Distributed storage method for face image data

Abstract

The invention discloses a face image data distributed storage method, belongs to the field of artificial intelligence, and aims to realize accurate identity verification and state monitoring of a driver by capturing and processing a face image of the driver in real time in a distributed manner. The method comprises the following steps: capturing a driver face image in real time through front-end acquisition equipment; an edge computing node is used for judging the illumination condition of the captured face image, dynamic normalization processing is carried out according to the illumination condition, adaptive graying is carried out on the face image based on the illumination condition, the processed face image is uploaded to a data center after zooming and cutting, then feature extraction is carried out, and a feature set is constructed; performing primary processing based on the feature set, and then creating redundant backup and establishing an index in a data center through distributed storage; and the personalized processing unit compares the current driver feature set with the feature set stored in the data center, and judges whether the face image of the driver is matched so as to determine the control authority of the vehicle.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and more particularly to a distributed storage method for facial image data. Background Art

[0002] In today's automotive industry, intelligent driving technology is rapidly developing, aiming to provide a safer and more convenient driving experience. Facial recognition technology, as an important biometric identification method, has been widely used for driver authentication and personalized settings adjustments. However, with the continuous increase in data volume, how to efficiently and securely store and manage this facial data has become a pressing issue. Distributed storage technology has attracted attention due to its high reliability, high scalability, and high performance. By distributing data across multiple independent devices, distributed storage not only improves data disaster recovery capabilities but also dynamically expands storage capacity according to demand. This is of great significance for intelligent driving systems that need to process large amounts of driver facial data. In intelligent driving scenarios, driver facial image data can be collected in real time by in-vehicle cameras and processed and analyzed using advanced facial recognition algorithms. This data is not only used for authentication but also for configuring vehicle control permissions, such as automatically adjusting seat position and rearview mirror angle to enhance the driving experience. Simultaneously, by recognizing the driver's emotional state, the system can also make corresponding environmental adjustments, such as playing music or activating seat massage functions, further increasing driving comfort and safety. To protect drivers' privacy and data security, it is crucial to employ distributed storage technology for encrypted storage and management of facial image data. This not only prevents data leaks and unauthorized access but also ensures compliance with relevant laws and regulations. Furthermore, the high availability and scalability of distributed storage systems guarantee fast data processing even with large datasets, which is essential for intelligent driving systems that require real-time feedback.

[0003] The invention patent with announcement number CN114063909B discloses an intelligent distributed storage method and system for image data, including an image storage module, an image restoration module, and an image reading and parsing module; combining the spatial redundancy characteristics of the image data itself, on the basis of the existing redundancy mechanism, combined with intelligent algorithms to provide data security protection.

[0004] Although existing technologies have improved the spatial availability of image data in distributed storage and the ability to repair data after image storage data blocks are lost, they still fail to capture and process driver facial images in real time and achieve accurate identity verification and status monitoring. Therefore, to overcome these limitations, the present invention proposes a distributed storage method for facial image data. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention aims to provide a distributed storage method for facial image data, solving the problems of real-time capture and distributed processing of driver facial images to achieve accurate identity verification and status monitoring. The method involves capturing driver facial images in real time via a front-end device, and then using edge computing nodes to perform dynamic normalization processes such as grayscale conversion, scaling, and cropping before uploading to a data center. The data center extracts local and global features from the normalized images to construct a feature set. A distributed storage architecture is employed, configuring a replication factor based on feature importance for redundant backups, and using load balancing technology to distribute data access requests. Based on network connectivity status, remote or local feature matching is selected for identity verification, providing vehicle control permissions. Building upon identity verification, the driver's status is continuously monitored to ensure driving safety.

[0006] To achieve the above object, the present invention provides the following technical solutions: A distributed storage method for facial image data includes the following steps: Step S1: Capture the driver's face image in real time through the front-end acquisition device; Step S2: Using the edge computing node to determine the lighting conditions of the captured facial image, dynamically normalize the image based on the lighting conditions, adaptively grayscale the facial image based on the lighting conditions, and upload the processed facial image to the data center after scaling and cropping. Step S3: The data center performs feature extraction on the face image data after dynamic normalization, wherein the feature extraction includes local feature extraction and global feature extraction, and constructs a feature set; Step S4: Based on the feature set, perform preliminary processing, including feature set splitting, encoding, compression, encryption, and then create redundant backups and establish indexes through distributed storage in the data center; Step S5: The personalized processing unit compares the current driver feature set with the feature set stored in the data center to determine whether the driver's facial image matches to determine the control authority of the vehicle.

[0007] Specifically, the specific steps of dynamic normalization processing include: To determine lighting conditions, acquire a face image and calculate statistical measures of brightness distribution, contrast, and color saturation. The calculation formulas are as follows: ,in, It is a comprehensive evaluation value of lighting conditions. is the mean brightness of the face image, is the standard deviation of the brightness of the face image, is the color saturation of the face image, It refers to the contrast of facial images. is a positive constant; Configure illumination threshold and ,and Less than ,like , then it is determined that the face image is in a low-light environment. If , then the face image is judged to be in a high-light environment, otherwise it is in a normal-light environment.

