CN116798076A - Fingerprint sampling data storage system based on GPGPU - Google Patents
Fingerprint sampling data storage system based on GPGPU Download PDFInfo
- Publication number
- CN116798076A CN116798076A CN202310795505.0A CN202310795505A CN116798076A CN 116798076 A CN116798076 A CN 116798076A CN 202310795505 A CN202310795505 A CN 202310795505A CN 116798076 A CN116798076 A CN 116798076A
- Authority
- CN
- China
- Prior art keywords
- data
- fingerprint
- module
- feature
- 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.)
- Pending
Links
- 238000013500 data storage Methods 0.000 title claims abstract description 63
- 238000005070 sampling Methods 0.000 title claims abstract description 28
- 238000000605 extraction Methods 0.000 claims abstract description 82
- 238000000034 method Methods 0.000 claims abstract description 42
- 238000012545 processing Methods 0.000 claims abstract description 36
- 238000007781 pre-processing Methods 0.000 claims abstract description 27
- 238000004364 calculation method Methods 0.000 claims abstract description 8
- 230000008569 process Effects 0.000 claims abstract description 6
- 238000005516 engineering process Methods 0.000 claims description 18
- 230000007246 mechanism Effects 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 12
- 238000013527 convolutional neural network Methods 0.000 claims description 11
- 230000006835 compression Effects 0.000 claims description 9
- 238000007906 compression Methods 0.000 claims description 9
- 238000013136 deep learning model Methods 0.000 claims description 9
- 238000013461 design Methods 0.000 claims description 9
- 238000011084 recovery Methods 0.000 claims description 9
- 238000013526 transfer learning Methods 0.000 claims description 9
- 230000005540 biological transmission Effects 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 7
- 238000007726 management method Methods 0.000 claims description 7
- 230000003044 adaptive effect Effects 0.000 claims description 6
- 230000002708 enhancing effect Effects 0.000 claims description 6
- 239000000284 extract Substances 0.000 claims description 6
- 238000013523 data management Methods 0.000 claims description 4
- 230000003993 interaction Effects 0.000 claims description 4
- 230000003042 antagnostic effect Effects 0.000 claims description 3
- 239000003990 capacitor Substances 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 3
- 238000004891 communication Methods 0.000 claims description 3
- 238000012217 deletion Methods 0.000 claims description 3
- 230000037430 deletion Effects 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 238000005286 illumination Methods 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 238000013508 migration Methods 0.000 claims description 3
- 230000005012 migration Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 230000000737 periodic effect Effects 0.000 claims description 3
- 230000002688 persistence Effects 0.000 claims description 3
- 238000013434 data augmentation Methods 0.000 claims 1
- 238000005265 energy consumption Methods 0.000 abstract description 4
- 238000001914 filtration Methods 0.000 description 7
- 238000009499 grossing Methods 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 230000003321 amplification Effects 0.000 description 2
- 230000008485 antagonism Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000003199 nucleic acid amplification method Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 235000002566 Capsicum Nutrition 0.000 description 1
- 239000006002 Pepper Substances 0.000 description 1
- 235000016761 Piper aduncum Nutrition 0.000 description 1
- 235000017804 Piper guineense Nutrition 0.000 description 1
- 244000203593 Piper nigrum Species 0.000 description 1
- 235000008184 Piper nigrum Nutrition 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 150000003839 salts Chemical class 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1347—Preprocessing; Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/901—Indexing; Data structures therefor; Storage structures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/778—Active pattern-learning, e.g. online learning of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1365—Matching; Classification
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Medical Informatics (AREA)
- Human Computer Interaction (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Bioethics (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention discloses a fingerprint sampling data storage system based on a GPGPU, which comprises a data acquisition module, an image preprocessing module, a feature extraction module, a feature matching module, a data storage module and a user interface module; the invention can remarkably accelerate the processing speed of fingerprint images and the characteristic extraction process by utilizing the parallel computing capability of the GPU. Traditional CPU calculation cannot be comparable with the parallel processing capacity of the GPU, so that the performance of the GPGPU-based system can be greatly improved; the invention can effectively process large-scale data, fully utilize the parallel computing capability of the GPU and improve the throughput and efficiency of data processing; the invention can provide rapid fingerprint identification and data processing in a real-time application scene; the invention has good expandability, and the quantity of the GPUs and the computing resources can be increased according to the requirements; while GPGPU systems provide high performance, they may also have certain advantages in terms of energy consumption.
Description
Technical Field
The invention relates to the field of fingerprint sampling data storage systems, in particular to a fingerprint sampling data storage system based on a GPGPU.
