neurons in the input layer is the same as the length of the comprehensive feature vector. Multiple hidden layers are set, and the number of neurons in each layer can be adjusted according to the complexity of the task and the amount of data.
[0083] When training a neural network, a large amount of sample data is used to train the network through a back-propagation algorithm and an optimizer (such as the Adam optimizer) in a way that minimizes the loss function (such as cross entropy loss or mean square error loss), and finally outputs a low-dimensional user biometric vector that can highly summarize the user's unique biometric features.
[0084] S3. Build a biometric database containing authorized user feature information, and match the user's biometric features with the user's feature information in the biometric database.
[0085] Build a secure and reliable database to store the biometric vectors of authorized users. For the stored user biometric vectors, use encrypted storage and use hash functions to irreversibly encrypt the feature vectors to ensure user privacy.
[0086] For the new user's biometrics, the most similar feature vector is searched in the database. A similarity metric such as Euclidean distance, cosine similarity or Mahalanobis distance is used to calculate the distance between the new user's biometrics and the feature vectors stored in the database.
[0087] S4. Output the biometric recognition result according to the matching result.
[0088] A similarity threshold is set, which can be adjusted according to experiments and actual application scenarios.
[0089] If the calculated similarity exceeds the threshold, the match is considered successful, and the output recognition result is the corresponding user identity; if it is lower than the threshold, the match is considered failed, and a secondary verification mechanism can be triggered, such as requiring the user to perform additional eye movements or provide other auxiliary information, or denying access, to ensure the security and accuracy of the system.
[0090] On the other hand, reference Figure 2 The present invention also proposes a contact lens-based intelligent biometric identification system, which is used to implement the above-mentioned contact lens-based intelligent biometric identification method, comprising:
[0091] A data processing module, used to obtain user eye information according to the sensor integrated on the contact lens, pre-process the user eye information, and obtain pre-processed data;
[0092] A feature extraction module, for extracting user biometric features based on preprocessed data;
[0093] A matching module, used to build a biometric database containing authorized user feature information, and match the user's biometric features with the user's feature information in the biometric database;
[0094] The biometric recognition module is used to output a biometric recognition result according to the matching result.
[0095] Preferably, the feature extraction module includes:
[0096] A local feature extraction unit is used to divide the preprocessed data into regions, extract features from the preprocessed data in each region, and obtain local bioelectric time domain features, local corneal texture features, and local motion features respectively;
[0097] A feature fusion unit is used to fuse local bioelectric time domain features with local corneal texture features and local motion features of the corresponding area to obtain regional features;
[0098] The feature integration unit is used to integrate the regional features of adjacent regions to obtain comprehensive features;
[0099] The biometric output unit is used to input the comprehensive features into the neural network algorithm and output the user's biometric features.
[0100] Preferably, the feature fusion unit includes:
[0101] A feature alignment subunit, used to align local bioelectric time domain features, local corneal texture features, and local motion features of the same area;
[0102] The feature stitching subunit is used to use the concatenate function to horizontally stitch the aligned local bioelectric time domain features, local corneal texture features and local motion features of the same area to obtain the final regional features.
[0103] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.
[0104] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. An intelligent biometric identification method based on contact lenses, characterized in that: The following steps are involved: Acquiring user eye information according to a sensor integrated on the contact lens, and preprocessing the user eye information to obtain preprocessed data; extracting user biometrics based on the preprocessed data; Constructing a biometric database containing feature information of authorized users, and matching the user's biometric features with the user's feature information in the biometric database; Output the biometric result based on the matching result.
2. The contact lens-based intelligent biometric identification method according to claim 1, characterized in that: The sensors include bioelectric sensors, optical sensors and pressure sensors; Acquire eye bioelectric signals based on the bioelectric sensor; acquire corneal information based on the optical sensor; acquire eye movement information based on the pressure sensor; The preprocessed data includes bioelectric preprocessed data, corneal preprocessed data and motion preprocessed data.
3. The contact lens-based intelligent biometric identification method according to claim 2, characterized in that: Extracting user biometric features based on the preprocessed data includes: Dividing the preprocessed data into regions, extracting features from the preprocessed data in each region, and respectively obtaining local bioelectric time domain features, local corneal texture features, and local motion features; Fusing the local bioelectric time domain features with the local corneal texture features and local motion features of the corresponding region to obtain regional features; Integrate the regional features of adjacent regions to obtain comprehensive features; The comprehensive features are input into a neural network algorithm to output the user's biometric features.
4. The contact lens-based intelligent biometric identification method according to claim 3, characterized in that: The local bioelectric time domain features and the local corneal texture features and local motion features of the corresponding area are integrated to obtain regional features, including: Align the local bioelectric time domain features, local corneal texture features and local motion features of the same area; The concatenate function is used to horizontally splice the aligned local bioelectric time domain features, local corneal texture features and local motion features of the same area to obtain the final regional features.
5. The contact lens-based intelligent biometric identification method according to claim 3, characterized in that: The BiLSTM network is used to extract local bioelectric time domain features and local motion features; the textCNN network is used to extract local corneal texture features.
6. An intelligent biometric recognition system based on contact lenses, characterized in that: include: A data processing module, used to obtain user eye information according to a sensor integrated on the contact lens, and pre-process the user eye information to obtain pre-processed data; A feature extraction module, used to extract user biometric features based on the preprocessed data; A matching module, used to construct a biometric database containing feature information of authorized users, and match the user's biometric features with the user's feature information in the biometric database; The biometric recognition module is used to output a biometric recognition result according to the matching result.
7. The contact lens-based intelligent biometric identification system according to claim 6, characterized in that: The feature extraction module comprises: A local feature extraction unit, used to divide the preprocessed data into regions, extract features from the preprocessed data in each region, and respectively obtain local bioelectric time domain features, local corneal texture features, and local motion features; A feature fusion unit, used for fusing the local bioelectric time domain feature with the local corneal texture feature and the local motion feature of the corresponding area to obtain a regional feature; A feature integration unit, used for integrating the regional features of adjacent regions to obtain comprehensive features; The biometric feature output unit is used to input the comprehensive feature into the neural network algorithm and output the user's biometric feature.
8. The contact lens-based intelligent biometric identification system according to claim 7, characterized in that: The feature fusion unit comprises: A feature alignment subunit, used to align local bioelectric time domain features, local corneal texture features, and local motion features of the same area; The feature stitching subunit is used to use the concatenate function to horizontally stitch the aligned local bioelectric time domain features, local corneal texture features and local motion features of the same area to obtain the final regional features.