[0008] Specifically, the specific steps of local feature extraction include: Key points in the image are identified. For each detected key point, a window is configured, the pixels within the window are divided into sub-windows, and the gradient histogram of each sub-window is calculated. The histograms of all sub-windows are connected to form a descriptor vector. All descriptor vectors are stacked into a matrix, singular value decomposition is performed, and the eigenvectors corresponding to the largest k eigenvalues ​​are selected for data projection to achieve dimensionality reduction; The distance between the descriptor vector to be matched and each descriptor vector in the face database is calculated, and the local facial feature label of the descriptor vector with the smallest distance is used as the label of the descriptor vector to be matched.

[0009] Specifically, the specific steps of global feature extraction include: Global texture extraction is performed on face images. Neighborhood windows of different sizes are applied to face images for multi-scale extraction. The LBP feature value of each window is calculated, and the LBP feature values ​​of all scales are merged to form a multi-scale LBP feature vector. Global edge shape feature extraction is performed on the face image, dividing the face image into multiple layers of regions. HOG features are extracted independently in each layer of regions, and the features from all layers are merged into a long vector.

[0010] Specifically, the steps for global texture extraction include: Configure the neighborhood window size and apply the neighborhood window to each pixel with it as the center; Calculate the LBP feature value for each window, i.e.: ,in, is the coordinate of the center pixel, It is The coordinates of the surrounding pixels Representing coordinates The pixel value at Representing coordinates The pixel value at It is the number of pixels in the neighborhood. is a step function defined as:

[0011] at this time, Pick When, if Greater than ,but The value is 1, if Less than or equal to ,but The value is 0; Configuring the Offset Angle Threshold , for each pixel point, calculate its LBP value within the offset angle threshold range, and take its minimum value as the final LBP feature value of the pixel point, that is: ,in, It is the rotation angle, and its value range is... , Is the rotation angle LBP eigenvalues ​​under; Merge the LBP eigenvalues ​​of all neighborhood window configurations and merge the LBP eigenvalues ​​obtained at each scale into a long vector.

[0012] Specifically, the specific steps of global edge shape feature extraction include: Divide the face image into multiple hierarchical regions to form a multi-level grid structure; For each level of the region, the gradient magnitude and gradient direction of each pixel in the region are calculated. For any pixel in the region, , the calculation formula of gradient amplitude and gradient direction is:

[0013]

[0014] in, It is a pixel. gradient magnitude, It's a pixel gradient direction, and It's a pixel horizontal and vertical gradients; The gradient direction is divided into interval directions. For each pixel, the gradient magnitude is accumulated into the histogram of the corresponding interval according to its gradient direction, and the gradient direction histogram of each region is dynamically normalized. Combine the gradient direction histograms in the regions of all levels into one long vector.

[0015] Specifically, the steps of encrypting the feature set include: Based on the encoded and compressed data segments, the original face image, local features, and global features are encrypted at different levels, resulting in multi-level encryption. Specifically, this includes: The first level of encryption uses an encryption algorithm to encrypt the encoded and compressed original face image data fragments, generates a key and encodes it, and saves it as a key feature vector to the local feature vector. The second level of encryption uses an encryption algorithm to perform secondary encryption on the local feature data fragments, generates a new key and encodes it as a key feature vector, and saves it to the global feature vector; The third level of encryption uses an encryption algorithm to encrypt the global feature data fragment three times, generating the final key and encoding it as a key feature vector, which is then saved to a secure storage area.

[0016] Specifically, the steps involved in distributed storage include: The data center performs deep processing on the received facial feature set, wherein the deep processing includes feature set splitting, feature set encoding, feature set compression, and feature set encryption; Based on the encryption level, a replication factor is configured for the original face image, local features, and global features, and redundant backups are performed based on the replication factor. Configure a distributed storage architecture for the data center, and perform distributed storage of the original face images, local features, and global features after redundant backups; By using load balancing technology, data access requests are distributed across multiple data nodes in a distributed storage architecture to avoid overload issues on a single node. Local storage devices are installed inside the vehicle to cache normalized face images obtained by edge computing nodes and global features extracted from the data center; Configuring Update Thresholds , Used to measure the time interval that needs to be updated. Used to measure the memory capacity that needs to be updated. When the interval between local storage device cache updates is greater than When the local storage device's memory capacity is less than At that time, the normalized face images cached locally are adaptively synchronized to the data center.

[0017] Specifically, the steps for identity verification include: The system checks whether the vehicle is connected to the network. If the network is connected, it sends a request to the remote data center to obtain the real-time updated facial image feature set. It then performs complex feature matching between the captured driver facial image features and the feature set provided by the data center. Based on the matching result, it determines whether the current driver already exists in the vehicle's data center. If the network is not connected, it will enter local verification mode, read the pre-cached normalized facial image and global features from the vehicle's local storage device, use local data for simple feature matching to determine the driver's identity, and determine whether the current driver already exists in the vehicle's local storage device based on the local matching results.

[0018] Specifically, simple feature matching includes: Obtain the normalized facial image of the current driver through the edge computing node, perform similarity measurement on it with the normalized facial image in the local storage device, configure a similarity threshold, and output the identifier of the facial image whose similarity measurement is greater than the similarity threshold; When the number of output face image identifiers is 0: the current driver's access request is rejected and the vehicle control authority is not provided; When the number of facial image identifiers output is 1: the recognition result is displayed through a visual interface, and the current driver is requested to confirm. When the number of output facial image identifiers is greater than 1: for each output facial image identifier, obtain its corresponding global features, call the edge computing node to obtain the global features of the current driver's facial image, perform similarity measurement, output the facial image identifier with the highest similarity, and request the current driver to confirm.