Background
The advent of fingerprint sampling data storage systems has resulted from the widespread use and development of fingerprint identification technology. Fingerprint as a unique biological feature has wide application and high reliability in the fields of personal identification and authentication. With the progress of technology and the popularization of hardware equipment, the performances of fingerprint acquisition equipment and fingerprint identification algorithms are continuously improved, and the demands and development of fingerprint sampling data storage systems are promoted. The fingerprint sampling data storage system based on the GPGPU can improve fingerprint identification performance, efficiently process large-scale data, support real-time application, have expandability and flexibility, and provide data security and privacy protection.
However, the conventional fingerprint sampling data storage system based on the GPGPU still has many defects: fingerprint data processing and computation with a GPGPU typically requires writing and optimizing the underlying GPU core code, which may have a certain learning curve and technical threshold for non-professional developers. In addition, GPGPU programming also needs to consider the transmission and synchronization problems of data, and complexity of system design and implementation is increased; the performance of GPGPU systems is often limited by the transmission bandwidth and latency of the data. In fingerprint sampling data storage systems, data transfer from the fingerprint acquisition device to the GPU may become a performance bottleneck, especially when processing large-scale data; GPUs typically have limited memory capacity, which may limit the size of fingerprint data that can be processed simultaneously. For large-scale fingerprint data storage systems, additional mechanisms may be required to manage and expand storage capacity; the acquisition and query requests of fingerprint data cannot be processed and responded to in real time. Particularly in the case of large-scale data sets or high concurrent access, the performance of the system may not meet the requirements of real-time; GPGPU systems typically require significant power consumption and hardware resources, which can result in high power consumption and cost. For certain application scenarios, in particular mobile devices or embedded systems, energy consumption and cost may become limiting factors.
Disclosure of Invention
The invention aims to provide a fingerprint sampling data storage system based on GPGPU, which solves the problems set forth in the background art:
fingerprint data processing and computation by a GPGPU typically requires writing and optimizing the underlying GPU core code, which may have a certain learning curve and technical threshold for non-professional developers. In addition, GPGPU programming also needs to consider the transmission and synchronization problems of data, and complexity of system design and implementation is increased; the performance of GPGPU systems is often limited by the transmission bandwidth and latency of the data. In fingerprint sampling data storage systems, data transfer from the fingerprint acquisition device to the GPU may become a performance bottleneck, especially when processing large-scale data; GPUs typically have limited memory capacity, which may limit the size of fingerprint data that can be processed simultaneously. For large-scale fingerprint data storage systems, additional mechanisms may be required to manage and expand storage capacity; the acquisition and query requests of fingerprint data cannot be processed and responded to in real time. Particularly in the case of large-scale data sets or high concurrent access, the performance of the system may not meet the requirements of real-time; GPGPU systems typically require significant power consumption and hardware resources, which can result in high power consumption and cost. For certain application scenarios, in particular mobile devices or embedded systems, energy consumption and cost may become limiting factors.
The fingerprint sampling data storage system based on the GPGPU is characterized by comprising a data acquisition module, an image preprocessing module, a feature extraction module, a feature matching module, a data storage module and a user interface module;
the data acquisition module transmits the acquired original fingerprint image data to the image preprocessing module for processing;
the image preprocessing module is used for receiving the original fingerprint image data from the data acquisition module, carrying out denoising, enhancing and normalizing preprocessing steps on the image, generating preprocessed fingerprint image data, and transmitting the preprocessed fingerprint image data to the feature extraction module;
the feature extraction module receives the preprocessed fingerprint image data from the image preprocessing module, extracts fingerprint features including minutiae points and the direction of a ridge from the image by using a fingerprint feature extraction algorithm, generates extracted fingerprint feature data, and transmits the extracted fingerprint feature data to the feature matching module;
the feature matching module receives the fingerprint feature data from the feature extraction module, compares and matches the acquired fingerprint features with stored fingerprint features, generates a matching result, and transmits the result to the user interface module;
the data storage module is used for storing the collected original fingerprint image data, the fingerprint image data generated by preprocessing and the fingerprint feature data, receiving the fingerprint feature data extracted by the feature extraction module and storing the fingerprint feature data into a database and a file system;
the user interface module provides an interface for interaction between a user and the system, and comprises functions of triggering fingerprint acquisition, inquiring stored fingerprint data and displaying matching results, and interacts with the data acquisition module, the feature matching module and the data storage module to transmit instructions of the user and acquire corresponding results.
Preferably, the data acquisition module is responsible for acquiring original fingerprint image data from fingerprint acquisition equipment based on communication with interfaces of the fingerprint scanner and the sensor, and acquiring a high-quality fingerprint image;
the data acquisition module takes the format, sampling rate and sampling quality of the data into account.
Preferably, the image preprocessing module performs preprocessing on the acquired fingerprint image data, and the preprocessing step may include image denoising, enhancement and normalization.
Preferably, the feature extraction module extracts representative features from the preprocessed fingerprint image, a common fingerprint feature extraction algorithm comprises a Minutiae-based method and a Ridge-based method, and the fingerprint extraction algorithm identifies detail features including Minutiae and Ridge directions in the fingerprint image.
