CN117315833B - Palm vein recognition module for intelligent door lock and method thereof - Google Patents
Palm vein recognition module for intelligent door lock and method thereof Download PDFInfo
- Publication number
- CN117315833B CN117315833B CN202311274387.5A CN202311274387A CN117315833B CN 117315833 B CN117315833 B CN 117315833B CN 202311274387 A CN202311274387 A CN 202311274387A CN 117315833 B CN117315833 B CN 117315833B
- Authority
- CN
- China
- Prior art keywords
- palm vein
- feature
- training
- blood vessel
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 210000003462 vein Anatomy 0.000 title claims abstract description 227
- 238000000034 method Methods 0.000 title claims abstract description 39
- 230000002792 vascular Effects 0.000 claims abstract description 49
- 238000012549 training Methods 0.000 claims description 88
- 239000013598 vector Substances 0.000 claims description 82
- 239000011159 matrix material Substances 0.000 claims description 42
- 238000000605 extraction Methods 0.000 claims description 31
- 238000013507 mapping Methods 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 6
- 238000012546 transfer Methods 0.000 claims description 4
- 238000011176 pooling Methods 0.000 claims description 2
- 238000012545 processing Methods 0.000 abstract description 17
- 238000005516 engineering process Methods 0.000 abstract description 16
- 238000004422 calculation algorithm Methods 0.000 abstract description 12
- 238000013135 deep learning Methods 0.000 abstract description 5
- 238000012795 verification Methods 0.000 description 24
- 230000008569 process Effects 0.000 description 15
- 238000010586 diagram Methods 0.000 description 9
- 210000004204 blood vessel Anatomy 0.000 description 5
- 230000006698 induction Effects 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000005856 abnormality Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 2
- 239000003795 chemical substances by application Substances 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 230000015654 memory Effects 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000012098 association analyses Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000001815 facial effect Effects 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 230000036620 skin dryness Effects 0.000 description 1
- 230000003238 somatosensory effect Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
- 230000002618 waking effect Effects 0.000 description 1
Abstract
The invention discloses a palm vein recognition module for an intelligent door lock and a method thereof, wherein the palm vein recognition module comprises the following steps: a circuit board; the photosensitive chip is electrically connected to the circuit board; an infrared camera held in a photosensitive path of the photosensitive chip; and a processor. And in the processor, combining a deep learning algorithm and an image processing technology, extracting the characteristics of the palm vein vascular distribution image, and intelligently determining whether the user to be verified is an authorized user.
Description
Technical Field
The invention relates to the technical field of intelligent door locks, in particular to a palm vein recognition module for an intelligent door lock and a method thereof.
Background
Palm vein recognition has higher security than other biometric recognition techniques because palm vein vascularity is unique to everyone and is not susceptible to the external environment.
Along with the rapid development of the fields of intelligent doors and intelligent door locks, the electronic door opening mode gradually replaces the mechanical key door opening mode with the characteristics of simplicity, rapidness and the like. The palm vein recognition technology is combined with the intelligent door and the intelligent door lock, so that a more efficient and safer unlocking mode can be realized, and a better intelligent door lock service is provided for users. However, the palm vein recognition schemes on the market at present have some problems, such as low recognition accuracy, and the like, which limit the application and development of the palm vein recognition technology in the field of intelligent door locks.
Thus, an optimized palm vein identification scheme for intelligent door locks is desired.
Disclosure of Invention
The embodiment of the invention provides a palm vein recognition module for an intelligent door lock and a method thereof, wherein the palm vein recognition module comprises the following steps: a circuit board; the photosensitive chip is electrically connected to the circuit board; an infrared camera held in a photosensitive path of the photosensitive chip; and a processor. And in the processor, combining a deep learning algorithm and an image processing technology, extracting the characteristics of the palm vein vascular distribution image, and intelligently determining whether the user to be verified is an authorized user.
The embodiment of the invention also provides a palm vein recognition module for the intelligent door lock, which comprises:
A circuit board;
The photosensitive chip is electrically connected to the circuit board;
an infrared camera held in a photosensitive path of the photosensitive chip; and
A processor.
The embodiment of the invention also provides a palm vein recognition method for the intelligent door lock, which comprises the following steps:
Acquiring CPU occupancy rate, memory occupancy rate, disk storage amount and network bandwidth value of a plurality of preset time points in a preset time period;
Performing time sequence association analysis on CPU occupancy rate, memory occupancy rate, disk storage amount and network bandwidth value of the plurality of preset time points to obtain an operation state feature vector; and
And determining whether the running state of the multimedia intelligent dispatcher is normal or not based on the running state feature vector.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. In the drawings:
Fig. 1 is a schematic structural diagram of a palm vein recognition module for an intelligent door lock according to an embodiment of the present invention.
Fig. 2 is a schematic diagram of recognition of a recognition module according to an embodiment of the present invention.
Fig. 3 is a block diagram of the processor in the palm vein recognition module for the intelligent door lock according to the embodiment of the invention.
Fig. 4 is a flowchart of a palm vein recognition method for an intelligent door lock according to an embodiment of the present invention.
Fig. 5 is a schematic diagram of a system architecture of a palm vein recognition method for an intelligent door lock according to an embodiment of the present invention.
Fig. 6 is an application scenario diagram of a palm vein recognition module for an intelligent door lock according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the embodiments of the present invention will be described in further detail with reference to the accompanying drawings. The exemplary embodiments of the present invention and their descriptions herein are for the purpose of explaining the present invention, but are not to be construed as limiting the invention.
Unless defined otherwise, all technical and scientific terms used in the embodiments of the application have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the present application is for the purpose of describing particular embodiments only and is not intended to limit the scope of the present application.