[0019] Beneficial effects of the present invention: 1. Dynamically normalizing captured face images via edge computing nodes, including grayscale conversion, scaling, and cropping, effectively reduces data transmission volume and improves transmission efficiency. The data center then extracts features from the dynamically normalized face image data, including local and global feature extraction, to construct a feature set. This helps improve the accuracy and efficiency of face recognition. The feature set is then split, encoded, compressed, and encrypted. Furthermore, distributed storage within the data center creates redundant backups and establishes an index, enhancing data reliability and security.

[0020] 2. The personalization processing unit compares the current driver's profile with the profile stored in the data center to determine if the driver's facial image matches and grants vehicle control rights, thereby enhancing the user experience. During the authentication process, whether in online or offline mode, the driver is effectively identified and the corresponding vehicle control rights are granted, ensuring continuous and personalized service.

[0021] 3. The application of distributed storage architecture and load balancing technology can avoid the overload problem of a single node, improving the system's response speed and stability. Local storage devices are used to cache normalized face images obtained by edge computing nodes and global features extracted from the data center, reducing reliance on the central server and lowering network latency. An adaptive synchronization mechanism ensures that the normalized face images cached locally are updated to the data center in a timely manner, maintaining data consistency. Attached Figure Description

[0022] Figure 1 is a flow chart of a distributed storage method for facial image data; Figure 2 This is the flow chart of adaptive grayscale; Figure 3 Flowchart for feature extraction; Figure 4 Flowchart of the personalized processing unit. DETAILED DESCRIPTION

[0023] See also Figure 1 This embodiment introduces a distributed storage method for face image data, including: Step S1: Capture the driver's face image in real time through the front-end acquisition device; In this embodiment, a high-resolution camera is installed inside the vehicle to ensure clear capture of the driver's facial details. Leveraging the camera's real-time video streaming capability, the driver's facial image is continuously monitored and captured, providing a high-quality data source for subsequent feature extraction and identity verification.

[0024] Step S2: Use the edge computing node to dynamically normalize the captured facial image and upload it to the data center, including grayscale, scaling, and cropping to eliminate the influence of factors such as lighting and posture; In this embodiment, the edge computing node receives image data from the camera and performs preprocessing. First, the color image is converted to grayscale to reduce computational complexity. Then, the image size is adjusted using scaling and cropping techniques to ensure that all images have a uniform size and aspect ratio. This ensures that the subsequent feature extraction process is not affected by external variables, improving the accuracy and reliability of recognition, and enabling stable operation in various environments.

[0025] Preferably, the specific steps of dynamic normalization processing include: See also Figure 2 , obtain a face image, and judge the lighting conditions by calculating the lighting statistics of the brightness distribution, contrast and color saturation of the face image. The calculation formula of the lighting statistics is as follows,

[0026] in, is the comprehensive evaluation value of lighting conditions, is the mean brightness of the face image, and the calculation formula is:

[0027] is the standard deviation of the brightness of the face image, and the calculation formula is:

[0028] is the color saturation of the face image, and the calculation formula is:

[0029]

[0030] It refers to the contrast of a face image, calculated using the following formula:

[0031] in, Is a face image The length and width, The color channel of the face image is , and the coordinates are Pixel value, color channel of face image The value range is R, G, B; It is a positive constant used to prevent division by zero. The ratio of the brightness mean to the standard deviation reflects the size of the average brightness of the face image relative to its variation range. If the ratio is high, it means that the face image is bright and uniform overall; if the ratio is low, it means that the face image is dark or there are large brightness fluctuations. The ratio of color saturation to contrast reflects the size of the image's color saturation relative to the contrast. If the ratio is high, it means that the image color is brighter and the contrast is obvious; if the ratio is low, it means that the image color is relatively simple or the contrast is insufficient. Nonlinear function Used to adjust the effect of brightness standard deviation on comprehensive evaluation. When the brightness standard deviation is large, the function value is close to 1, emphasizing the effect of brightness changes on lighting conditions; when the brightness standard deviation is small, the function value is close to 0, weakening the effect of brightness changes. By adding the above two ratios and multiplying them by the nonlinear adjustment factor, a comprehensive evaluation value is obtained. It takes into account multiple factors such as brightness, color saturation, and contrast, and can more comprehensively reflect lighting conditions.

[0032] Configure illumination threshold and ,and Less than ,like , then it is determined that the face image is in a low-light environment. If If the light intensity is low, the face image is determined to be in a bright environment; otherwise, it is in a normal lighting environment. In a low-light environment, the overall brightness of the face image is low, resulting in unclear facial features and increased recognition difficulty. In a bright environment, the face image may be overexposed, leading to loss of facial details.

[0033] Adaptive grayscale conversion is performed based on lighting conditions. A baseline weighting coefficient for grayscale conversion of facial images under normal lighting conditions is configured, and a weighted average is calculated using this baseline weighting coefficient to obtain the grayscale converted facial image.