The feature extraction module automatically learns features in the fingerprint image using a Convolutional Neural Network (CNN) and other deep learning models;
the deep learning model further comprises a self-encoder, sparse coding, a pre-training model and transfer learning;
convolutional neural network (Convolutional Neural Networks, CNN): CNN is a deep learning model widely used for image processing tasks. The method can automatically learn the characteristics in the image through components such as a convolution layer, a pooling layer, a full connection layer and the like. For fingerprint images, CNN can learn the shape, texture, and local pattern of the ridge line.
Self-encoder (Autoencoders): the self-encoder is an unsupervised learning model for learning a compact representation of the input data. It consists of an encoder and decoder that can compress input data into a low-dimensional representation of features by training and reconstruct the input data therefrom. The self-encoder may be used to learn an abstract feature representation of the fingerprint image.
Sparse Coding (Sparse Coding): sparse coding is an unsupervised learning method for learning sparse representation of data. It represents the input data as a linear combination of basis vectors by learning a set of these basis vectors and sparse coefficients. Sparse coding can be applied to fingerprint images, learn sparse representation of fingerprints, and extract key features.
Pre-training models and transfer learning: the pre-training model is a deep learning model which is trained in advance on a large-scale data set, such as ResNet, VGG, acceptance and the like in an image classification task. These models may act as feature extractors, inputting the fingerprint image into the middle layer of the model, extracting a high-level feature representation of the image. The transfer learning is to apply the weight and structure of the pre-training model to the task of fingerprint image, accelerate model training and improve performance.
The feature extraction module is based on an image enhancement and denoising method, uses an image enhancement and denoising technology to improve the quality of fingerprint images, and applies an image denoising algorithm, enhances contrast and adjusts image brightness to improve the reliability of fingerprint features;
the feature extraction module introduces context information, introduces the context information of surrounding and adjacent fingerprint images except a single fingerprint image by means of multi-mode fusion, multi-scale processing or sequence model and the like, performs joint processing by using a plurality of fingerprint images, and captures richer feature information;
the feature extraction module increases robustness and anti-interference capability, a robustness algorithm is used for processing the problems that a fingerprint image faces various interferences and noises including fingerprint rotation deformation and local shielding in a feature extraction process, and a method including rotation invariance, local ridge tracking and ridge subdivision is introduced in the feature extraction process;
the robustness algorithm is able to identify and process outliers in the data without being seriously affected by them. This means that the algorithm can handle outliers appropriately, avoiding that outliers have an excessive impact on the results of the algorithm.
Robustness to noise: the robustness algorithm can still produce accurate and reliable results in the presence of noise. They can handle noise by reducing its impact or by employing appropriate filtering and smoothing techniques.
Robustness to violations of model assumptions: the robustness algorithm can still provide reasonable results in the case where the data does not meet the model assumptions. They can adapt to the diversity of data and have certain fault tolerance;
the feature extraction module further optimizes the computing performance, reduces the computing complexity by utilizing the parallel computing capability of the GPU and using an approximate computing method, and reduces the computing amount and accelerates the feature extraction process by using an efficient computing method and algorithm.
Preferably, the feature matching module compares the collected fingerprint features with stored fingerprint features to determine whether matching exists;
the feature matching module uses a matching algorithm comprising a feature-based matching algorithm and a model-based matching algorithm;
the feature matching module initializes the feature extraction module by using a pre-trained deep learning model (such as a model on ImageNet) based on feature extraction of transfer learning, and adapts fingerprint image data through fine tuning and training in a specific field;
the transfer learning method utilizes the feature representation capability of a model pre-trained on a large-scale data set and performs effective feature extraction on limited fingerprint data;
the feature matching module performs weak supervision learning, and reduces dependence on labeling data by using a weak supervision learning method comprising semi-supervision and unsupervised learning through self-supervision learning and technology of generating an antagonistic network or clustering;
the feature matching module combines the multi-mode information with the temperature, the pressure and the additional biological feature information of the capacitor of the fingerprint image to perform feature extraction, and provides more auxiliary information to enhance the accuracy and the robustness of feature extraction.