In describing embodiments of the present application, unless otherwise indicated and limited thereto, the term "connected" should be construed broadly, for example, it may be an electrical connection, or may be a communication between two elements, or may be a direct connection, or may be an indirect connection via an intermediate medium, and it will be understood by those skilled in the art that the specific meaning of the term may be interpreted according to circumstances.
It should be noted that, the term "first\second\third" related to the embodiment of the present application is merely to distinguish similar objects, and does not represent a specific order for the objects, it is to be understood that "first\second\third" may interchange a specific order or sequence where allowed. It is to be understood that the "first\second\third" distinguishing objects may be interchanged where appropriate such that embodiments of the application described herein may be practiced in sequences other than those illustrated or described herein.
The traditional fingerprint module is a hardware device integrating fingerprint acquisition and recognition functions, and is generally composed of a fingerprint sensor and a related processing chip, and is used for acquiring and processing fingerprint images and comparing the fingerprint images with a pre-stored fingerprint template so as to realize identity authentication or access control. The processing chip of the fingerprint module is responsible for processing and comparing the acquired fingerprint images. Typically including image processing algorithms and fingerprint matching algorithms, fingerprint features may be extracted and compared to stored fingerprint templates to determine the degree of matching of the fingerprints.
Drawbacks of conventional fingerprint modules include:
1. fingerprint identification for wet environment and hand humidity influence
2. Deviation of contact position, resulting in recognition of abnormality
3. The contact area of the fingerprint module is limited, the characteristic points are not as many as the palm, and the safety is not as high as that of the palm vein module
Along with the rapid development of the fields of intelligent doors and intelligent door locks, the electronic door opening mode gradually replaces the mechanical key door opening mode with the characteristics of simplicity, rapidness and the like, and how to rapidly and safely finish identity verification and unlock is a continuously pursued target in the industry.
The invention provides a novel unlocking mode, which adopts a camera to shoot a hand picture, and further completes the identity verification of the palm by extracting and comparing the palm vein and palm print characteristics of the hand, and adopts a millisecond wave radar human body induction sensor for waking up a system.
Because the door lock system is a low-power-consumption system powered by a battery, in order to reduce the power consumption of the system, other peripheral devices of the palm vein recognition system are all in a power-off state, and only the human body induction sensor is in a power-on state. When the human body is detected to be close to the palm vein module, the somatosensory module wakes up the system in a high-level output mode, so that the palm vein module starts the palm identification verification.
The scheme is similar to a fingerprint verification scheme, adopts a non-contact type identification mode and is different from the fingerprint contact type verification scheme.
Palm vein recognition is a biological feature recognition technology, and identity authentication is performed by analyzing and comparing vein distribution patterns in the palm of a human body. The palm vein vascularity of everyone is unique and fixed, similar to fingerprint and iris biometric features, and can be used for authentication and access control. The principle of palm vein recognition is that near infrared rays penetrate through palm skin and are absorbed by blood, and the transmitted rays are absorbed by palm vein blood vessels. By collecting and processing these transmitted light images, an image of the palm vein vascularity can be obtained.
Palm vein recognition has the following advantages over other biometric recognition techniques:
High security: the palm vein vascularity is unique to everyone and is difficult to forge and replicate. Palm vein recognition is more difficult to impersonate or spoof than techniques such as fingerprint, facial recognition, etc.
The external environment is prevented from being influenced: the palmar vein is distributed under the skin and is not easily affected by external environment, such as temperature, humidity, illumination change, etc. Thus, palm vein recognition can provide stable recognition performance under different environments.
Non-contact identification: compared with the technologies such as fingerprint and iris recognition which need contact acquisition, the palm vein recognition can carry out non-contact acquisition through near infrared rays, and is more convenient and sanitary.
And (3) quick identification: the palm vein recognition collection and comparison process is usually fast, can complete recognition within a few seconds, and is suitable for an efficient identity verification scene.
The palm vein recognition technology is combined with the intelligent door and the intelligent door lock, so that a more efficient and safer unlocking mode can be realized, and a better intelligent door lock service is provided for users. As a biological feature recognition technology, the palm vein recognition has higher safety, and the palm vein blood vessel distribution of each person is unique and is not easily influenced by external environment and is difficult to forge or copy. Compared with the traditional unlocking modes such as passwords, keys and the like, the palm vein recognition technology provides higher safety, and effectively prevents unauthorized personnel from entering.
The palm vein recognition technology is used for unlocking, a user only needs to put the palm on the equipment for scanning, a key is not required to be carried or a complex password is not required to be remembered, and the non-contact unlocking mode is very convenient and fast, and is particularly suitable for places needing frequent entry and exit, such as residential communities, office buildings and the like. The palm vein recognition technology has higher recognition speed and accuracy, and can carry out quick and accurate feature extraction and comparison on palm vein distribution images by combining a deep learning algorithm and an image processing technology, so that quick identity verification is realized, the waiting time of a user can be saved, and the unlocking efficiency is improved. The palm vein recognition technology has relatively low requirements on hand gestures and skin states, and can effectively recognize the palm vein regardless of the angles and positions of the palms. Meanwhile, the hand skin care product is not influenced by factors such as the degree of dryness and humidity of the hand skin, the temperature and the like, and has good adaptability. The palm vein recognition technology can record unlocking information of each time, including time, place and the like, and the traceability can provide important reference data for safety management, so that access personnel can be conveniently supervised and managed.
The invention has the beneficial effects that: the identification security level is improved, and the characteristic point information is increased; the problem of contact position deviation and abnormality recognition is solved; the fingerprint identification method solves the problems of fingerprint wetness and abnormal identification.