[0034] in, is the grayscale face image, These are the pixel values ​​in the R channel of a face image. These are the pixel values ​​in the G channel of a face image. These are the pixel values ​​in the B channel of a face image. , , These are the baseline weight coefficients corresponding to the pixel values ​​in the R, G, and B channels, respectively. These weight coefficients are adaptively modified based on lighting conditions to achieve adaptive grayscale. The specific formula for adaptively modifying the weight coefficients is as follows:

[0035]

[0036]

[0037] in, , , These are the baseline weighting coefficients corresponding to the pixel values ​​in the modified R, G, and B channels. is the comprehensive evaluation value of the average lighting conditions under normal lighting conditions, , , They are the corresponding modification parameters under R, G, and B channels, and the value range is [0,1]. Is a mapping function used to transform The value of is mapped to [0,1], and the grayscale coefficient of the face image is adaptively modified according to the lighting conditions of the face image, changing the influence of different channels on the grayscale image, so that the grayscale image can maintain the best visual effect and information integrity under various lighting conditions. Specifically, this process achieves dynamic adaptation to different lighting environments by adjusting the baseline weight coefficients of the R, G, and B channels, thereby optimizing the quality of the grayscale image.

[0038] By using a face detection algorithm, including the Haar feature classifier in OpenCV, the face location in the grayscale image is detected. The detected face boundary coordinates are calculated to determine the minimum rectangular area containing the entire face. Based on the determined rectangular area, the original image is cropped to retain the face image within the rectangular area. The face image within the rectangular area is then scaled according to the target size. This removes background information from the driver's face image captured in real time by the front-end acquisition device, making subsequent feature extraction and recognition more accurate. The cropped face image is then scaled according to the target size to fit the required input size. The processed image will have a uniform size and content, facilitating further analysis and comparison.

[0039] Step S3: The data center performs feature extraction on the face image data after dynamic normalization, wherein the feature extraction includes local feature extraction and global feature extraction, and constructs a feature set; In this embodiment, please refer to Figure 3 Feature extraction is performed on the dynamically normalized image. Local feature extraction algorithms are used to identify feature points in key areas such as the eyes, nose, and mouth. Simultaneously, global feature extraction algorithms are used to analyze the texture and contour information of the entire face. These feature points and global features are combined to construct a comprehensive facial feature set, providing a solid foundation for identity verification.

[0040] Preferably, the specific steps for local feature extraction include: The Harris corner detector is used to identify key points in face images. For each detected key point, a window is configured, the pixels in the window are divided into sub-windows, and the gradient histogram of each sub-window is calculated. The histograms of all sub-windows are connected to form a descriptor vector. The gradient magnitude and direction of a face image are calculated. The Harris corner detection algorithm is applied, and potential keypoints are identified by calculating interest values. The interest values ​​are determined by the eigenvalues ​​of the gradient covariance matrix. The formula for calculating interest values ​​is as follows:

[0041] in, It represents the interest value, and M is the gradient covariance matrix. It is the determinant of the gradient covariance matrix. As a regulating factor, It is the trace of the gradient covariance matrix.

[0042] Through non-maximum suppression, the keypoint with the maximum response is retained in the adjacent region of each keypoint. Keypoints correspond to the corner points of facial features such as the eyes, nose, and mouth. Identifying these keypoints provides the basis for subsequent feature extraction. Within a certain range around each keypoint, namely a 3x3 or 5x5 window, the keypoint with the maximum interest value is obtained as the final keypoint. A fixed-size window, such as 8x8, is selected with the keypoint as the center. This window is divided into multiple subwindows. For example, an 8x8 window can be divided into four subwindows. For each pixel in each subwindow, the gradient histogram in eight directions is calculated. The gradient histograms of all subwindows are concatenated to form a long vector that serves as the descriptor vector for the keypoint. The descriptor vector contains local feature information around the keypoint, including edge direction and texture. In this way, the keypoint is associated with the local features surrounding it, providing a basis for subsequent feature matching.

[0043] Stack all descriptor vectors into a matrix, perform singular value decomposition, select the eigenvectors corresponding to the k largest eigenvalues ​​for data projection to achieve dimensionality reduction and obtain local descriptor eigenvectors. Create an empty matrix with the number of columns equal to the dimensions of a single descriptor vector and the number of rows equal to the number of keypoints. Iterate over the descriptor vector for each keypoint and add it as a row to the matrix. After iterating over all keypoints, stack the descriptor vectors for all keypoints into a single matrix. Perform singular value decomposition on the matrix, retaining the top k largest eigenvalues ​​and their corresponding eigenvectors. Use these eigenvectors to project the original data, achieving dimensionality reduction. This reduces the data dimension and computational complexity while preserving the data's key features.

[0044] Calculate the distance between the feature vector of the local descriptor to be matched and each descriptor vector in the face database, and take the face local feature label of the descriptor vector with the smallest distance as the label of the feature vector of the local descriptor to be matched; Calculate the Euclidean distance or other similarity measure between the local descriptor feature vector to be matched and each descriptor vector in the face database. Find the descriptor vector pair with the smallest distance and regard it as the best match. Considering the descriptor vector pair with the smallest distance as the best match means that the two descriptor vectors are closest in the feature space and have the highest similarity. By comparing the similarity between the local descriptor feature vectors in the current image and the local descriptor feature vectors stored in the face database, find the most similar pair of descriptor vectors, obtain the local facial feature label corresponding to the descriptor vector, and thus achieve label matching; Preferably, the specific steps for global feature extraction include: Perform global texture extraction on the face image, apply neighborhood windows of different sizes to the face image, perform multi-scale extraction, calculate the LBP eigenvalue of each window, merge the LBP eigenvalues ​​of all scales, and form a multi-scale LBP feature vector; The global edge shape feature extraction is performed on the face image, the face image is divided into multiple levels of regions, the HOG feature values ​​are extracted independently in each level of the region, and the feature values ​​of all levels are merged into a multi-level HOG feature vector.

[0045] Preferably, the steps for global texture extraction of a face image include: Configure the neighborhood window size, and apply a neighborhood window centered on each pixel; Calculate the LBP feature value of each window, that is:

[0046] in, is the coordinate of the center pixel, It is The coordinates of the surrounding pixels, Representing coordinates The pixel value at Representing coordinates The pixel value at It is the number of pixels in the neighborhood, excluding the center point. is a step function defined as:

[0047] at this time Pick When, if Greater than ,but The value is 1, if Less than or equal to ,but The value is 0; Configuring the Offset Angle Threshold For each pixel, calculate its LBP value within the offset angle threshold range, and take the minimum value as the final LBP feature value for that pixel, i.e.:

[0048] in, is the rotation angle, and its value range is , exemplary, The value is 90°. Is the rotation angle LBP eigenvalue under ; The LBP eigenvalues ​​of all neighborhood window configurations are merged, and the LBP eigenvalues ​​obtained at each scale are combined into a long vector. This results in an LBP feature vector that contains information from multiple scales and is rotationally invariant. This feature vector captures the texture information of the image at different scales, making the features more discriminative. By minimizing the LBP value of each pixel at different rotation angles, the features are made robust to rotational changes in the image. The combination of multi-scale and rotational invariance allows this feature vector to more comprehensively describe the image's texture information, making it suitable for a variety of applications requiring high robustness and accuracy.

[0049] Preferably, the specific steps of extracting global edge shape features include: The facial image is divided into multiple hierarchical regions, which gradually decrease in size, forming a multi-level grid structure. Starting with a larger grid, such as 8x8 pixels, the grid size is gradually reduced to 4x4 pixels and then 2x2 pixels. This multi-level division method captures information at different scales in the image. This multi-level region division method makes the features more robust to subtle changes in the image, maintaining good performance in various scenarios. For each level of the region, the gradient magnitude and gradient direction of each pixel in the region are calculated. For any pixel in the region, , the calculation formula of gradient amplitude and gradient direction is:

[0050]

[0051] in, It's a pixel gradient magnitude, It's a pixel gradient direction, and It's a pixel horizontal and vertical gradients; The gradient direction is divided into intervals, each interval representing a specific angle range. For each pixel, the gradient magnitude is accumulated into the histogram of the corresponding interval according to its gradient direction. The gradient direction histogram of each region is dynamically normalized to reduce the impact of changes in lighting and contrast. The gradient direction histograms from all regions are merged into a single long vector, forming the HOG feature vector. Through multi-level region partitioning, accurate gradient calculation, effective interval direction partitioning, dynamic normalization, and feature merging, the global edge shape feature extraction method can effectively capture edge and local shape information in images.

[0052] Preferably, the specific steps of constructing the feature set include: Construct a temporary feature set database, whose fields include: face image identifier, original face image, normalized face image, face image file path, face image acquisition time, local descriptor feature vector, local descriptor feature vector label, global multi-scale LBP feature vector, global multi-level HOG feature vector, which is used to store various types of information related to face features for subsequent processing and analysis.

[0053] The feature vectors, including local descriptor feature vectors, global multi-scale LBP feature vectors, and multi-level HOG feature vectors, along with their feature information including vector labels and file paths, are updated in the feature set database. All extracted feature data is systematically stored to facilitate subsequent authentication or recognition tasks.

[0054] After integrating all feature data into the database, data cleaning and optimization are performed to remove redundant data and standardize feature vectors to improve the efficiency and accuracy of data processing.

[0055] Optimize queries by creating indexes on key fields, including face image identifiers and local descriptor feature vector labels, to speed up the matching process during identity verification and optimize database query performance.

[0056] Step S4: Based on the feature set, perform preliminary processing, including feature set splitting, encoding, compression, encryption, and then create redundant backups and establish indexes through distributed storage in the data center; In this embodiment, the extracted feature set temporary database is split, encoded, and compressed to reduce data transmission and storage space requirements. The compressed data is then encrypted to ensure data privacy and security. The encrypted feature sets are stored in multiple data centers via a distributed storage system, creating redundant backups to prevent data loss. Furthermore, an index is created for each data segment to improve data retrieval efficiency.

[0057] Preferably, the specific steps of encoding and compressing the feature set include: Based on the feature set temporary database, the original face image, the normalized face image, the local descriptor feature vector, the global multi-scale LBP feature vector, and the global multi-level HOG feature vector are each split into independent data segments for separate processing and storage.

[0058] Each data segment is encoded and converted into binary format. The pickle module in Python is used to serialize the data segment and convert it into binary format. Compress the encoded data segments using the Gzip compression algorithm to ensure that the data takes up less space during transmission and storage. Furthermore, a distributed computing framework is used to parallelize the encoding and compression tasks of multiple data segments to further increase processing speed. Preferably, the specific steps of encrypting the feature set include: Based on the encoded and compressed data fragments, the original face image, local descriptor feature vector, global multi-scale LBP feature vector, and global multi-level HOG feature vector are encrypted at multiple levels, including: The first level of encryption uses a strong encryption algorithm, including AES-256-bit encryption, to encrypt the encoded and compressed raw facial image data fragment. A key is generated from the first-level encryption process, encoded, and converted into a hexadecimal string. The encoded key is used as a key feature vector and stored in the local descriptor feature vector. By encoding and compressing the raw facial image data for the first level of encryption, even if the data is intercepted during transmission, unauthorized third parties cannot directly decipher the image content, thereby protecting user privacy. The generated key feature vector, as part of the local feature, enhances its security because even if the local feature data is leaked, the original image cannot be decrypted without the correct key.

[0059] The second level of encryption uses an encryption algorithm to perform secondary encryption on the extracted local descriptor feature vector data fragments, including local area features such as eyes, nose, and mouth. A new key is generated from the secondary encryption process and encoded. The encoded new key is used as a key feature vector and saved in the global feature vector, including the global multi-scale LBP feature vector and the global multi-level HOG feature vector. The secondary encryption of local features provides an additional layer of protection for data security. This means that even if the local feature data fragments are leaked, the attacker will still not be able to obtain useful information because they have been encrypted. Each level of encryption generates a new key and encodes it. These keys are independent of each other, which increases the difficulty of cracking because the attacker needs to obtain all keys at the same time to fully decrypt the data.

[0060] The third level of encryption uses the same encryption algorithm to perform three levels of encryption on the extracted global feature data fragments. The final key is generated from the three-level encryption process and encoded. The encoded final key is stored as a key feature vector in a secure storage area to ensure its security and integrity. The three-level encryption of global features ensures that even if the first two levels of encryption are breached, the final key remains securely stored in the system's secure storage, providing a final line of defense for the entire system. This layered encryption strategy significantly improves the overall security of the system, as each level of encryption increases the obstacles that potential attackers must overcome.

[0061] Preferably, the specific steps of distributed storage include: The data center performs in-depth processing on the received facial feature set, including feature set splitting, encoding, compression, and encryption. Based on the encryption level, replication factors are configured for the original facial image, local features, and global features, and redundant backups are performed based on the replication factors. The replication factor determines the number of redundant copies of each data segment in distributed storage. For global feature data, a higher replication factor is configured to improve data reliability and fault tolerance. Configure a distributed storage architecture for the data center, and store the original face images, local features and global features after redundant backup in a distributed manner. Use load balancing technology to distribute data access requests to multiple data nodes in the distributed storage architecture to avoid overload problems of a single node.

[0062] Among them, a local storage device is set up inside the vehicle to cache the normalized face images obtained by the edge computing nodes and the global features extracted by the data center, so as to support basic face recognition functions when there is no network connection. At the same time, configure the update threshold , Used to measure the time interval required for update, Used to measure the amount of memory that needs to be updated, when the interval between local storage device cache updates is greater than... When or the memory capacity of the local storage device is less than At that time, the normalized face images cached locally are adaptively synchronized to the data center.

[0063] Step S5: The personalization processing unit compares the current driver feature set with the feature set stored in the data center to determine whether the driver's face image matches and thus determine the vehicle's control authority.

[0064] In this embodiment, please refer to Figure 4The personalization processing unit receives the current driver's facial image and verifies the driver's identity using a feature set. If identity verification is successful, the system determines vehicle control permissions, such as automatically adjusting the seat position and playing the driver's preferred music. Simultaneously, the system continuously monitors the driver's condition, including fatigue level and inattention. By analyzing changes in facial features, such as the frequency and duration of eye closure, the system can detect driver fatigue or distraction and issue timely alerts to remind the driver to rest or refocus, ensuring driving safety. This achieves accurate driver identity verification, provision of vehicle control permissions, and real-time monitoring of driving safety.

[0065] Preferably, the specific steps for identity verification include: The system checks whether the vehicle is connected to the network. If the network is connected, it sends a request to the remote data center to obtain the real-time updated facial image feature set. It then performs complex feature matching between the captured driver facial image features and the feature set provided by the data center. Based on the matching result, it determines whether the current driver already exists in the vehicle's data center.

[0066] The complex feature matching includes: By sending a request to a remote data center, a real-time updated set of facial image features is obtained, including facial images corresponding to different facial image identifiers stored in the data center, as well as their local and global features, etc., which are used to match the captured driver's facial image. Using edge computing nodes, the current driver's facial image is captured, and the face image is normalized before being transmitted to the data center for feature extraction, which is then used to match the feature set provided by the data center. Similarity metrics such as cosine similarity and Euclidean distance are used to match the captured driver facial image features with features provided by the data center. By comparing the similarity between different features, it is possible to more accurately determine whether the current driver already exists in the vehicle's data center. Based on the matching results, the identity of the current driver is determined. If the match is successful, the current driver is considered to exist in the vehicle's data center; otherwise, the current driver is considered not to exist in the data center.

[0067] Based on the driver's identity determination, the corresponding operation is performed. That is, if the current driver already exists in the data center, then control permissions for the vehicle can be granted to him / her; otherwise, his / her access request can be denied or he / she can be prompted to try again.

[0068] If the network is not connected, it will enter local verification mode, read the pre-cached normalized facial image and global features from the vehicle's local storage device, use local data for simple feature matching to determine the driver's identity, and determine whether the current driver already exists in the vehicle's local storage device based on the local matching results.

[0069] The simple feature matching includes: Obtain the normalized facial image of the current driver through the edge computing node, perform similarity measurement on it with the normalized facial image in the local storage device, configure a similarity threshold, and output the identifier of the facial image whose similarity measurement is greater than the similarity threshold; When the number of output facial image identifiers is 0: the current driver's access request is rejected and the vehicle control authority is not provided; the driver is prompted to try again or contact the administrator for processing, and the access attempt is recorded, including the time, location and driver's facial image information.

[0070] When the number of output facial image identifiers is 1: the recognition result is displayed through a visual interface, and the current driver is requested to confirm. When the driver confirms the recognition result, the control authority of the vehicle corresponding to the current facial image identifier is provided, including adjusting the seat position and playing the driver's favorite music; when the driver fails the recognition result, the driver is prompted to try again or contact the administrator for processing, and the access attempt is recorded, including the time, place and driver's facial image information.

[0071] When the number of output facial image identifiers is greater than 1: for each output facial image identifier, its corresponding global features are obtained, and the edge computing node is called to obtain the global features of the current driver's facial image, perform similarity measurement, output the facial image identifier with the highest similarity, and request the current driver to confirm. When the driver confirms the recognition result, the control authority of the vehicle corresponding to the current facial image identifier is provided; when the driver fails the recognition result, the driver is prompted to try again or contact the administrator for processing, and the access attempt is recorded, including the time, place and driver's facial image information.

[0072] Preferably, the specific steps of driver status monitoring include: Based on distributed storage, local features of the current driver's facial image are obtained, including the position and shape of the eyes, nose, and mouth, for driver status analysis and comparison; The driver's eye closure frequency and duration are calculated, and a closure frequency threshold and a closure time threshold are configured. When the eye closure frequency and duration exceed the threshold, the driver is determined to be in a state of fatigue. If the driver is fatigued, monitor the driver's line of sight and gaze point to determine if the driver is distracted. If the driver does not look ahead for a long time but looks at other areas, it is determined that the driver is distracted. Feedback is provided to the driver through a visual panel. When fatigue or inattention is detected, the vehicle can alert the driver through audible alarms, seat vibrations, or flashing interior lights. Simultaneously, warning messages are displayed on the instrument panel or center console screen, informing the driver of the current risk level and recommended actions.

[0073] Working principle and its effect: The distributed storage method for facial image data captures driver facial images in real time using front-end acquisition devices and performs dynamic normalization processing using edge computing nodes, including grayscale conversion, scaling, and cropping. The data center receives this data and extracts features, including local and global features, to construct a feature set. Based on these feature sets, the data center performs splitting, encoding, compression, and encryption processing, and then uses a distributed storage architecture for redundant backup and index creation to ensure data security. The personalization processing unit compares the current driver's feature set with the feature set stored in the data center to achieve identity verification. Once verification is successful, the system grants vehicle control permissions, such as adjusting seat position and playing music, and monitors the driver's status.

[0074] This system improves authentication accuracy, enhances data security, provides vehicle control permissions, optimizes system performance, adapts to different lighting environments, enables real-time updates and synchronization, and features flexible network connection modes. It enhances authentication accuracy by combining local and global features and employing complex similarity measurement methods. Multi-level encryption protects user privacy and data security, while vehicle control permissions improve the driving experience. Load balancing and local caching strategies optimize system performance, and dynamic normalization adapts to different lighting environments. Real-time updates and synchronization mechanisms ensure data timeliness, while flexible network connection modes guarantee system reliability and availability. Multi-level feature extraction comprehensively describes facial images, improving recognition accuracy and robustness. Multi-scale LBP and HOG feature vectors capture global texture information and edge shape features, respectively. The key feature vector generated during encryption ensures encryption security and traceability. Distributed storage architecture improves data reliability and accessibility, while a visual interface enhances user experience and interactivity.

[0075] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the concept of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A distributed storage method for face image data, characterized in that: include Step S1: Capturing the driver's face image in real time through the front-end acquisition device; Step S2: using the edge computing node to determine the lighting conditions of the captured face image, performing dynamic normalization processing according to the lighting conditions, adaptively graying the face image based on the lighting conditions, and uploading the processed face image to the data center after scaling and cropping; Step S3: the data center performs feature extraction on the face image data after dynamic normalization processing, wherein the feature extraction includes local feature extraction and global feature extraction, and constructs a feature set; Step S4: Based on the feature set, preliminary processing is performed, including feature set splitting, encoding, compression, encryption, and then redundant backup is created and indexing is established in the data center through distributed storage; Step S5: The personalized processing unit compares the current driver feature set with the feature set stored in the data center to determine whether the driver's facial image matches to determine the control authority of the vehicle.

2. The distributed storage method for facial image data according to claim 1, characterized in that: The specific steps of the dynamic normalization process include: Get a face image and determine the lighting conditions by calculating the statistics of the brightness distribution, contrast, and color saturation of the face image. The calculation formula is as follows: ,in, is the comprehensive evaluation value of lighting conditions. is the mean brightness of the face image, is the standard deviation of face image brightness, is the color saturation of the face image, is the face image contrast, is a positive constant; Configuring light thresholds and ,and Less than ,like , then the face image is judged to be in a low-light environment. If , then the face image is judged to be in a high-light environment, otherwise it is in a normal-light environment.

3. The distributed storage method for facial image data according to claim 1, characterized in that: The specific steps of extracting local features include: Identify key points in the image. For each detected key point, configure a window, divide the pixels in the window into sub-windows, and calculate the gradient histogram of each sub-window. Connect the histograms of all sub-windows to form a descriptor vector. All descriptor vectors are stacked into a matrix, singular value decomposition is performed, and the eigenvectors corresponding to the largest k eigenvalues ​​are selected for data projection to achieve dimensionality reduction; The distance between the descriptor vector to be matched and each descriptor vector in the face database is calculated, and the local facial feature label of the descriptor vector with the smallest distance is used as the label of the descriptor vector to be matched.

4. The distributed storage method for facial image data according to claim 1, characterized in that: The specific steps of the global feature extraction include: Perform global texture extraction on face images, apply neighborhood windows of different sizes to face images, perform multi-scale extraction, calculate the LBP eigenvalue of each window, merge the LBP eigenvalues ​​of all scales, and form a multi-scale LBP feature vector; Perform global edge shape feature extraction on face images, divide the face images into multiple levels of regions, extract HOG features independently on each level of region, and merge features of all levels into a long vector.

5. The distributed storage method for facial image data according to claim 4, characterized in that: The step of global texture extraction comprises: Configure the neighborhood window size and apply the neighborhood window to each pixel with it as the center; Calculate the LBP feature value of each window, that is: ,in, is the coordinate of the center pixel, It is The coordinates of the surrounding pixels, Representing coordinates The pixel value at Representing coordinates The pixel value at is the number of pixels in the neighborhood, is a step function, defined as: ,at this time, Pick When Greater than ,but The value is 1, if Less than or equal to ,but The value is 0; Configuring the Offset Angle Threshold , for each pixel point, calculate its LBP value within the offset angle threshold range, and take its minimum value as the final LBP feature value of the pixel point, that is: ,in, is the rotation angle, and its value range is , is the rotation angle The LBP eigenvalue under ; Merge the LBP eigenvalues ​​of all neighborhood window configurations and merge the LBP eigenvalues ​​obtained at each scale into a long vector.

6. The distributed storage method for facial image data according to claim 4, characterized in that: The specific steps of extracting the global edge shape feature include: Divide the face image into multiple hierarchical regions to form a multi-level grid structure; For each level of the region, the gradient amplitude and gradient direction of each pixel in the region are calculated. For any pixel in the region , the calculation formula of gradient amplitude and gradient direction is: , ,in, It's a pixel The gradient amplitude of It's a pixel The gradient direction of and It's a pixel Horizontal and vertical gradients; The gradient direction is divided into interval directions. For each pixel point, its gradient amplitude is accumulated into the histogram of the corresponding interval according to its gradient direction, and the gradient direction histogram of each area is dynamically normalized. Combine the gradient direction histograms in the regions of all levels into one long vector.

7. The distributed storage method for facial image data according to claim 1, characterized in that: The specific steps of encrypting the feature set include: Based on the encoded and compressed data fragments, different encryption levels are used for the original face image, local features and global features, and multi-level encryption is performed separately, including: The first level of encryption uses an encryption algorithm to perform a first-level encryption on the encoded and compressed original face image data fragment, generates a key and performs key encoding, which is used as a key feature vector and saved in the local feature vector; The second level of encryption uses an encryption algorithm to perform secondary encryption on the local feature data fragments, generates a new key and encodes the key as a key feature vector, which is saved in the global feature vector; The third level of encryption uses an encryption algorithm to perform three-level encryption on the global feature data fragment, generate the final key and encode the key as a key feature vector, and save it in a secure storage area.

8. The distributed storage method for facial image data according to claim 7, characterized in that: The specific steps of the distributed storage include: The data center performs deep processing on the received facial feature set, wherein the deep processing includes feature set splitting, feature set encoding, feature set compression, and feature set encryption; According to the encryption level, respectively configure the replication factor for the original face image, the local features and the global features, and perform redundant backup according to the replication factor; Configure a distributed storage architecture for the data center and perform distributed storage of the redundantly backed-up original face images, local features, and global features; Through load balancing technology, data access requests are distributed to multiple data nodes in the distributed storage architecture to avoid overloading of a single node; A local storage device is set up inside the vehicle to cache the normalized face images obtained by the edge computing node and the global features extracted by the data center; Configuring Update Thresholds , Used to measure the time interval required for updating. Used to measure the memory capacity that needs to be updated. When the interval between local storage device cache updates is greater than When the local storage device has less memory than When , the normalized face images cached locally are adaptively synchronized to the data center.

9. The distributed storage method for facial image data according to claim 1, characterized in that: The specific steps of determining whether the driver's facial image matches include: Detect whether the current vehicle network is connected. If the network is connected, send a request to the remote data center to obtain the real-time updated facial image feature set, perform complex feature matching on the captured driver's facial image features and the feature set provided by the data center, and determine whether the current driver already exists in the data center of the vehicle based on the matching results; If the network is not connected, it will enter local verification mode, read the pre-cached normalized facial image and global features from the vehicle's local storage device, use local data for simple feature matching to determine the driver's identity, and determine whether the current driver already exists in the vehicle's local storage device based on the local matching results.

10. The distributed storage method for facial image data according to claim 9, characterized in that: The simple feature matching includes: Obtain the normalized face image of the current driver through the edge computing node, measure the similarity with the normalized face image in the local storage device, configure the similarity threshold, and output the face image identifier whose similarity measurement is greater than the similarity threshold; When the number of output face image identifiers is 0: the access request of the current driver is rejected and the vehicle control authority is not provided; When the number of output face image identifiers is 1: the recognition result is displayed through a visual interface, and the current driver is requested to confirm; When the number of output facial image identifiers is greater than 1: for each output facial image identifier, obtain its corresponding global features, call the edge computing node to obtain the global features of the current driver's facial image, perform similarity measurement, output the facial image identifier with the highest similarity, and request the current driver to confirm.

Images