The feature matching module uses non-local features, and introduces a non-local feature extraction method, wherein the extraction method comprises the overall line direction, the layout of lines and a topological structure;
the feature matching module is based on the feature extraction of self-adaptive learning, and aims at the difference between individuals, and the self-adaptive feature is extracted by adopting a self-adaptive learning method through personalized model parameter adjustment, dynamically adjusted feature weights or an adaptive mechanism using individual feature expression;
the robustness of the feature matching module is enhanced, and the robustness of feature extraction is improved by using image enhancement, denoising and restoration technologies, so that various interferences and noises including dirt, illumination change and distortion are faced to the fingerprint image;
the feature matching module also introduces an antagonism training or data amplification technology;
preferably, the data storage module designs an effective data storage system for storing the collected fingerprint data and the corresponding characteristic data;
the data storage module selects a database system comprising a relational database, a NoSQL database and a graph database and a file system comprising a distributed file system, an object storage system and a block storage system to manage and organize data, and ensures the safety and expandability of the data;
the data storage module stores and processes in a distributed manner: consider the use of a distributed storage system and a processing framework such as Apache Hadoop, apache Spark, google Cloud Bigtable, or the like. The system can realize the distributed storage and processing of data, improve the expandability and the performance of the system and support the storage and processing requirements of large-scale data.
The data storage module uses lossless and lossy compression algorithms to reduce occupation of storage space and bandwidth consumption of data transmission;
the data storage module uses lossless and lossy compression algorithms, and introduces a data redundancy and fault tolerance mechanism; using Redundant Array (RAID) and distributed data backup lossless and lossy compression algorithms to prevent data loss and system failure;
the data storage module designs an efficient data index structure and a query optimization algorithm by using the B+ tree, inverted index and hash-based index technology, so that the retrieval and query operation of data are accelerated;
the data storage module introduces a security and privacy protection mechanism comprising data encryption, access control, authority management and anonymization technologies, protects confidentiality and integrity of fingerprint sampling data, and complies with related privacy regulations and standards;
the data storage module performs data backup and recovery: reliable data backup and recovery strategies are designed to cope with the situation of data loss or system failure. Periodic data backup, snapshot techniques, disaster recovery backup, or the like may be considered to ensure data recoverability and persistence.
The data storage module manages data and lifecycle: data management and lifecycle management policies are introduced to manage lifecycle, storage costs, and access frequency of data. The storage location of the data, the data migration policy, and the data deletion mechanism may be determined according to the importance of the data, the access pattern, and the storage requirements.
Preferably, the user interface module is responsible for interaction with a user, including triggering of fingerprint acquisition, data query and result display functions;
the user interface module designs a user-friendly graphical interface or command line interface, which is convenient for a user to operate a system and acquire results
Compared with the prior art, the invention has the advantages that:
(1) By utilizing the parallel computing capability of the GPU, the processing speed of the fingerprint image and the characteristic extraction process can be remarkably accelerated. Traditional CPU calculation cannot be comparable with the parallel processing capacity of the GPU, so that the performance of the GPGPU-based system can be greatly improved.
(2) The system based on the GPGPU can effectively process large-scale data, fully utilizes the parallel computing capacity of the GPU, and improves the throughput and the efficiency of data processing.
(3) The GPGPU-based system can provide rapid fingerprint identification and data processing in a real-time application scene.
(4) GPGPU-based systems generally have good scalability, and the number of GPUs and computing resources can be increased according to requirements to accommodate ever-increasing fingerprint data and ever-changing application requirements.
(5) While GPGPU systems provide high performance, they may also have certain advantages in terms of energy consumption.
Drawings
FIG. 1 is a schematic diagram of the overall system of the present invention.
Detailed Description
Examples: referring to fig. 1, the fingerprint sampling data storage system based on the GPGPU includes a data acquisition module, an image preprocessing module, a feature extraction module, a feature matching module, a data storage module, and a user interface module;
the data acquisition module transmits the acquired original fingerprint image data to the image preprocessing module for processing;
the image preprocessing module is used for receiving the original fingerprint image data from the data acquisition module, carrying out preprocessing steps of denoising, enhancing and normalizing on the image, generating preprocessed fingerprint image data, and transmitting the preprocessed fingerprint image data to the feature extraction module;
the feature extraction module is used for receiving the preprocessed fingerprint image data from the image preprocessing module, extracting fingerprint features including minutiae points and the direction of the ridge from the image by using a fingerprint feature extraction algorithm, generating extracted fingerprint feature data, and transmitting the extracted fingerprint feature data to the feature matching module;
the feature matching module receives the fingerprint feature data from the feature extraction module, compares and matches the acquired fingerprint features with stored fingerprint features, generates a matching result, and transmits the result to the user interface module;
the data storage module is used for storing the acquired original fingerprint image data, the fingerprint image data generated by preprocessing and the fingerprint feature data, receiving the fingerprint feature data extracted by the feature extraction module and storing the fingerprint feature data into a database and a file system;
the user interface module provides an interface for interaction between a user and the system, and comprises functions of triggering fingerprint acquisition, inquiring stored fingerprint data and displaying matching results, and interacts with the data acquisition module, the feature matching module and the data storage module to transmit instructions of the user and acquire corresponding results.
The data acquisition module is responsible for acquiring original fingerprint image data from fingerprint acquisition equipment based on communication with interfaces of the fingerprint scanner and the sensor, and acquiring a high-quality fingerprint image;
the data acquisition module takes the format, sampling rate and sampling quality of the data into account.