In an embodiment of the present invention, fig. 1 is a schematic structural diagram of a palm vein recognition module for an intelligent door lock according to an embodiment of the present invention. As shown in fig. 1, a palm vein recognition module 100 for an intelligent door lock according to an embodiment of the present invention includes: a wiring board 1; a photosensitive chip 2 electrically connected to the wiring board 1; an infrared camera 3 held on a photosensitive path of the photosensitive chip 2; and a processor 4.
The module and the door lock control system work cooperatively, as shown in fig. 2, palm information needs to be input before identity verification, the palm information is stored in a local identity verification database, and as a condition of subsequent identity verification, only a person who inputs the palm information can pass the identity verification, and a user who does not input the palm information cannot pass the identity verification. The palm information data is stored in a local database of the identification module, and the personal privacy data is stored locally, so that the personal privacy data is prevented from being diffused.
Under the dormant state, the identification module is in a power-off state, the human body induction sensor and the door lock control system are in a low-power consumption state, when a human body approaches the identification module, the human body induction sensor is triggered to work, the door lock control system is awakened by the electric level, the door lock system supplies power to the identification module, the identification action is started, at the moment, if palm information is detected in an identification area, identity verification is carried out, the acquired palm data is compared with the palm information of a local database, if the acquired palm data is compared with the palm information of the local database, the identity verification is passed, and then the door lock control system is communicated, so that the door opening action is realized.
The invention recognizes by non-contact, reduces sensitivity to wet fingers and positions, has enough information for comparison verification because the detection area is the whole palm, thereby effectively solving the problems of wet fingerprints, deviation of contact positions, few characteristic points and the like, and improving the safety of identity verification.
In the scheme, the recognition system is awakened in a human body induction mode; and the identity verification is completed by identifying the palm vein and the palmprint of the hand.
Here, the palm vein distribution image of the user to be verified is obtained through the infrared camera, and then the palm vein distribution image is received through the photosensitive chip and transmitted to the processor for data processing and analysis, so that the user verification is intelligently completed.
In particular, in order to realize the efficient data processing and analysis of the palm vein vascular distribution image by the processor, the technical conception of the application is as follows: and in the processor, combining a deep learning algorithm and an image processing technology, extracting the characteristics of the palm vein vascular distribution image, and intelligently determining whether the user to be verified is an authorized user.
Fig. 3 is a block diagram of the processor in the palm vein recognition module for the intelligent door lock according to the embodiment of the invention. As shown in fig. 3, the processor 4 includes: an image acquisition module 110, configured to acquire a palm vein distribution image of a user to be authenticated, which is transmitted by the photosensitive chip; a multi-scale feature extraction module 120, configured to perform multi-scale feature extraction on the palm vein blood vessel distribution image to obtain a plurality of palm vein blood vessel distribution feature maps; a context feature extraction module 130, configured to extract context features between the plurality of palm vein blood vessel distribution feature maps to obtain a context palm vein blood vessel distribution feature vector; and an authorized user determining module 140, configured to determine whether the user to be authenticated is an authorized user based on the contextual palm vein vascular distribution feature vector.
In the image acquisition module 110, a photosensitive chip is used to receive and transmit image data, and accuracy and sensitivity of the photosensitive chip are ensured to obtain clear, high-quality palm vein images. In this way, basic data for acquiring a palm vein image of a user is provided, providing the necessary input for subsequent feature extraction and verification. In the multi-scale feature extraction module 120, multi-scale feature extraction may capture different levels of detail and features by analyzing images at different scales, which selects appropriate scales and feature extraction methods to ensure that the extracted features are discriminative and robust. Thus, the diversity analysis of the palm vein vascular distribution is increased, and more comprehensive and rich characteristic information is provided. In the contextual feature extraction module 130, the contextual features may include information such as relevance, relative position, and spatial layout among different feature maps, and suitable contextual feature extraction algorithms are designed to capture relevance and contextual information between features. In this way, a more comprehensive and global feature representation is provided, and the discrimination capability of the palm vein vascularity is enhanced. In the authorized user determination module 140, an accurate authorized user database is established, and a suitable matching algorithm and decision rule are selected to achieve accurate user authentication. Thus, the identity authentication method based on the palm vein blood vessel distribution characteristics is provided, and has higher safety and reliability.
Based on this, in the technical solution of the present application, the encoding process of the processor includes: firstly, palm vein distribution images of users to be verified, which are transmitted by the photosensitive chip, are acquired.
The changes are reflected on different scales, considering that the shape, size, direction, position, etc. of the palmar vein blood vessel are different. For example, the shape of the palmar vein can be seen on a larger scale for its overall curvature and trend, and on a smaller scale for its local detail and texture. Therefore, in the technical scheme of the application, the palm vein blood vessel distribution image is subjected to multi-scale feature extraction to obtain a plurality of palm vein blood vessel distribution feature images.
The palm vein vascularity image may display a vascularity pattern inside the palm of the user, each person's vascularity pattern being unique, similar to a fingerprint or iris, and thus may be used for identity authentication and individual identification. Through palm vein vascularity images, the shape, curve and branch structure of the blood vessel can be observed and analyzed, and the information can be used for blood vessel feature extraction and matching for user identity verification. The vascularity image also provides information on the density and spacing of the blood vessels, which can be used to determine the details and characteristics of the palmar vein image, further enhancing the ability to discriminate the identity of the user. The vascularity image may show the texture features of the blood vessel, such as the texture and texture variations of the vessel wall, which may be used to enhance the uniqueness and discrimination of the palmar vein image. The palm vein angiogram also provides the context environment information of the palm, such as fingers, palm edges and the like, and the context information can be combined with the blood vessel characteristics, so that the accuracy of user identity authentication is further improved.
By extracting and analyzing the useful information, a palm vein feature model of the user can be established and compared with a pre-stored authorized user model to determine whether the user to be authenticated is an authorized user. The identity authentication method based on the palm vein vascularity image has high safety and identification degree, and is not influenced by external factors (such as skin dryness, wound and the like).
The palm vein blood vessel distribution map of the user to be verified, which is transmitted by the photosensitive chip, plays an important role in finally determining whether the user to be verified is an authorized user. Firstly, ensuring that the photosensitive chip can accurately capture and transmit the palm vein vascular distribution image, the image quality is important for subsequent feature extraction and verification, so that attention is required to factors such as illumination conditions, definition, noise and the like so as to obtain a high-quality image. Then, it is ensured that the photosensitive chip can accurately transmit the palm vein vascular distribution image of the user to be verified, any data transmission error or distortion may cause misjudgment of the verification result, and therefore appropriate measures need to be taken to ensure the accuracy and integrity of the data. When acquiring the palm vein image, the palm needs to be ensured to be correctly placed by the user and the stability of the palm needs to be maintained, and the palm can be realized through proper guidance and indication, so that the acquired image can accurately reflect the palm vein blood vessel distribution of the user.
By acquiring the palm vein vascularity image, subsequent feature extraction can be performed, and the features can include information of the shape, density, direction and the like of the vascularity, so as to construct a palm vein feature model of the user. Based on the acquired palm vein vascular distribution image, the authentication of the authorized user can be performed, and whether the user to be authenticated is the authorized user can be determined by comparing the palm vein characteristics of the user to be authenticated with the characteristic model of the authorized user. The identity verification mode has high safety and accuracy, and can effectively prevent impersonation and fraudulent conduct.
Acquiring the palm vein vascular distribution image of the user to be authenticated, which is transmitted by the photosensitive chip, is important for finally determining whether the user to be authenticated is an authorized user. Basic data required for constructing a characteristic model and carrying out identity verification are provided, and the accuracy and the reliability of the whole palm vein recognition system are ensured.
In a specific example of the present application, the multi-scale feature extraction module includes: and the image feature extraction unit is used for enabling the palm vein blood vessel distribution image to pass through an image feature extractor based on a pyramid network so as to obtain the plurality of palm vein blood vessel distribution feature graphs.
The pyramid network-based image feature extractor adopts four different scales of maximum pooling operation on the palm vein blood vessel distribution image to obtain the plurality of palm vein blood vessel distribution feature graphs, wherein the four different scales are 13x13, 9x9, 5x5 and 1x1 respectively.
A Pyramid Network (Pyramid Network) is a deep neural Network structure for extracting features of an image, and a Pyramid-shaped Network is constructed by extracting features on different scales, so that information of different levels and scales in the image is captured. The main idea of a pyramid network is to process an input image through feature maps of multiple resolutions, at each resolution the network will extract features of different scales, from coarse to fine, forming a hierarchical structure that captures object shape, texture and context information at different scales.
Pyramid networks are typically composed of multiple branches, each with a different receptive field and output resolution. Lower level branches handle larger scale features, while higher level branches handle smaller scale features. These branches may share some low-level features to improve the efficiency of feature utilization.
Specifically, by constructing an image pyramid, palm vein vascularity images can be acquired at different scales, which can be used for feature extraction at different resolutions to obtain more comprehensive and rich palm vein features. By extracting features on different branches in the pyramid network, multiple palm vein vascularity feature maps can be obtained, which can contain information of different scales and levels for further feature representation and matching.
The pyramid network has the advantage that it can process multi-scale information, thereby improving the expressive power and robustness of features. Through the image feature extractor of the pyramid network, the features in the palm vein blood vessel distribution image can be more comprehensively captured, and the accuracy and the reliability of the palm vein recognition system are improved.
Then, extracting the context features among the plurality of palm vein blood vessel distribution feature maps to obtain a context palm vein blood vessel distribution feature vector. That is, the association relationship between the respective local palm vein blood vessel distribution features is established such that the contextual palm vein blood vessel distribution feature vector contains the overall feature expression of the palm vein blood vessel distribution.
In a specific example of the present application, the context feature extraction module is configured to extract context features between the plurality of palm vein blood vessel distribution feature maps to obtain a coding process of a context palm vein blood vessel distribution feature vector, and the coding process includes: firstly, the palm vein blood vessel distribution feature images are respectively passed through a feature full-perception module based on a full-connection layer to obtain a plurality of palm vein blood vessel distribution full-perception feature vectors; and then the palm vein blood vessel distribution full-perception feature vectors are processed through a Bi-LSTM-based image local feature context encoder to obtain the context palm vein blood vessel distribution feature vectors.
Bi-LSTM (two-way long and short term memory network) based image local feature context encoder is a neural network structure for image feature encoding and context modeling, which can be modeled using the temporal processing capabilities of Bi-LSTM to capture context information between features. Bi-LSTM is a variant of a recurrent neural network (Recurrent Neural Network, RNN) that can take into account both forward and backward context information when processing sequence data. In image processing, the image features may be divided into partial blocks in a certain order, and the features of each partial block may be used as input of Bi-LSTM.
Specifically, first, local features are extracted from a plurality of palm vein vascularity feature maps, and these features may be pixel values of a specific region or feature representations obtained through convolution operation. The extracted local feature sequence is then input into Bi-LSTM. The Bi-LSTM processes the feature blocks one by one in sequence and outputs a hidden state vector at each time step. Since Bi-LSTM is Bi-directional, both the forward and backward contexts of the current feature block are considered. Finally, in each time step of Bi-LSTM, the hidden state vectors capture the context information of the current feature block, and these hidden state vectors can be used as the context representation of the local features to form the context palm vein vascular distribution feature vector.
The image local feature context encoder based on Bi-LSTM can conduct context modeling on the palm vein vascular distribution features, so that the dependency relationship between the features and the context information are captured. This helps to improve the accuracy and robustness of the palmar vein recognition, especially in the presence of noise or variations.
And then, the contextual palm vein vascular distribution feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the user to be verified is an authorized user. The training process of the classifier is to learn a decision boundary capable of distinguishing different categories by optimizing a loss function by using the known palm vein vascular distribution feature vectors of the authorized user and the unauthorized user as training data. The prediction process of the classifier is to input the palm vein distribution feature vector of the user to be verified into the classifier, and output which category the user belongs to according to the decision boundary, namely whether the user is an authorized user or not.
In one embodiment of the present application, the authorized user determining module is configured to: and the contextual palm vein vascular distribution feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the user to be verified is an authorized user.
The classification of the contextual palm vein vascular distribution feature vector by the classifier can be used for judging whether the user to be verified is an authorized user or not. By training a classifier, it is possible to learn to map feature vectors of different users to corresponding categories, such as "authorized users" or "unauthorized users".
Such classifiers may be implemented using various machine learning algorithms or deep learning models, such as support vector machines (Support Vector Machine, SVM), random Forest (Random Forest), multi-layer perceptrons (Multilayer Perceptron), and the like. These classifiers can be trained based on feature vectors and corresponding labels in the training data to build a classification model. The contextual palm vein distribution feature vector of the user to be verified is input into a trained classifier, and the classifier predicts the category of the feature vector according to the learned mode and rule. If the classification result is "authorized user", it means that the user to be authenticated is considered to be a legal authorized user.
In one embodiment of the present application, the palm vein recognition module for intelligent door lock further includes a training module for training the pyramid network-based image feature extractor, the full-connection layer-based feature full-perception module, the Bi-LSTM-based image local feature context encoder, and the classifier; wherein, training module includes: the training data acquisition unit is used for acquiring training data, wherein the training data comprises training palm vein vascular distribution images of a user to be verified, which are transmitted by the photosensitive chip, and whether the user to be verified is a true value of an authorized user or not; the training image feature extraction unit is used for enabling the training palm vein blood vessel distribution image to pass through the pyramid network-based image feature extractor to obtain a plurality of training palm vein blood vessel distribution feature graphs; the training feature full-perception unit is used for enabling the training palm vein blood vessel distribution feature graphs to respectively pass through the feature full-perception module based on the full-connection layer so as to obtain a plurality of training palm vein blood vessel distribution full-perception feature vectors; the training context coding unit is used for enabling the training palm vein vascular distribution full-perception feature vectors to pass through the Bi-LSTM-based image local feature context coder to obtain training context palm vein vascular distribution feature vectors; the training classification unit is used for enabling the training context palm vein vascular distribution feature vectors to pass through a classifier to obtain a classification loss function value; and a training unit, configured to train the pyramid network-based image feature extractor, the full-connection layer-based feature full-perception module, the Bi-LSTM-based image local feature context encoder, and the classifier with the classification loss function value, where, in each iteration of the training, a weight matrix that performs domain mapping on the training context palm vein distribution feature vector performs a weight space exploration constraint iteration based on regularization of a class matrix.
In the technical scheme of the application, after the training palm vein blood vessel distribution image passes through the image feature extractor based on the pyramid network, the obtained plurality of training palm vein blood vessel distribution feature images can respectively express image semantic features under different depths of image feature association scales based on the pyramid network, so that when the plurality of training palm vein blood vessel distribution feature images respectively pass through the feature full-perception module based on the full-connection layer to obtain a plurality of training palm vein blood vessel distribution full-perception feature vectors, and the plurality of training palm vein blood vessel distribution full-perception feature vectors pass through the image local feature context encoder based on the Bi-LSTM to obtain training palm vein blood vessel distribution feature vectors, the training palm vein blood vessel distribution feature vectors can express short-distance-long-distance dual-context associated image semantic features of cross-scale and cross-depth, and then the training palm vein blood vessel distribution feature vectors pass through the classifier to be classified, so that the probability distribution corresponding to the label can be obtained based on the space domain from the image feature space serving as the semantic feature regression process of the image semantic feature to the probability distribution space. Here, considering that the training context palm vein distribution feature vector synchronously expresses the cross-scale and cross-depth image semantic features and the high-order context association features thereof, the label distribution enrichment corresponding to the feature distribution diversification of different image semantic distribution dimensions in the probability distribution domain of the classification result may be caused in the space domain mapping process, so that the mapping convergence effect to the probability distribution space in the classification process is affected.
Based on the above, in the training process of the contextual palm vein vascular distribution feature vector through the classifier, the applicant performs weight space exploration constraint based on regularization of a class matrix on a weight matrix for performing domain mapping on the contextual palm vein vascular distribution feature vector to obtain an improved semantic understanding feature vector of a potential invader behavior pattern, which is specifically expressed as follows: performing weight space exploration constraint iteration based on regularization of a class matrix on a weight matrix of the training context palm vein vascular distribution feature vector in a domain mapping mode according to the following optimization formula; wherein, the optimization formula is: Wherein/> The training context palm vein vascular distribution feature vector is specifically expressed as a column vector,/>Is an improved training context semantic pollution state feature vector, which is a row vector,/>Is a domain transfer matrix which can be learned, and/>Is a learnable weight matrix,/>For the weight matrix of the last iteration,/>Is a weight matrix after iteration,/>Transposed vector or transposed matrix representing vector or matrix,/>Representing a matrix multiplication.
Here, the weight space domain of the weight matrix and the palm vein vascular distribution feature vector of the training context are consideredDomain differences (domain gap) between probability distribution domains of the domain mapping result of (a) by weight matrix/>Palm vein vascular distribution feature vector/>, relative to the training contextThe class matrix regularization representation of (3) is used as an inter-domain migration agent (inter-domain TRANSFERRING AGENT) to transfer the probability distribution of the valuable weight constraint into the weight space, so that excessive exploration (over-explloit) of the weight distribution in the weight space by the tag distribution enriched (rich label distributed) probability distribution domain in the domain mapping process based on the weight space is avoided, the mapping convergence effect to the probability distribution space in the classification process is improved, and the training effect of the classifier is also improved.
In summary, the palm vein recognition module 100 for an intelligent door lock according to an embodiment of the present invention is illustrated, in which, in the processor, the feature extraction is performed on the palm vein distribution image in combination with the deep learning algorithm and the image processing technology, and whether the user to be authenticated is an authorized user is intelligently determined.
In one embodiment of the present invention, fig. 4 is a flowchart of a method for palm vein recognition for an intelligent door lock according to an embodiment of the present invention. Fig. 5 is a schematic diagram of a system architecture of a palm vein recognition method for an intelligent door lock according to an embodiment of the present invention. As shown in fig. 4 and 5, a palm vein recognition method for an intelligent door lock according to an embodiment of the present invention includes: 210, acquiring a palm vein vascular distribution image of a user to be verified, which is transmitted by the photosensitive chip; 220, performing multi-scale feature extraction on the palm vein blood vessel distribution image to obtain a plurality of palm vein blood vessel distribution feature images; 230, extracting contextual features between the plurality of palm vein vascularity feature maps to obtain contextual palm vein vascularity feature vectors; and 240, determining whether the user to be verified is an authorized user based on the contextual palm vein vascularity feature vector.
In the palm vein recognition method for the intelligent door lock, the multi-scale feature extraction is performed on the palm vein distribution image to obtain a plurality of palm vein distribution feature images, including: and passing the palm vein blood vessel distribution image through an image feature extractor based on a pyramid network to obtain a plurality of palm vein blood vessel distribution feature graphs.
It will be appreciated by those skilled in the art that the specific operations of the respective steps in the above-described palm vein recognition method for an intelligent door lock have been described in detail above with reference to the description of the palm vein recognition module for an intelligent door lock of fig. 1 to 3, and thus, repetitive descriptions thereof will be omitted.
Fig. 6 is an application scenario diagram of a palm vein recognition module for an intelligent door lock according to an embodiment of the present invention. As shown in fig. 6, in this application scenario, first, a palm vein vascularity image of a user to be authenticated transmitted by the photosensitive chip is acquired (e.g., C as illustrated in fig. 6); the acquired palm vein profile image is then input into a server (e.g., S as illustrated in fig. 6) deployed with a palm vein recognition algorithm for the smart door lock, wherein the server is capable of processing the palm vein profile image based on the palm vein recognition algorithm for the smart door lock to determine whether the user to be authenticated is an authorized user.
The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the invention, and is not meant to limit the scope of the invention, but to limit the invention to the particular embodiments, and any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (7)
1. A palm vein recognition module for intelligent lock, its characterized in that includes:
A circuit board;
The photosensitive chip is electrically connected to the circuit board;
an infrared camera held in a photosensitive path of the photosensitive chip; and
A processor;
Wherein the processor comprises:
the image acquisition module is used for acquiring palm vein vascular distribution images of the user to be verified, which are transmitted by the photosensitive chip;
the multi-scale feature extraction module is used for carrying out multi-scale feature extraction on the palm vein blood vessel distribution image so as to obtain a plurality of palm vein blood vessel distribution feature images;
The context feature extraction module is used for extracting context features among the plurality of palm vein vascular distribution feature graphs to obtain a context palm vein vascular distribution feature vector; and
The authorized user determining module is used for determining whether the user to be verified is an authorized user or not based on the contextual palm vein vascular distribution feature vector;
the system further comprises a training module for training an image feature extractor based on a pyramid network, a feature full-perception module based on a full-connection layer, an image local feature context encoder based on Bi-LSTM and a classifier;
Wherein, training module includes:
the training data acquisition unit is used for acquiring training data, wherein the training data comprises training palm vein vascular distribution images of a user to be verified, which are transmitted by the photosensitive chip, and whether the user to be verified is a true value of an authorized user or not;
the training image feature extraction unit is used for enabling the training palm vein blood vessel distribution image to pass through the pyramid network-based image feature extractor to obtain a plurality of training palm vein blood vessel distribution feature graphs;
The training feature full-perception unit is used for enabling the training palm vein blood vessel distribution feature graphs to respectively pass through the feature full-perception module based on the full-connection layer so as to obtain a plurality of training palm vein blood vessel distribution full-perception feature vectors;
the training context coding unit is used for enabling the training palm vein vascular distribution full-perception feature vectors to pass through the Bi-LSTM-based image local feature context coder to obtain training context palm vein vascular distribution feature vectors;
the training classification unit is used for enabling the training context palm vein vascular distribution feature vectors to pass through a classifier to obtain a classification loss function value; and
The training unit is used for training the pyramid network-based image feature extractor, the full-connection layer-based feature full-perception module, the Bi-LSTM-based image local feature context encoder and the classifier according to the classification loss function value, wherein in each round of iteration of the training, a weight matrix based on regularization of a class matrix is performed on a weight space exploration constraint iteration of the training context palm vein vascular distribution feature vector;
In each iteration of the training, performing a weighted space exploration constraint iteration based on regularization of a class matrix on a weight matrix of domain mapping of the training context palm vein vascular distribution feature vector, wherein the method comprises the following steps of: performing weight space exploration constraint iteration based on regularization of a class matrix on a weight matrix of the training context palm vein vascular distribution feature vector in a domain mapping mode according to the following optimization formula;
wherein, the optimization formula is:
Wherein V is the palm vein vascular distribution feature vector of the training context, specifically expressed as a column vector, V' is the semantic pollution state feature vector of the improved training context, as a row vector, M t∈RL×L is a domain transfer matrix capable of learning, and For a learnable weight matrix, M is the weight matrix of the last iteration, M' is the weight matrix after the iteration, (. Cndot.) T represents the transposed vector or transposed matrix of the vector or matrix,/>Representing a matrix multiplication.
2. The palm vein recognition module for an intelligent door lock of claim 1, wherein the multi-scale feature extraction module comprises:
and the image feature extraction unit is used for enabling the palm vein blood vessel distribution image to pass through an image feature extractor based on a pyramid network so as to obtain the plurality of palm vein blood vessel distribution feature graphs.
3. The palm vein recognition module for an intelligent door lock according to claim 2, wherein the pyramid network-based image feature extractor applies four different scales of maximum pooling operations to the palm vein blood vessel distribution image to obtain the plurality of palm vein blood vessel distribution feature maps, wherein the four different scales are 13x13, 9x9, 5x5, 1x1, respectively.
4. The palm vein recognition module for an intelligent door lock of claim 3, wherein the contextual feature extraction module is configured to:
the palm vein blood vessel distribution feature images are respectively passed through a feature full-perception module based on a full-connection layer to obtain a plurality of palm vein blood vessel distribution full-perception feature vectors; and
And passing the plurality of palm vein blood vessel distribution full-perception feature vectors through a Bi-LSTM-based image local feature context encoder to obtain the contextual palm vein blood vessel distribution feature vectors.
5. The palm vein recognition module for an intelligent door lock of claim 4, wherein the authorized user determination module is configured to:
and the contextual palm vein vascular distribution feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the user to be verified is an authorized user.
6. A palm vein recognition method for an intelligent door lock, comprising:
acquiring a palm vein vascular distribution image of a user to be verified, which is transmitted by the photosensitive chip;
Performing multi-scale feature extraction on the palm vein blood vessel distribution image to obtain a plurality of palm vein blood vessel distribution feature images;
Extracting contextual features among the plurality of palm vein vascular distribution feature maps to obtain contextual palm vein vascular distribution feature vectors; and
Determining whether the user to be verified is an authorized user or not based on the contextual palm vein vascular distribution feature vector;
Wherein, still include: training an image feature extractor based on a pyramid network, a feature full-perception module based on a full-connection layer, an image local feature context encoder based on Bi-LSTM and a classifier;
The training method for the image feature extractor based on the pyramid network, the feature full-perception module based on the full-connection layer, the Bi-LSTM-based image local feature context encoder and the classifier comprises the following steps of:
Acquiring training data, wherein the training data comprises training palm vein vascular distribution images of a user to be verified, which are transmitted by the photosensitive chip, and whether the user to be verified is a true value of an authorized user or not;
Passing the training palm vein blood vessel distribution image through the pyramid network-based image feature extractor to obtain a plurality of training palm vein blood vessel distribution feature graphs;
Respectively passing the training palm vein blood vessel distribution feature maps through the feature full-perception module based on the full-connection layer to obtain a plurality of training palm vein blood vessel distribution full-perception feature vectors;
Passing the training palm vein blood vessel distribution full-perception feature vectors through the Bi-LSTM-based image local feature context encoder to obtain training context palm vein blood vessel distribution feature vectors;
the palm vein distribution feature vector of the training context passes through a classifier to obtain a classification loss function value;
Training the pyramid network-based image feature extractor, the full-connection layer-based feature full-perception module, the Bi-LSTM-based image local feature context encoder and the classifier by using the classification loss function values, wherein in each round of iteration of the training, a weight matrix for performing domain mapping on the training context palm vein vascular distribution feature vector is subjected to weight space exploration constraint iteration based on regularization of a class matrix;
In each iteration of the training, performing a weighted space exploration constraint iteration based on regularization of a class matrix on a weight matrix of domain mapping of the training context palm vein vascular distribution feature vector, wherein the method comprises the following steps of: performing weight space exploration constraint iteration based on regularization of a class matrix on a weight matrix of the training context palm vein vascular distribution feature vector in a domain mapping mode according to the following optimization formula;
wherein, the optimization formula is:
Wherein V is the palm vein vascular distribution feature vector of the training context, specifically expressed as a column vector, V' is the semantic pollution state feature vector of the improved training context, as a row vector, M t∈RL×L is a domain transfer matrix capable of learning, and For a learnable weight matrix, M is the weight matrix of the last iteration, M' is the weight matrix after the iteration, (. Cndot.) T represents the transposed vector or transposed matrix of the vector or matrix,/>Representing a matrix multiplication.
7. The method for intelligent door lock palm vein recognition according to claim 6, wherein the multi-scale feature extraction of the palm vein blood vessel distribution image to obtain a plurality of palm vein blood vessel distribution feature maps comprises:
And passing the palm vein blood vessel distribution image through an image feature extractor based on a pyramid network to obtain a plurality of palm vein blood vessel distribution feature graphs.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311274387.5A CN117315833B (en) | 2023-09-28 | Palm vein recognition module for intelligent door lock and method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311274387.5A CN117315833B (en) | 2023-09-28 | Palm vein recognition module for intelligent door lock and method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117315833A CN117315833A (en) | 2023-12-29 |
| CN117315833B true CN117315833B (en) | 2024-06-04 |
Family
ID=
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103793692A (en) * | 2014-01-29 | 2014-05-14 | 五邑大学 | Low-resolution multi-spectral palm print and palm vein real-time identity recognition method and system |
| CN108985231A (en) * | 2018-07-12 | 2018-12-11 | 广州麦仑信息科技有限公司 | A kind of vena metacarpea feature extracting method based on multiple dimensioned convolution kernel |
| CN109034034A (en) * | 2018-07-12 | 2018-12-18 | 广州麦仑信息科技有限公司 | A kind of vein identification method based on nitrification enhancement optimization convolutional neural networks |
| CN110321870A (en) * | 2019-07-11 | 2019-10-11 | 西北民族大学 | A kind of vena metacarpea recognition methods based on LSTM |
| CN110321871A (en) * | 2019-07-11 | 2019-10-11 | 电子科技大学成都学院 | A kind of palm vein identification system and method based on LSTM |
| CN111354118A (en) * | 2018-12-21 | 2020-06-30 | 广州市天汇计算机科技有限公司 | Door handle with palm vein recognition function and application method |
| KR102375593B1 (en) * | 2021-08-26 | 2022-03-17 | 전북대학교산학협력단 | Apparatus and method for authenticating user based on a palm composite image |
| CN116189319A (en) * | 2022-12-15 | 2023-05-30 | 云丁网络技术(北京)有限公司 | Identification module, intelligent door lock and unlocking control method |
| CN116620219A (en) * | 2023-07-21 | 2023-08-22 | 宁波芯然科技有限公司 | Low-power-consumption vehicle-mounted palm vein unlocking module and unlocking method |
| CN116631068A (en) * | 2023-07-25 | 2023-08-22 | 江苏圣点世纪科技有限公司 | Palm vein living body detection method based on deep learning feature fusion |
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103793692A (en) * | 2014-01-29 | 2014-05-14 | 五邑大学 | Low-resolution multi-spectral palm print and palm vein real-time identity recognition method and system |
| CN108985231A (en) * | 2018-07-12 | 2018-12-11 | 广州麦仑信息科技有限公司 | A kind of vena metacarpea feature extracting method based on multiple dimensioned convolution kernel |
| CN109034034A (en) * | 2018-07-12 | 2018-12-18 | 广州麦仑信息科技有限公司 | A kind of vein identification method based on nitrification enhancement optimization convolutional neural networks |
| CN111354118A (en) * | 2018-12-21 | 2020-06-30 | 广州市天汇计算机科技有限公司 | Door handle with palm vein recognition function and application method |
| CN110321870A (en) * | 2019-07-11 | 2019-10-11 | 西北民族大学 | A kind of vena metacarpea recognition methods based on LSTM |
| CN110321871A (en) * | 2019-07-11 | 2019-10-11 | 电子科技大学成都学院 | A kind of palm vein identification system and method based on LSTM |
| KR102375593B1 (en) * | 2021-08-26 | 2022-03-17 | 전북대학교산학협력단 | Apparatus and method for authenticating user based on a palm composite image |
| CN116189319A (en) * | 2022-12-15 | 2023-05-30 | 云丁网络技术(北京)有限公司 | Identification module, intelligent door lock and unlocking control method |
| CN116620219A (en) * | 2023-07-21 | 2023-08-22 | 宁波芯然科技有限公司 | Low-power-consumption vehicle-mounted palm vein unlocking module and unlocking method |
| CN116631068A (en) * | 2023-07-25 | 2023-08-22 | 江苏圣点世纪科技有限公司 | Palm vein living body detection method based on deep learning feature fusion |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Radzi et al. | IoT based facial recognition door access control home security system using raspberry pi | |
| Basheer et al. | FESD: An approach for biometric human footprint matching using fuzzy ensemble learning | |
| Hussain et al. | Internet of things with deep learning-based face recognition approach for authentication in control medical systems | |
| Kamaraju et al. | Wireless fingerprint attendance management system | |
| Jung et al. | Fingerprint liveness detection by a template-probe convolutional neural network | |
| Nagwanshi et al. | Estimation of centroid, ensembles, anomaly and association for the uniqueness of human footprint features | |
| Zhang et al. | Gait learning based authentication for intelligent things | |
| Evangelin et al. | Biometric authentication of physical characteristics recognition using artificial neural network with PSO algorithm | |
| Srinivas et al. | Artificial intelligence based optimal biometric security system using palm veins | |
| Daramola et al. | Algorithm for fingerprint verification system | |
| CN117315833B (en) | Palm vein recognition module for intelligent door lock and method thereof | |
| Manocha et al. | Palm vein recognition for human identification using NN | |
| CN117315833A (en) | Palm vein recognition module for intelligent door lock and method thereof | |
| Chaitanya et al. | A human gait recognition against information theft in smartphone using residual convolutional neural network | |
| CN113190819A (en) | Authentication method and device based on living body fusion characteristics and computer equipment | |
| Liu | Fingerprint analysis and singular point definition by deep neural network | |
| Videkar et al. | Finger vein identification based on minutiae feature extraction with spurious minutiae removal | |
| Cenys et al. | Genetic algorithm based palm recognition method for biometric authentication systems | |
| Fadlil et al. | Room Security System Using Machine Learning with Face Recognition Verification. | |
| CN111709312A (en) | Local feature face recognition method based on joint main mode | |
| Li et al. | Palmprint matching using line features | |
| Sagiroglu et al. | Artificial neural network based automatic face parts prediction system from only fingerprints | |
| Ozkaya et al. | Intelligent face border generation system from fingerprints | |
| Hashmi et al. | Accessnet: a three layered visual based access authentication system for restricted zones | |
| Yashashwini et al. | Face Recognition Based Smart Door Lock System using Convolution Neural Network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |