CN115935423A - Driving recording method and system with desensitized key privacy information and storage medium - Google Patents
Driving recording method and system with desensitized key privacy information and storage medium Download PDFInfo
- Publication number
- CN115935423A CN115935423A CN202211671105.0A CN202211671105A CN115935423A CN 115935423 A CN115935423 A CN 115935423A CN 202211671105 A CN202211671105 A CN 202211671105A CN 115935423 A CN115935423 A CN 115935423A
- Authority
- CN
- China
- Prior art keywords
- image
- privacy information
- desensitized
- information
- module
- 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
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Abstract
The invention discloses a driving recording method, a system and a storage medium with desensitized key privacy information, wherein the method comprises the following steps: acquiring a data set of a driving record image; training a network model architecture by using a data set to obtain a key privacy information positioning model, and determining the positioning of key privacy information in a driving record image by using the model to obtain a region to be desensitized; inputting the area to be desensitized into a privacy information desensitization module based on a chaotic encryption system to generate an encrypted image and realize desensitization of the area to be desensitized; the acquired images in the actual vehicle running process are input into a key privacy information positioning model, the output region to be desensitized is input into a privacy information desensitization module, the generated encrypted images are covered with information of the corresponding positions of the original images, and the desensitized images are obtained and stored in a storage medium. According to the invention, through constructing the network model architecture and the privacy information desensitization module, desensitization of key privacy information in the driving record is realized, and the key information of the driving record is reserved.
Description
Technical Field
The invention belongs to the technical field of image processing and deep learning, and relates to a driving recording method and system with desensitized key privacy information, terminal equipment and a storage medium.
Background
In order to standardize automobile data processing activities, protect legal rights and interests of individuals and organizations, maintain national security and social public interests and promote reasonable development and utilization of automobile data, the nation establishes regulations (trial) of automobile data security management, and the regulations are implemented from 2021 to 10/1. According to the sixth fourth point in the regulation, the nation encourages the legal and effective utilization of the automobile data, advocates that automobile data processors adhere to the desensitization treatment principle in developing the automobile data processing activities, and performs anonymization, de-identification and other treatments as far as possible.
The driving recording method can record the time, speed, image, sound and other related information of the vehicle during driving, not only can maintain the legal rights and interests of a driver, but also can supervise the legal driving of the driver, and meanwhile, provides effective and accurate basis for the traffic police, the court, the insurance company and other mechanisms to process traffic accident cases, and is convenient for the traffic police, the court, the insurance company and other mechanisms to process the cases quickly and accurately.
With the continuous development of automobile data recording technology in recent years, different automobile data recorders are suitable for different scenes, such as high definition, night vision, wide angle, double lenses and the like, and the recording range and the recording capacity are continuously expanded and improved. However, since the driving route of the vehicle is random and wide in range, the continuous recording of the driving recorder will inevitably result in a great amount of private information and key information related to public safety being recorded. Because the recorded information of the automobile data recorder is stored in the data storage medium on the automobile data recorder and can be easily acquired by individuals, the data security of key privacy information such as human faces, license plates and the like cannot be guaranteed, and before the data is stored in the storage medium, desensitization treatment needs to be carried out on the automobile data.
Different from a camera for limiting the position and the angle, the automobile data recorder is influenced by various factors during actual working, such as a license plate and a face image with a large offset angle, such as a face image of a mask, and the like.
Disclosure of Invention
In view of this, the invention provides a driving recording method, a system, a terminal device and a storage medium with desensitized key privacy information, which determine the location of the key privacy information in driving recording images by constructing a network model architecture, thereby solving the problems of numerous interference of the recorded images and easy omission and false detection of the key privacy information in the driving process; by constructing the privacy information desensitization module and performing desensitization processing on the positioned key privacy information by adopting a chaotic encryption algorithm, the condition that an individual cannot acquire the key privacy information through a storage medium is ensured, the desensitization problem of the key privacy information in the driving record is solved, and the key information recording function of the driving record is kept.
The first purpose of the invention is to provide a driving recording method with desensitized key privacy information.
A second object of the invention is to provide a vehicle recording system with desensitization of key private information.
A third object of the present invention is to provide a terminal device.
It is a fourth object of the present invention to provide a storage medium.
The first purpose of the invention can be achieved by adopting the following technical scheme:
a method of vehicle recording with desensitized critical private information, the method comprising:
acquiring a data set of a driving record image; the driving recording image at least comprises one of human eyes and a license plate;
building a network model architecture, and identifying human eyes and license plate information in the driving recording image through the network model architecture; training a network model architecture by using the data set to obtain a key privacy information positioning model, and determining the positioning of key privacy information in a driving record image by using the key privacy information positioning model to obtain an area to be desensitized; the key privacy information comprises a face and a license plate;
inputting the area to be desensitized into a privacy information desensitization module based on a chaotic encryption system to generate an encrypted image so as to desensitize the area to be desensitized;
inputting the collected images in the actual vehicle running process into the key privacy information positioning model, inputting the output region to be desensitized into a privacy information desensitization module, and covering the generated encrypted images with the information of the corresponding positions of the original images to obtain desensitized images; storing the desensitized image in a storage medium.
Further, the determining the location of the key privacy information in the driving record image by using the key privacy information location model to obtain the area to be desensitized includes:
positioning human eyes and a license plate in the driving recording image by using a key privacy information positioning model;
determining the positioning of the human face according to the positioning of human eyes;
and combining the positioning information of the face and the license plate to obtain an area to be desensitized.
Further, the network model architecture comprises a preprocessing module, a skeleton module for feature extraction, a pyramid module for feature fusion, and a Loss and NMS module at an output end, wherein:
the preprocessing module is used for preprocessing the driving record image;
the skeleton module for feature extraction adopts a Focus structure and a CSP structure in CSPNet and is used for extracting features of the preprocessed image; the CSP structure comprises a convolution layer and a residual error unit and is used for extracting features of the input feature diagram;
the pyramid module with the fused features adopts a fused structure of an FPN structure and a PAN structure, wherein the FPN structure is an up-sampling process and conveys strong semantic information from top to bottom; the PAN structure is a characteristic pyramid from bottom to top, and strong positioning information is transmitted from bottom to top; the pyramid module integrates parameters and extracts features of the feature layers with different sizes from different trunk layers, so that more information is extracted according to the extracted features;
and the Loss adopts GIOU _ Loss as a Loss function of the target detection frame, and the NMS module is used for screening the predicted target detection frame.
Further, positioning the human eyes in the driving recording image by using a key privacy information positioning model comprises:
screening the obtained prediction target detection frame by using an NMS module, and using the screened prediction target detection frame as a prediction frame of human eyes to realize the positioning of the human eyes;
and in the same way, the screened prediction target detection frame is used as the prediction frame of the license plate, so that the license plate is positioned.
Further, let intersection a be the area of the overlapping portion of the real target detection frame and the predicted target detection frame, and set B be the total area occupied by the real target detection frame and the predicted target detection frame, then:
and recording the minimum external rectangular area of the real target detection frame and the predicted target detection frame as C, obtaining a Loss function GIOU _ Loss:
further, the preprocessing comprises Mosaic data enhancement, cmBN and a self-adaptive image scaling processing mode, wherein the Mosaic data enhancement comprises random arrangement, random scaling and random cutting of the image.
Further, the area to be desensitized is used as an image to be encrypted;
inputting the area to be desensitized into a privacy information desensitization module based on a chaotic encryption system to generate an encrypted image, wherein the steps comprise:
generating an encryption sequence according to the chaotic mapping function and the pixel size of the image to be encrypted;
obtaining a chaotic sequence according to the encryption sequence;
generating an information sequence to be encrypted according to the image to be encrypted;
and encrypting the information to be encrypted by using the chaotic sequence, and generating an encrypted image according to the encrypted sequence. The second purpose of the invention can be achieved by adopting the following technical scheme:
a vehicle recording system with desensitization to critical private information, the system comprising:
the data set acquisition module is used for acquiring a data set of the driving record image; the driving recording image at least comprises one of human eyes and a license plate;
the key privacy information positioning module is used for building a network model architecture and identifying human eyes and license plate information in the driving recording image through the network model architecture; training a network model architecture by using the data set to obtain a key privacy information positioning model, and determining the positioning of key privacy information in a driving record image by using the key privacy information positioning model to obtain an area to be desensitized; the key privacy information comprises a face and a license plate;
the key privacy information desensitization module is used for inputting the area to be desensitized into the privacy information desensitization module based on the chaotic encryption system to generate an encrypted image so as to realize desensitization of the area to be desensitized;
the desensitized image generation module is used for inputting the acquired image in the actual vehicle driving process into the key privacy information positioning model, inputting the output region to be desensitized into the privacy information desensitization module, and covering the generated encrypted image with the information of the corresponding position of the original image to obtain the desensitized image; storing the desensitized image in a storage medium.
The third purpose of the invention can be achieved by adopting the following technical scheme:
the terminal equipment comprises a processor and a memory for storing a program executable by the processor, wherein when the processor executes the program stored in the memory, the vehicle recording method with desensitized key privacy information is realized.
The fourth purpose of the invention can be achieved by adopting the following technical scheme:
a storage medium stores a program which, when executed by a processor, implements the above-described method of vehicle recording with desensitization of key privacy information.
Compared with the prior art, the invention has the following beneficial effects:
1. according to the method, the key privacy information in the driving record image is positioned by constructing a network model architecture. Aiming at the conditions of a face, a large-angle side face, a new energy license plate, a large-angle license plate and the like of a wearer with a mask, the problems that images recorded in the driving process are interfered in a lot, and key privacy information is easy to miss detection and false detection are solved.
2. According to the invention, by constructing the privacy information desensitization module and adopting the chaotic encryption algorithm to desensitize the key privacy information positioned by the key privacy information positioning model, it is ensured that an individual cannot acquire the key privacy information through a storage medium, but law enforcement agencies such as public security and the like can acquire the corresponding key privacy information through decrypting an image at a necessary moment. The desensitization problem of key privacy information is solved, and the key information recording function of driving record is kept.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the structures shown in the drawings without creative efforts.
Fig. 1 is a flowchart of a driving recording method with desensitized key privacy information according to embodiment 1 of the present invention.
Fig. 2 is a flowchart of a key privacy information positioning module according to embodiment 1 of the present invention.
Fig. 3 is a flowchart of a key privacy information desensitization module according to embodiment 1 of the present invention.
Fig. 4 is a block diagram of a structure of a vehicle recording system with desensitized key privacy information according to embodiment 2 of the present invention.
Fig. 5 is a block diagram of a terminal device according to embodiment 3 of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer and more complete, the technical solutions in the embodiments of the present invention will be described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some embodiments of the present invention, but not all embodiments, and all other embodiments obtained by a person of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the protection scope of the present invention. It should be understood that the description of the specific embodiments is intended to be illustrative only and is not intended to be limiting.
Example 1:
the method is developed based on a Pycharm development environment, an OpenCV computer vision library and a Pytorch deep learning framework, wherein OpenCV covers a large number of packaging function interfaces related to image processing, and can complete related image processing tasks; the Pythroch deep learning frame is a deep learning frame with Python priority, can realize strong GPU acceleration and simultaneously supports a dynamic neural network. The Pycharm development environment under the Windows platform is one of the preferable development environments for completing image processing and machine learning tasks at present.
As shown in fig. 1, the driving recording method with desensitized key privacy information provided in this embodiment includes:
and S101, acquiring a data set of the driving record image.
The method comprises the steps of collecting a data set of related images of human eyes and license plates of a vehicle in running by using high-definition cameras, cameras and other equipment in a preset environment, wherein the data set comprises images of human eye feature points, conventional license plates, large-angle license plates and new energy license plates. The preset environment refers to an environment which is in line with road vehicles and can basically reflect real road conditions. And randomly selecting the data set according to the proportion of 8:2 to obtain a training set and a test set.
In the embodiment, the data set comprises 10000 driving recording images of human eyes and a license plate, wherein the training set comprises 4000 images of human eyes and the license plate, only 2000 images of human eyes and only 2000 images of the license plate; the test set contains 1000 images of human eyes and license plates, only 500 images of human eyes and 500 images of license plates.
S102, building a network model architecture, and identifying human eyes and license plate information in the driving record image through the network model architecture; and training a network model architecture by using the data set to obtain a key privacy information positioning model, and determining the positioning of key privacy information in the driving record image by using the key privacy information positioning model to obtain the area to be desensitized.
Building a network model architecture capable of identifying relevant characteristics of human eyes and a license plate, and training relevant parameters of the human eyes and the license plate by using a training set and a testing set to obtain a key privacy information positioning model, wherein the model is used for detecting the positions of the human eyes and the license plate, and further determining the position of a human face according to the positions of the human eyes, so that the condition of wrong human face position identification or large error caused by wearing a mask is avoided; and meanwhile, the whole key privacy information is positioned by combining the license plate position.
(1) And building a network model architecture.
And (5) building a network model architecture by utilizing a deep learning technology.
The network model architecture comprises a preprocessing module, a feature extraction framework module, a feature fusion pyramid module and a Loss and NMS module at an output end, wherein the output of each module is the input of the next module, and the following modules are adopted as the input of the next module:
(1-1) a pretreatment module.
The preprocessing module is used for processing an original input picture, and aims to highlight the characteristics of the picture, enrich a data set, reduce the training requirement of a GPU and the like.
The preprocessing comprises processing modes such as Mosaic data enhancement, cmBN and self-adaptive image scaling. The Mosaic data enhancement method is used for carrying out image splicing on pictures in a data set through random arrangement, random scaling and random cutting, and the self-adaptive image scaling is used for uniformly scaling input pictures to the same standard size so as to facilitate the subsequent network processing and is an effective image preprocessing mode.
In this embodiment, the preprocessing module processes the image by using a method such as Mosaic data enhancement and adaptive image scaling, the Mosaic data enhancement method includes randomly arranging, randomly scaling, and randomly cutting to perform image splicing on the images in the data set, and the adaptive image scaling is to scale the input images to the same standard size so that the images can be processed by subsequent modules conveniently.
And (1-2) a skeleton module for feature extraction.
The skeleton module for feature extraction is used for extracting features of an image by using a network structure, and a Focus structure and a CSP structure in CSPNet are adopted.
The Focus structure is to perform a series of slicing operations on an input picture, and perform 32 convolution operations on the input picture, so as to achieve the purpose of adjusting the size of the feature map.
In this embodiment, an originally input 608 × 608 × 3 image is subjected to slicing operation of a Focus structure to be changed into a 304 × 304 × 12 feature map, and is subjected to convolution operation of 32 convolution kernels to be changed into a 304 × 304 × 32 feature map.
The CSP structure is composed of a convolutional layer and a residual unit and is used for extracting features of an input feature map. The convolution kernel size of each CSP module is 3 multiplied by 3, stride =2, and the function of down sampling is achieved. After the feature map passes through five CSP blocks, that is, the input feature map is down-sampled five times to become a feature map of 19 × 19 size.
And (1-3) a pyramid module with feature fusion.
The pyramid module of feature fusion is implemented using a fusion structure of FPN and PAN. The FPN structure is an up-sampling process, strong semantic information is conveyed from top to bottom, the size of a feature map is up-sampled from 19 x 19 to 76 x 76, a PAN structure is a feature pyramid from bottom to top, strong positioning information is conveyed, the size of the feature map is down-sampled from 76 x 76 to 19 x 19 and corresponds to a minimum anchor box and a maximum anchor box respectively, and the two pyramid models perform parameter integration and feature extraction on feature layers with different sizes from different stem layers, so that more information of the feature map can be extracted.
(1-4) Loss and NMS module at output.
And the Loss and NMS module at the output end selects a Loss function and screens a prediction box according to a target detection task. The Loss of the target detection box by adopting the GIOU _ Loss is improved and supplemented on the IOU Loss function. The intersection A is recorded as the area of the overlapped part of the real target frame and the prediction frame, and the union B is the total area occupied by the two frames together, including:
noting the minimum bounding rectangle area of the two boxes as C, we obtain a representation of the loss function as follows:
and finally, screening the prediction frames by using an NMS method, and removing redundant, overlapped and inaccurate prediction frames.
(2) And training a network model architecture by using the data set to obtain a key privacy information positioning model, and determining the positioning of key privacy information in the driving record image by using the key privacy information positioning model to obtain the area to be desensitized.
And training related parameters of human eyes and license plates in the key privacy information positioning model by utilizing a human eye and license plate training set and a testing set to obtain the trained key privacy information positioning model.
In the embodiment, an Adam algorithm is used as an optimization parameter of the model, the training period is set to be 300, the batch-size is set to be 128, the learning rate is 0.0001, the model is converged after the model training is completed, and the positioning of human eyes and a license plate is realized.
Positioning and calibrating the human face according to the normal proportion of the human eyes and the human face through the recognized coordinate array of the human eyes, determining the position coordinates of the human face and calibrating the position coordinates as a prediction frame; and outputting the combination of the human face and the prediction frame of the license plate. A flow diagram of a key privacy information location model is shown in fig. 2.
S103, inputting the acquired image in the actual vehicle running process into a key privacy information positioning model, and inputting the output region to be desensitized into a privacy information desensitization module to realize desensitization of the region to be desensitized; and obtaining a desensitized image according to the desensitized region to be desensitized.
The key privacy information positioning module outputs an area to be desensitized (namely a prediction box), and each prediction box represents an image to be desensitized. And the key privacy information desensitization module is used for desensitizing the image to be desensitized.
As shown in fig. 3, a chaos encryption algorithm is used to perform desensitization processing on key privacy information (i.e., an area to be desensitized). The chaotic encryption algorithm needs part of known parameters as encrypted parameter seeds, and the part of parameters are randomly generated and uniformly stored when each device leaves a factory and are used for subsequent data encryption and necessary data decryption.
In this embodiment, it is known that the encrypted parameter seed c (0) =0.4, the chaotic range parameter r =0.3, and the mapping parameter seed t =0.5, and the number n of pixels of the image to be encrypted is 100.
Further, the desensitization treatment is completed by using a chaotic encryption algorithm, and the desensitization treatment comprises the following steps:
(1) The encrypted sequence c (k + 1) is generated by the function F (c (k)) as follows:
f (c (k)) is a chaotic mapping function, k is more than or equal to 0 and less than or equal to 8n, n is the pixel size of an image to be encrypted, and c (k) is the value of the kth bit in an encryption sequence; r is defined as a number in the range of [0,1], which is a known chaotic range parameter. When k =0, c (0) is a known encryption parameter seed.
In this embodiment, an encryption sequence [0.4,0.67,0.44,0.82,0.93, …,0.01] composed of values in the range [0,1] can be obtained by using the chaotic mapping function.
(2) And obtaining the chaotic sequence through the encryption sequence.
The length of the chaotic sequence depends on the length of data to be encrypted.
The expression of the chaotic sequence d (k) is as follows:
wherein G (c (k)) is the mapping function, and t is a value in the range of [0,1], which is the known mapping parameter seed.
In this embodiment, the encrypted sequence [0.4,0.67,0.44,0.82,0.93, …,0.01] is mapped to the values 0 and 1 by G (c (k)), and the chaotic sequence d (k) = [0,1,0,1,1, …,0] is obtained.
The encrypted sequence c (k) can be mapped to values 0 and 1 through G (c (k)), and the chaotic sequence d (k) is obtained.
(3) And generating an information sequence to be encrypted according to the image to be encrypted.
n = i × j, i, j being the total number of rows and the total number of columns, respectively, of the image to be encrypted.
The information sequence b (l) to be encrypted corresponds to eight-bit binary numbers converted from gray values of pixels in the p-th row and the q-th column in the image to be encrypted, wherein:
l＝(p-1)×j+q
wherein l is more than or equal to 0 and less than or equal to n, p is more than or equal to 0 and less than or equal to i, and q is more than or equal to 0 and less than or equal to j.
In the present embodiment, an information sequence to be encrypted b (k) = [1,0,0,1,0,0,0,1, …,1] is generated from an image to be encrypted.
(4) And encrypting the information to be encrypted by utilizing the chaotic sequence, and generating an encrypted image according to the encrypted sequence.
The encrypted positioning information is a sequence w (l):
wherein the content of the first and second substances,
is an exclusive or operation.
The encrypted image is generated by a sequence w (k) corresponding to the eight-bit binary number of the grey value of the pixel of the p-th row, q-th column in the encrypted image.
In this embodiment, the sequence w (k) is [0,0,1,1,0,1,0,1, …,0], the gray level of the pixel in the first row and the first column in the encrypted image is 53 according to the first 8 bits of w (k), and so on, the gray level of the pixel in the entire encrypted image is obtained.
(5) And covering the encrypted image with the information of the corresponding position of the original image to obtain the desensitized image, thereby realizing the desensitization of the key privacy information.
And covering the corresponding encrypted position of the original image by using the encrypted image to obtain an image desensitized to key privacy information.
(6) In the decryption process, the same chaotic range parameter r, the encryption parameter seed c (0) and the mapping parameter seed t are only needed to be used, the steps (1) to (4) are repeated on the image to be decrypted, and the output image is the decrypted image.
And S104, storing the desensitized image into a storage medium.
The individual user can only view the desensitized image stored in the storage medium, and if a judicial authority or a related department needs to call to view the key privacy information in the storage medium, the image can be obtained only by the decryption step in S36.
Those skilled in the art will appreciate that all or part of the steps in the method for implementing the above embodiments may be implemented by a program to instruct associated hardware, and the corresponding program may be stored in a computer-readable storage medium.
It should be noted that although the method operations of the above-described embodiments are depicted in the drawings in a particular order, this does not require or imply that these operations must be performed in this particular order, or that all of the illustrated operations must be performed, to achieve desirable results. Rather, the depicted steps may change the order of execution. Additionally or alternatively, certain steps may be omitted, multiple steps combined into one step execution, and/or one step broken down into multiple step executions.
Example 2:
as shown in fig. 4, this embodiment provides a driving recording system with desensitized key privacy information, which includes a data set obtaining module 401, a key privacy information positioning module 402, a key privacy information desensitizing module 403, and a desensitized image generating module 404, where:
a data set obtaining module 401, configured to obtain a data set of a driving record image; the driving recording image at least comprises one of human eyes and a license plate;
the key privacy information positioning module 402 is used for building a network model architecture, and identifying the information of human eyes and license plates in the driving record image through the network model architecture; training a network model architecture by using the data set to obtain a key privacy information positioning model, and determining the positioning of key privacy information in a driving record image by using the key privacy information positioning model to obtain an area to be desensitized; the key privacy information comprises a human face and a license plate;
the key privacy information desensitization module 403 is configured to input the area to be desensitized into the privacy information desensitization module based on the chaotic encryption system, generate an encrypted image, and implement desensitization of the area to be desensitized;
a desensitized image generation module 404, configured to input the acquired image in the actual vehicle driving process into the key privacy information positioning model, input the output region to be desensitized into the privacy information desensitization module, and cover the generated encrypted image with information of a corresponding position of the original image to obtain a desensitized image; storing the desensitized image in a storage medium.
The specific implementation of each module in this embodiment may refer to embodiment 1, which is not described herein any more; it should be noted that, the apparatus provided in this embodiment is only illustrated by dividing the functional modules, and in practical applications, the functions may be distributed by different functional modules according to needs, that is, the internal structure is divided into different functional modules to complete all or part of the functions described above.
Example 3:
the present embodiment provides a terminal device, which may be a computer, as shown in fig. 5, and includes a processor 502, a memory, an input device 503, a display 504, and a network interface 505, which are connected through a system bus 501, the processor is configured to provide computing and controlling capabilities, the memory includes a nonvolatile storage medium 506 and an internal memory 507, the nonvolatile storage medium 506 stores an operating system, a computer program, and a database, the internal memory 507 provides an environment for the operating system and the computer program in the nonvolatile storage medium to run, and when the processor 502 executes the computer program stored in the memory, the vehicle traveling recording method of embodiment 1 described above is implemented as follows:
acquiring a data set of a driving record image; the driving recording image at least comprises one of human eyes and a license plate;
building a network model architecture, and identifying human eyes and license plate information in the driving recording image through the network model architecture; training a network model architecture by using the data set to obtain a key privacy information positioning model, and determining the positioning of key privacy information in a driving record image by using the key privacy information positioning model to obtain an area to be desensitized; the key privacy information comprises a human face and a license plate;
inputting the area to be desensitized into a privacy information desensitization module based on a chaotic encryption system to generate an encrypted image so as to desensitize the area to be desensitized;
inputting the collected images in the actual vehicle running process into the key privacy information positioning model, inputting the output region to be desensitized into a privacy information desensitization module, and covering the generated encrypted images with the information of the corresponding positions of the original images to obtain desensitized images; storing the desensitized image in a storage medium.
Example 4:
the present embodiment provides a storage medium, which is a computer-readable storage medium, and stores a computer program, and when the computer program is executed by a processor, the method for recording driving according to embodiment 1 is implemented as follows:
acquiring a data set of a driving record image; the driving recording image at least comprises one of human eyes and a license plate;
building a network model architecture, and identifying human eyes and license plate information in the driving recording image through the network model architecture; training a network model architecture by using the data set to obtain a key privacy information positioning model, and determining the positioning of key privacy information in a driving record image by using the key privacy information positioning model to obtain an area to be desensitized; the key privacy information comprises a human face and a license plate;
inputting the area to be desensitized into a privacy information desensitization module based on a chaotic encryption system to generate an encrypted image so as to desensitize the area to be desensitized;
inputting the acquired image in the actual vehicle running process into the key privacy information positioning model, inputting the output region to be desensitized into a privacy information desensitization module, and covering the generated encrypted image with information of the corresponding position of the original image to obtain the desensitized image; storing the desensitized image in a storage medium.
It should be noted that the computer readable storage medium of the present embodiment may be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
In summary, the method provided by the present invention mainly includes acquiring data sets of images of human eyes and license plates: dividing a data set into a training set and a testing set according to a certain proportion according to an existing calibration image containing human eye and license plate information; training a network model architecture to obtain a key privacy information positioning model: training parameters of a network model architecture, and obtaining a model of coordinate information of a prediction frame capable of outputting key privacy information after the parameters are converged; inputting the collected images in the normal running process of the vehicle into a key privacy information positioning model, and outputting a prediction frame of a face to be desensitized and a license plate; and inputting the coordinates of the prediction box into a key privacy information desensitization module for desensitization, generating a desensitized image and storing the desensitized image into a data storage medium to realize the desensitization of the key privacy information.
The above description is only for the preferred embodiments of the present invention, but the protection scope of the present invention is not limited thereto, and any person skilled in the art can substitute or change the technical solution and the inventive concept of the present invention within the scope of the present invention.
Claims (10)
1. A method of vehicle recording with desensitized key privacy information, the method comprising:
acquiring a data set of a driving record image; the driving recording image at least comprises one of human eyes and a license plate;
building a network model architecture, and identifying human eyes and license plate information in the driving recording image through the network model architecture; training a network model architecture by using the data set to obtain a key privacy information positioning model, and determining the positioning of key privacy information in a driving record image by using the key privacy information positioning model to obtain an area to be desensitized; the key privacy information comprises a human face and a license plate;
inputting the area to be desensitized into a privacy information desensitization module based on a chaotic encryption system to generate an encrypted image and realize desensitization of the area to be desensitized;
inputting the collected images in the actual vehicle running process into the key privacy information positioning model, inputting the output region to be desensitized into a privacy information desensitization module, and covering the generated encrypted images with the information of the corresponding positions of the original images to obtain desensitized images; storing the desensitized image in a storage medium.
2. The driving recording method according to claim 1, wherein the determining, by using the key privacy information positioning model, the positioning of the key privacy information in the driving recording image to obtain the area to be desensitized comprises:
positioning human eyes and a license plate in the driving recording image by using a key privacy information positioning model;
determining the positioning of the human face according to the positioning of human eyes;
and combining the positioning information of the face and the license plate to obtain an area to be desensitized.
3. The vehicle recording method according to claim 2, wherein the network model architecture comprises a preprocessing module, a skeleton module for feature extraction, a pyramid module for feature fusion, and a Loss and NMS module at output, wherein:
the preprocessing module is used for preprocessing the driving record image;
the skeleton module for feature extraction adopts a Focus structure and a CSP structure in CSPNet and is used for extracting features of the preprocessed image; the CSP structure comprises a convolution layer and a residual error unit and is used for extracting features of the input feature diagram;
the pyramid module with the fused features adopts a fused structure of an FPN structure and a PAN structure, wherein the FPN structure is an up-sampling process and conveys strong semantic information from top to bottom; the PAN structure is a characteristic pyramid from bottom to top, and strong positioning information is transmitted from bottom to top; the pyramid module integrates parameters and extracts features of the feature layers with different sizes from different trunk layers, so that more information is extracted according to the extracted features;
and the Loss adopts GIOU _ Loss as a Loss function of the target detection box, and the NMS module is used for screening the predicted target detection box.
4. The driving recording method according to claim 3, wherein the locating human eyes in the driving recording image by using the key privacy information locating model comprises:
screening the obtained prediction target detection frame by using an NMS module, and using the screened prediction target detection frame as a prediction frame of human eyes to realize the positioning of the human eyes;
and in the same way, the screened prediction target detection frame is used as the prediction frame of the license plate, so that the license plate is positioned.
5. The driving recording method according to claim 3, wherein the intersection a is an area of an overlapping portion of the real target detection frame and the predicted target detection frame, and the intersection B is a total area occupied by the real target detection frame and the predicted target detection frame, then:
recording the minimum circumscribed rectangle area of the real target detection frame and the predicted target detection frame as C, obtaining a Loss function GIOU _ Loss:
6. the driving recording method according to claim 3, wherein the preprocessing comprises a Mosaic data enhancement, cmBN, and an adaptive image scaling processing mode, wherein the Mosaic data enhancement comprises random arrangement, random scaling, and random cropping of the image.
7. The driving recording method according to any one of claims 1 to 6, characterized in that the area to be desensitized is taken as the image to be encrypted;
inputting the area to be desensitized into a privacy information desensitization module based on a chaotic encryption system to generate an encrypted image, wherein the steps comprise:
generating an encryption sequence according to the chaotic mapping function and the pixel size of the image to be encrypted;
obtaining a chaotic sequence according to the encrypted sequence;
generating an information sequence to be encrypted according to the image to be encrypted;
and encrypting the information to be encrypted by using the chaotic sequence, and generating an encrypted image according to the encrypted sequence.
8. A vehicle recording system with desensitization to key private information, the system comprising:
the data set acquisition module is used for acquiring a data set of the driving record image; the driving recording image at least comprises one of human eyes and a license plate;
the key privacy information positioning module is used for building a network model architecture and identifying human eyes and license plate information in the driving recording image through the network model architecture; training a network model architecture by using the data set to obtain a key privacy information positioning model, and determining the positioning of key privacy information in a driving record image by using the key privacy information positioning model to obtain an area to be desensitized; the key privacy information comprises a human face and a license plate;
the key privacy information desensitization module is used for inputting the area to be desensitized into the privacy information desensitization module based on the chaotic encryption system to generate an encrypted image so as to realize desensitization of the area to be desensitized;
the desensitized image generation module is used for inputting the acquired image in the actual vehicle running process into the key privacy information positioning model, inputting the output region to be desensitized into the privacy information desensitization module, and covering the generated encrypted image with the information of the corresponding position of the original image to obtain the desensitized image; storing the desensitized image in a storage medium.
9. A terminal device comprising a processor and a memory for storing a program executable by the processor, wherein the processor implements the driving recording method according to any one of claims 1 to 7 when executing the program stored in the memory.
10. A storage medium storing a program, wherein the program, when executed by a processor, implements the driving recording method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211671105.0A CN115935423A (en) | 2022-12-26 | 2022-12-26 | Driving recording method and system with desensitized key privacy information and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211671105.0A CN115935423A (en) | 2022-12-26 | 2022-12-26 | Driving recording method and system with desensitized key privacy information and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115935423A true CN115935423A (en) | 2023-04-07 |
Family
ID=86700524
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211671105.0A Pending CN115935423A (en) | 2022-12-26 | 2022-12-26 | Driving recording method and system with desensitized key privacy information and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115935423A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116664849A (en) * | 2023-05-18 | 2023-08-29 | 中关村科学城城市大脑股份有限公司 | Data processing method, device, electronic equipment and computer readable medium |
| CN117132768A (en) * | 2023-10-27 | 2023-11-28 | 广汽埃安新能源汽车股份有限公司 | License plate and face detection and desensitization method and device, electronic equipment and storage medium |
-
2022
- 2022-12-26 CN CN202211671105.0A patent/CN115935423A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116664849A (en) * | 2023-05-18 | 2023-08-29 | 中关村科学城城市大脑股份有限公司 | Data processing method, device, electronic equipment and computer readable medium |
| CN116664849B (en) * | 2023-05-18 | 2024-01-16 | 中关村科学城城市大脑股份有限公司 | Data processing method, device, electronic equipment and computer readable medium |
| CN117132768A (en) * | 2023-10-27 | 2023-11-28 | 广汽埃安新能源汽车股份有限公司 | License plate and face detection and desensitization method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115935423A (en) | Driving recording method and system with desensitized key privacy information and storage medium | |
| CN110678873A (en) | Attention detection method based on cascade neural network, computer device and computer readable storage medium | |
| JP6597609B2 (en) | Image processing apparatus, monitoring system, image processing method, and program | |
| CN112216049A (en) | Construction warning area monitoring and early warning system and method based on image recognition | |
| CN109740420A (en) | Vehicle illegal recognition methods and Related product | |
| EP3920165A1 (en) | Sensor device and encryption method | |
| CN108875750B (en) | Object detection method, device and system and storage medium | |
| CN112364778A (en) | Power plant safety behavior information automatic detection method based on deep learning | |
| CN116186770A (en) | Image desensitizing method, device, electronic equipment and storage medium | |
| CN109831648A (en) | Antitheft long-distance monitoring method, device, equipment and storage medium | |
| CN112132216B (en) | Vehicle type recognition method and device, electronic equipment and storage medium | |
| CN112037142A (en) | Image denoising method and device, computer and readable storage medium | |
| CN115830449A (en) | Remote sensing target detection method with explicit contour guidance and spatial variation context enhancement | |
| JP7211428B2 (en) | Information processing device, control method, and program | |
| US11244168B2 (en) | Method of highlighting an object of interest in an image or video | |
| CN112883843A (en) | Driver visual salient region detection method and device and computer equipment | |
| CN111064924A (en) | Video monitoring method and system based on artificial intelligence | |
| CN115240168A (en) | Perception result obtaining method and device, computer equipment and storage medium | |
| KR102089230B1 (en) | Apparatus and method for predicting vehicle speed using image | |
| CN113762053A (en) | Image processing method and device, computer and readable storage medium | |
| CN110659384B (en) | Video structured analysis method and device | |
| CN113569785A (en) | Driving state sensing method and device | |
| CN117235819B (en) | Intelligent operation method and device of experimental equipment | |
| RU2775162C1 (en) | System and method for tracking moving objects by video data | |
| WO2023189544A1 (en) | Position estimation device, position estimation method, and program |
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 |