The image preprocessing module performs preprocessing on the acquired fingerprint image data, and the preprocessing step can comprise image denoising, enhancement and normalization.
The characteristic extraction module extracts representative characteristics from the preprocessed fingerprint image, a common fingerprint characteristic extraction algorithm comprises a Minutiae-based method and a Ridge-based method, and the fingerprint extraction algorithm identifies detail characteristics comprising Minutiae and a Ridge line direction in the fingerprint image;
specifically, the detailed operation of the Ridge-based method is as follows:
image enhancement: the fingerprint image is preprocessed and enhanced to improve image quality and contrast. This may include removing noise, smoothing images, enhancing ridges, etc.;
estimating the direction: the ridge line direction of each pixel point in the fingerprint image is estimated. This may be achieved by calculating the gradient and direction fields of the image, or applying filtering and direction estimation algorithms;
ridge extraction: based on the ridge line direction information, a ridge line in the fingerprint image is extracted. Common methods include ridge tracking based on directional gradients, local direction estimation, connected component analysis, and the like;
characterization: and generating a feature vector or a feature descriptor for fingerprint identification according to the morphology and the features of the ridge line. This may include extracting features of the start point, end point, length, direction, etc. of the ridge line;
specifically, the detailed operation of the Minutiae-based method is as follows;
histogram equalization (Histogram Equalization): the method enhances the contrast of the image by adjusting the gray level distribution of the image. Histogram equalization can be performed by redistributing the gray level of the image to make the pixel value distribution in the image more uniform, thereby enhancing the details of the image;
adaptive histogram equalization (Adaptive Histogram Equalization): the method divides the image into small blocks and performs histogram equalization within each small block. The adaptive histogram equalization can better handle local contrast variation in the image and prevent excessive enhancement of noise of the image compared to global histogram equalization;
gaussian filtering (Gaussian Filtering): gaussian filtering is a commonly used linear smoothing filter that suppresses high frequency noise in an image. By applying a Gaussian filter, details of the image can be blurred, noise in the image is reduced, and image quality is improved;
median Filtering (Median Filtering): the median filter is a nonlinear filter that replaces the value of each pixel with the median value in the neighborhood of the pixel. The median filter is very effective for removing random noise such as salt and pepper noise, and meanwhile, edge information in an image can be reserved;
converting a color image into a gray image: if a color fingerprint image is processed, it can be converted into a gray scale image to reduce the amount of computation and simplify the subsequent processing steps. Common conversion methods include weighting addition of three channels of an RGB image or using a gray conversion formula perceived by the human eye;
sharpening enhancement (Sharpening): by applying a sharpening filter, the edges and details of the image can be enhanced, making the image clearer. Sharpening filters may be implemented by adding high frequency components or enhancing gradients of the image.
The feature extraction module automatically learns features in the fingerprint image using a Convolutional Neural Network (CNN) and other deep learning models;
the feature extraction module is based on an image enhancement and denoising method, uses an image enhancement and denoising technology to improve the quality of fingerprint images, and applies an image denoising algorithm, enhances contrast and adjusts image brightness to improve the reliability of fingerprint features;
the feature extraction module introduces context information, introduces the context information of surrounding and adjacent fingerprint images except a single fingerprint image by means of multi-mode fusion, multi-scale processing or sequence model and the like, performs joint processing by using a plurality of fingerprint images, and captures richer feature information;
the feature extraction module increases robustness and anti-interference capability, a robustness algorithm is used for processing the problems that a fingerprint image faces various interferences and noises including fingerprint rotation deformation and local shielding in the feature extraction process, and a method including rotation invariance, local ridge tracking and ridge subdivision is introduced in the feature extraction process;
specifically, the robustness algorithm includes median and median absolute deviation (Median and Median Absolute Deviation, MAD), RANSAC (RANdom SAmple Consensus), huber loss function and M-estimation, robust covariance estimation (Robust Covariance Estimation), subspace robustness method (Subspace Robust Methods)
The feature extraction module further optimizes the computing performance, reduces the computing complexity by utilizing the parallel computing capability of the GPU and using an approximate computing method, and reduces the computing amount and accelerates the feature extraction process by using an efficient computing method and algorithm.
The characteristic matching module compares the collected fingerprint characteristics with stored fingerprint characteristics to determine whether matching exists or not;
the feature matching module uses a matching algorithm comprising a feature-based matching algorithm and a model-based matching algorithm;
the feature matching module uses a pre-trained deep learning model (such as a model on ImageNet) to initialize the feature extraction module based on the feature extraction of the transfer learning, and adapts fingerprint image data through fine tuning and training in a specific field;
the transfer learning method utilizes the feature representation capability of a model pre-trained on a large-scale data set and performs effective feature extraction on limited fingerprint data;
the feature matching module performs weak supervision learning, and reduces dependence on labeling data by using a weak supervision learning method comprising semi-supervision and unsupervised learning through self-supervision learning and technology of generating an antagonistic network or clustering;
the feature matching module combines the multi-mode information with the temperature, pressure and additional biological feature information of the capacitor of the fingerprint image to perform feature extraction, and provides more auxiliary information to enhance the accuracy and robustness of feature extraction.
The feature matching module uses non-local features, and introduces a non-local feature extraction method, wherein the extraction method comprises the overall line direction, the layout of the patterns and the topological structure;
the feature matching module is based on the feature extraction of self-adaptive learning, and aims at the difference between individuals, and the self-adaptive feature is extracted by adopting a self-adaptive learning method through personalized model parameter adjustment, dynamically adjusted feature weights or an adaptive mechanism using individual feature expression;
the robustness of the feature matching module is enhanced, and the robustness of feature extraction is improved by using the image enhancement, denoising and restoration technology, so that various interferences and noises including dirt, illumination change and distortion are faced to the fingerprint image;
the feature matching module also introduces an antagonism training or data amplification technology;
the data storage module is used for designing an effective data storage system and storing the collected fingerprint data and the corresponding characteristic data;
the data storage module selects a database system comprising a relational database, a NoSQL database and a graph database and a file system comprising a distributed file system, an object storage system and a block storage system to manage and organize data, and ensures the safety and expandability of the data;
data storage module distributed storage and processing: consider the use of a distributed storage system and a processing framework such as Apache Hadoop, apache Spark, google Cloud Bigtable, or the like. The system can realize the distributed storage and processing of data, improve the expandability and the performance of the system and support the storage and processing requirements of large-scale data.
The data storage module uses lossless and lossy compression algorithms to reduce occupation of storage space and bandwidth consumption of data transmission;
the data storage module uses lossless and lossy compression algorithms, and introduces a data redundancy and fault tolerance mechanism; using Redundant Array (RAID) and distributed data backup lossless and lossy compression algorithms to prevent data loss and system failure;
the data storage module designs an efficient data index structure and a query optimization algorithm by using the B+ tree, inverted index and hash-based index technology, so that the data retrieval and query operation are accelerated;
the data storage module introduces technical safety and privacy protection mechanisms including data encryption, access control, authority management and anonymization, protects confidentiality and integrity of fingerprint sampling data, and complies with related privacy regulations and standards;
data backup and recovery of data storage module: reliable data backup and recovery strategies are designed to cope with the situation of data loss or system failure. Periodic data backup, snapshot techniques, disaster recovery backup, or the like may be considered to ensure data recoverability and persistence.
Data storage module data management and lifecycle management: data management and lifecycle management policies are introduced to manage lifecycle, storage costs, and access frequency of data. The storage location of the data, the data migration policy, and the data deletion mechanism may be determined according to the importance of the data, the access pattern, and the storage requirements.
The user interface module is responsible for interacting with a user and comprises functions of triggering fingerprint acquisition, data query and result display;
the user interface module designs a user-friendly graphical interface or command line interface, which is convenient for a user to operate the system and obtain the result.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the above-described embodiments, and that the above-described embodiments and descriptions are only preferred embodiments of the present invention, and are not intended to limit the invention, and that various changes and modifications may be made therein without departing from the spirit and scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (7)
1. The fingerprint sampling data storage system based on the GPGPU is characterized by comprising a data acquisition module, an image preprocessing module, a feature extraction module, a feature matching module, a data storage module and a user interface module;
the data acquisition module transmits the acquired original fingerprint image data to the image preprocessing module for processing;
the image preprocessing module is used for receiving the original fingerprint image data from the data acquisition module, carrying out denoising, enhancing and normalizing preprocessing steps on the image, generating preprocessed fingerprint image data, and transmitting the preprocessed fingerprint image data to the feature extraction module;
the feature extraction module receives the preprocessed fingerprint image data from the image preprocessing module, extracts fingerprint features including minutiae points and the direction of a ridge from the image by using a fingerprint feature extraction algorithm, generates extracted fingerprint feature data, and transmits the extracted fingerprint feature data to the feature matching module;
the feature matching module receives the fingerprint feature data from the feature extraction module, compares and matches the acquired fingerprint features with stored fingerprint features, generates a matching result, and transmits the result to the user interface module;
the data storage module is used for storing the collected original fingerprint image data, the fingerprint image data generated by preprocessing and the fingerprint feature data, receiving the fingerprint feature data extracted by the feature extraction module and storing the fingerprint feature data into a database and a file system;
the user interface module provides an interface for interaction between a user and the system, and comprises functions of triggering fingerprint acquisition, inquiring stored fingerprint data and displaying matching results, and interacts with the data acquisition module, the feature matching module and the data storage module to transmit instructions of the user and acquire corresponding results.
2. The GPGPU-based fingerprint sampling data storage system of claim 1, wherein the data acquisition module is responsible for acquiring raw fingerprint image data from a fingerprint acquisition device based on communication with an interface of a fingerprint scanner and a sensor, acquiring a high quality fingerprint image;
the data acquisition module takes the format, sampling rate and sampling quality of the data into account.
3. The GPGPU-based fingerprint sample data storage system of claim 1, wherein the image preprocessing module preprocesses the acquired fingerprint image data, wherein the preprocessing step comprises image denoising, enhancement, and normalization.
4. The GPGPU-based fingerprint sample data storage system of claim 1, wherein the feature extraction module extracts representative features from the preprocessed fingerprint image, and a common fingerprint feature extraction algorithm comprises a Minutiae-based method and a Ridge-based method, and the fingerprint extraction algorithm identifies Minutiae features including Minutiae and Ridge directions in the fingerprint image.
The feature extraction module automatically learns features in the fingerprint image using a Convolutional Neural Network (CNN) and other deep learning models;
the feature extraction module is based on an image enhancement and denoising method, uses an image enhancement and denoising technology to improve the quality of fingerprint images, and applies an image denoising algorithm, enhances contrast and adjusts image brightness to improve the reliability of fingerprint features;
the feature extraction module introduces context information, introduces the context information of surrounding and adjacent fingerprint images except a single fingerprint image by means of multi-mode fusion, multi-scale processing or sequence model and the like, performs joint processing by using a plurality of fingerprint images, and captures richer feature information;
the feature extraction module increases robustness and anti-interference capability, a robustness algorithm is used for processing the problems that a fingerprint image faces various interferences and noises including fingerprint rotation deformation and local shielding in a feature extraction process, and a method including rotation invariance, local ridge tracking and ridge subdivision is introduced in the feature extraction process;
the feature extraction module further optimizes the computing performance, reduces the computing complexity by utilizing the parallel computing capability of the GPU and using an approximate computing method, and reduces the computing amount and accelerates the feature extraction process by using an efficient computing method and algorithm.
5. The GPGPU-based fingerprint sampling data storage system of claim 1, wherein the feature matching module compares the collected fingerprint features with stored fingerprint features to determine if there is a match;
the feature matching module uses a matching algorithm comprising a feature-based matching algorithm and a model-based matching algorithm;
the feature matching module initializes the feature extraction module by using a pre-trained deep learning model (such as a model on ImageNet) based on feature extraction of transfer learning, and adapts fingerprint image data through fine tuning and training in a specific field;
the transfer learning method utilizes the feature representation capability of a model pre-trained on a large-scale data set and performs effective feature extraction on limited fingerprint data;
the feature matching module performs weak supervision learning, and reduces dependence on labeling data by using a weak supervision learning method comprising semi-supervision and unsupervised learning through self-supervision learning and technology of generating an antagonistic network or clustering;
the feature matching module combines the multi-mode information with the temperature, the pressure and the additional biological feature information of the capacitor of the fingerprint image to perform feature extraction, and provides more auxiliary information to enhance the accuracy and the robustness of feature extraction.
The feature matching module uses non-local features, and introduces a non-local feature extraction method, wherein the extraction method comprises the overall line direction, the layout of lines and a topological structure;
the feature matching module is based on the feature extraction of self-adaptive learning, and aims at the difference between individuals, and the self-adaptive feature is extracted by adopting a self-adaptive learning method through personalized model parameter adjustment, dynamically adjusted feature weights or an adaptive mechanism using individual feature expression;
the robustness of the feature matching module is enhanced, and the robustness of feature extraction is improved by using image enhancement, denoising and restoration technologies, so that various interferences and noises including dirt, illumination change and distortion are faced to the fingerprint image;
the feature matching module also incorporates resistance training or data augmentation techniques.
6. The GPGPU-based fingerprint sampling data storage system of claim 1, wherein the data storage module is configured to provide an efficient data storage system for storing the collected fingerprint data and corresponding characteristic data;
the data storage module selects a database system comprising a relational database, a NoSQL database and a graph database and a file system comprising a distributed file system, an object storage system and a block storage system to manage and organize data, and ensures the safety and expandability of the data;
the data storage module stores and processes in a distributed manner: consider the use of a distributed storage system and a processing framework such as Apache Hadoop, apache Spark, google Cloud Bigtable, or the like. The system can realize the distributed storage and processing of data, improve the expandability and the performance of the system and support the storage and processing requirements of large-scale data.
The data storage module uses lossless and lossy compression algorithms to reduce occupation of storage space and bandwidth consumption of data transmission;
the data storage module uses lossless and lossy compression algorithms, and introduces a data redundancy and fault tolerance mechanism; using Redundant Array (RAID) and distributed data backup lossless and lossy compression algorithms to prevent data loss and system failure;
the data storage module designs an efficient data index structure and a query optimization algorithm by using the B+ tree, inverted index and hash-based index technology, so that the retrieval and query operation of data are accelerated;
the data storage module introduces a security and privacy protection mechanism comprising data encryption, access control, authority management and anonymization technologies, protects confidentiality and integrity of fingerprint sampling data, and complies with related privacy regulations and standards;
the data storage module performs data backup and recovery: reliable data backup and recovery strategies are designed to cope with the situation of data loss or system failure. Periodic data backup, snapshot techniques, disaster recovery backup, or the like may be considered to ensure data recoverability and persistence.
The data storage module manages data and lifecycle: data management and lifecycle management policies are introduced to manage lifecycle, storage costs, and access frequency of data. The storage location of the data, the data migration policy, and the data deletion mechanism may be determined according to the importance of the data, the access pattern, and the storage requirements.
7. The GPGPU-based fingerprint sampling data storage system of claim 1, wherein the user interface module is responsible for interacting with a user, including triggering of fingerprint acquisition, data querying, and results presentation functions;
the user interface module designs a user-friendly graphical interface or command line interface, which is convenient for a user to operate the system and acquire results.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310795505.0A CN116798076A (en) | 2023-06-30 | 2023-06-30 | Fingerprint sampling data storage system based on GPGPU |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310795505.0A CN116798076A (en) | 2023-06-30 | 2023-06-30 | Fingerprint sampling data storage system based on GPGPU |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116798076A true CN116798076A (en) | 2023-09-22 |
Family
ID=88037306
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310795505.0A Pending CN116798076A (en) | 2023-06-30 | 2023-06-30 | Fingerprint sampling data storage system based on GPGPU |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116798076A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117481667A (en) * | 2023-10-24 | 2024-02-02 | 沈阳工业大学 | Electroencephalogram signal acquisition system |
| CN118379203A (en) * | 2024-06-26 | 2024-07-23 | 杭州平祥数字技术有限公司 | Vehicle running image enhancement processing method and auxiliary running system in black light environment |
-
2023
- 2023-06-30 CN CN202310795505.0A patent/CN116798076A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117481667A (en) * | 2023-10-24 | 2024-02-02 | 沈阳工业大学 | Electroencephalogram signal acquisition system |
| CN118379203A (en) * | 2024-06-26 | 2024-07-23 | 杭州平祥数字技术有限公司 | Vehicle running image enhancement processing method and auxiliary running system in black light environment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Naikal et al. | Informative feature selection for object recognition via sparse PCA | |
| CN112437926B (en) | Fast robust friction ridge patch detail extraction using feedforward convolutional neural network | |
| Fourati et al. | Anti-spoofing in face recognition-based biometric authentication using image quality assessment | |
| CN116798076A (en) | Fingerprint sampling data storage system based on GPGPU | |
| CN111814574A (en) | Face living body detection system, terminal and storage medium applying double-branch three-dimensional convolution model | |
| CN109543548A (en) | A kind of face identification method, device and storage medium | |
| Kapoor et al. | Hybrid local phase quantization and grey wolf optimization based SVM for finger vein recognition | |
| US20230076017A1 (en) | Method for training neural network by using de-identified image and server providing same | |
| CN109241905B (en) | Image processing method and device | |
| Ilankumaran et al. | Multi-biometric authentication system using finger vein and iris in cloud computing | |
| González‐Soler et al. | On the generalisation capabilities of Fisher vector‐based face presentation attack detection | |
| Khalaf et al. | Robust partitioning and indexing for iris biometric database based on local features | |
| CN118172635A (en) | Image feature enhancement algorithm integrating frequency domain and spatial domain features | |
| Einy et al. | IoT Cloud‐Based Framework for Face Spoofing Detection with Deep Multicolor Feature Learning Model | |
| Fan et al. | A deep learning framework for face verification without alignment | |
| Al-Saidi et al. | Iris features via fractal functions for authentication protocols | |
| Kumar et al. | Face recognition with frame size reduction and DCT compression using PCA algorithm | |
| Juneja et al. | Compression-robust and fuzzy-based feature-fusion model for optimizing the iris recognition | |
| Shreya et al. | Gan-enable latent fingerprint enhancement model for human identification system | |
| CN110781724A (en) | Face recognition neural network, method, device, equipment and storage medium | |
| Xie et al. | Two-stage fusion of local binary pattern and discrete cosine transform for infrared and visible face recognition | |
| Wang et al. | Shadow compensation and illumination normalization of face image | |
| Wasmi et al. | Comparison between proposed convolutional neural network and KNN for finger vein and palm print | |
| Zou et al. | An OCaNet model based on octave convolution and attention mechanism for iris recognition | |
| CN114913610A (en) | Multi-mode identification method based on fingerprints and finger veins |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |