Data backup management method and system for automobile data recorder

Abstract

The invention discloses a data backup management method and system for an automobile data recorder, relates to the technical field of data backup management, solves the technical problems of data storage structuralization and safety deficiency and passive and low-efficiency storage space optimization, and aims to calculate the use frequency through a quantitative standard and accurately divide high-frequency / intermediate-frequency / low-frequency scenes. A differential backup strategy is matched, high-frequency scene data backup lagging is avoided, low-frequency scene redundancy is reduced, backup efficiency is improved, a three-level storage system is established, core event data are classified independently, retrieval time is shortened, meanwhile, three-level permission of an administrator, a common user and a temporary authorized user is set, unauthorized access is effectively prevented, and data backup efficiency is improved. The core data privacy security guarantee capability is improved, low-frequency non-important data is transferred to cold storage according to data access frequency and importance dynamic optimization, high-frequency data is encoded by H.265, and the local storage utilization rate is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data backup management, in particular to a data backup management method and system for a driving recorder. BACKGROUND

[0002] With the continuous growth of the number of cars and the improvement of road traffic safety awareness, driving recorders have become standard equipment for vehicles. The core value of driving recorders lies in recording daily driving trajectories, providing key evidence for traffic collisions / accidents, and meeting users' recording needs for scenery and special road conditions. Currently, driving recorders generally support 4K high-definition recording, and the data volume per hour can reach 3-5GB. In addition, users have diverse driving scenarios, and the amount of driving record data is growing explosively, which puts higher requirements on the efficiency, security, and storage optimization of data management.

[0003] In the prior art, data storage is based on a single dimension of date, lacks structured classification, and causes core event data to be mixed with daily data, resulting in low retrieval efficiency. The duplicate checking function is limited to comparing single attributes such as file names and file sizes, and cannot identify fragment duplicates or format difference duplicates, resulting in a high misjudgment rate. Secondly, the user is only reminded to manually delete when the local storage is full, and the data access frequency and importance are not dynamically optimized. Data recovery relies on a single backup node, and when core event data is damaged or the backup fails, there is no effective reconstruction mechanism, and the evidence value is easily lost. SUMMARY

[0004] To overcome the shortcomings of the prior art, the present application provides a data backup management method and system for a driving recorder, which solves the problems of lack of structured data storage and security, and passive and inefficient storage space optimization.

[0005] To achieve the above purpose, the present application realizes the following technical scheme: a data backup management method for a driving recorder, which specifically includes the following steps: Step one, counting the effective number of times of using the driving recorder within a preset period, calculating the frequency of use, comparing it with a preset comparison threshold, dividing it into three categories of high-frequency use, medium-frequency use, and low-frequency use, and performing corresponding backup processing to generate backup processing information corresponding to the scene; Step two, obtaining the driving record data after backup based on the backup processing information, performing hierarchical storage, obtaining the driving record data in each storage level, identifying the completely repeated data, fragment repeated data, and difference repeated data through a multi-level duplicate checking method, and then forming a repeated classification set; Step three, different repeat types in the repeated classification set are processed respectively, for the completely repeated data, the association with the abnormal event is judged, secondary confirmation or recycle bin transfer is performed according to the association result, and the storage node index is updated synchronously; for the fragment repeated data, the timeline view and key frame comparison information are generated, and the retention suggestion is generated in combination with the data integrity and key degree; for the difference repeated data, whether it is repeated data is determined according to the feature coincidence degree and time watermark consistency, and the merging or retention processing is performed; Step four, the remaining space of each storage node is checked regularly, when the remaining space is lower than the preset threshold, the space release mechanism is triggered, the data is divided into high-frequency access data, medium-frequency access data and low-frequency access data according to the access frequency, the low-frequency access data is transferred to the cold storage device or the cloud storage low-price storage layer; the high-frequency access data is checked and converted by using a more efficient compression algorithm; Step five, the integrity of the driving record data is checked regularly by using the checksum or hash value comparison method, if the data is damaged, the local backup or cloud backup is recovered preferentially; if the backup is not available, the data reconstruction process is started for the key data, the data recovery and reconstruction process information is recorded, and the process log is formed.

[0006] As a further scheme of the application, the way of dividing into three types of high-frequency use, medium-frequency use and low-frequency use is: If the use frequency is greater than P max , it is classified as high-frequency use, if the use frequency is in the comparison threshold interval, it is classified as medium-frequency use, and if the use frequency is less than P min , it is classified as low-frequency use.

[0007] As a further scheme of the application, the way of generating backup processing information corresponding to the scene is: For the high-frequency use scene, fixed backup once a week + emergency scene immediate backup are adopted, and the high-frequency backup processing information is generated; For the medium-frequency use scene, backup within twelve hours after each driving is adopted + monthly comprehensive arrangement is adopted, and the medium-frequency backup processing information is generated; For the low-frequency use scene, immediate backup after single trip and final backup before parking are adopted, and the low-frequency backup processing information is generated.

[0008] As a further scheme of the application, the multi-level duplicate detection method is: The first level, if the two data time stamps are completely same, it is determined as completely repeated data; if the time stamps are partially overlapped, it enters the second level verification; The second level, if the length, file size and resolution of the two data are completely consistent, it is determined as repeated data; if any parameter is different, it enters the third level identification; The third level determines whether the data content is completely repeated through image feature comparison or video content analysis, and further distinguiates the fragment repeated data and the difference repeated data.

[0009] As a further scheme of the present application, the processing details of the completely repeated data include: If the completely repeated data is associated with an abnormal event, a user secondary confirmation pop-up window is triggered, and the subsequent operation is executed after the user confirmation; If the completely repeated data is not associated with an abnormal event, it is directly transferred to the repeated data recycling station, and the default retention period t of the recycling station is set by the operator; After the recycling station is transferred, the indexes of the local core node, the local extension node and the cloud node are synchronously updated, and an index verification report is generated after synchronous completion.

[0010] As a further scheme of the present application, the processing details of the fragment repeated data include: In the time axis view, the original file is marked in red, the repeated fragment is marked in blue, and the overlapping area is highlighted in yellow. 3-5 key frames are automatically extracted, including the accident core frame or the road sign frame; If the fragment is an accident core frame, it is recommended to keep the original file + repeated fragment and mark the associated ID; if the fragment is a daily commuting ordinary fragment, it is recommended to keep only the original file.

[0011] As a further scheme of the present application, the processing details of the difference repeated data include: The video key frame is converted to an 8x8 grayscale image by a Python script, a dHash value is generated, the similarity is calculated by Hamming distance, and the similarity is ≥80% when the Hamming distance≤5. For videos with dHash similarity≥70%, sample at 1 frame per second, and extract the road marking and time watermark features of each frame using OpenCV; If the feature coincidence frame ratio≥75% and the time watermark is completely consistent, it is determined as repeated data, and one complete data is kept after merging, and a merging log is recorded; if the coincidence frame ratio is <75% or the time watermark is inconsistent, it is determined as non-repeated data, and two copies are kept and a difference explanation is added.

[0012] As a further scheme of the present application, the quantitative standard of data access frequency is: High-frequency access data, effective access≥3 times in the last 30 days; Medium-frequency access data, effective access 1-2 times in the last 30 days; Low-frequency access data, effective access 0 times in the last 30 days and effective access≤1 time in the last 90 days.

[0013] As a further scheme of the present application, the quantitative standard of important data is: Core important data, accident, scratch, violation related data, or user manually marked important data; Generally important data, special scene recording; Non-important data, daily commuting records without special scenes, redundant data backed up to 3 or more nodes.

[0014] The data backup management system for the driving recorder comprises: The scene classification calculation module is used for automatically counting the effective vehicle use times in a time period through the driving recorder supporting APP, and calculating the use frequency according to the formula use frequency = effective vehicle use times / time period, matching the use frequency with a preset comparison threshold value, dividing the use frequency into high-frequency use, medium-frequency use and low-frequency use scenes, and simultaneously transmitting the classified use scenes to the data backup processing module. The data backup processing module is used for executing corresponding backup processing according to different use scenes, adopting weekly fixed backup + emergency scene instant backup for the high-frequency use scene, adopting backup within twelve hours after the end of vehicle use + monthly overall arrangement for the medium-frequency use scene, and adopting immediate backup after the end of single trip + final backup before parking for the low-frequency use scene, generating backup processing information corresponding to the scenes, and transmitting the backup processing information to the hierarchical storage module. The hierarchical storage module is used for acquiring driving record data based on the backup processing information, storing the data in a three-level hierarchy of year-month-event type, setting access permissions for each level, and identifying complete repeated data, fragment repeated data and difference repeated data through a hierarchical process of time stamp comparison, time length / file size / resolution verification and data content identification, forming a repeated classification set, and transmitting the repeated classification set to the repeated data processing module. The repeated data processing module is used for processing the repeated classification set by type, performing secondary confirmation or recycle bin transfer for the complete repeated data according to the relevance with abnormal events, generating time axis view and key frame comparison information for the fragment repeated data, generating retention suggestions and persisting the associated relationship, and determining whether the difference repeated data is repeated or not through dHash algorithm and OpenCV frame feature extraction, and performing merging or retention processing, simultaneously generating repeated data processing information, and transmitting the repeated data processing information to the space optimization module. The space optimization module is used for regularly checking the remaining space of the storage node, triggering space release when the remaining space is lower than a preset threshold value, transferring non-important data in low-frequency access data beyond a preset period to cold storage, and adopting efficient compression algorithm / encoding format conversion for high-frequency access data.

[0015] The application provides a data backup management method and system for a driving recorder. The application calculates the use frequency by quantization standard, accurately divides high-frequency / middle-frequency / low-frequency scenes, and matches differentiated backup strategies, thereby avoiding backup lag of high-frequency scene data, reducing redundancy of low-frequency scene, improving backup efficiency, establishing a three-level storage system, separately classifying core event data, shortening search time, setting three levels of permissions of administrator, ordinary user and temporary authorized user to effectively prevent unauthorized access, improving core data privacy security protection capability, dynamically optimizing according to data access frequency and importance, transferring low-frequency non-important data to cold storage, and using H.265 coding for high-frequency data to improve local storage utilization. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The step method diagram of the application; Figure 2 The system block of the application. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the application.

[0018] First embodiment Please refer to Figure 1 The application provides a data backup management method for a driving recorder, which specifically includes the following steps: Step one, automatically counting the effective vehicle use times in a time period T by a driving recorder supporting APP, and the effective vehicle use times represent the number of times from vehicle ignition start to engine stop, and the driving time is greater than or equal to 5 minutes, and calculating the use frequency according to the formula use frequency = effective vehicle use times / time period, and retaining one decimal place in calculation, and matching the use frequency with the comparison threshold value preset by the operator, and the comparison threshold value is an interval value composed of P min and P max If the use frequency is greater than P max , it is classified as high-frequency use, if the use frequency is within the comparison threshold interval, it is classified as middle-frequency use, and if the use frequency is less than P min , it is classified as low-frequency use. The backup processing is based on the usage of different frequencies, and the backup processing includes backup frequency and backup mode. For high-frequency usage, a fixed backup once a week + an immediate backup in an emergency scenario is adopted, the emergency scenario includes scratching and accident, and high-frequency backup processing information is generated. For medium-frequency usage, a backup within 12 hours after each trip ends + a comprehensive arrangement once a month is adopted, and medium-frequency backup processing information is generated. For low-frequency usage, an immediate backup after a single trip ends + a final backup before parking is adopted, and low-frequency backup processing information is generated.

[0019] Step two, based on the generated backup processing information, storage management is carried out, the backed-up driving record data is obtained, and the driving record data is stored in the corresponding level of storage space according to the year-month-event type hierarchical storage, wherein the year is the highest level, which is used to distinguish the driving records of different years; the month is the intermediate level, which further subdivides the driving records of different months in the same year; the event type is the lowest level, which classifies and stores the driving records in each month according to specific event types such as normal driving, scratching and accident; at the same time, the corresponding access permission is set for each storage level to ensure the safety and privacy of the data, and only authorized users can access the data of a specific level; Then the data in the storage space is obtained, and the data is processed for duplicate checking. The specific duplicate checking method is to obtain the corresponding data timestamp. If the corresponding data timestamp is completely the same, it means that the data is duplicated. Otherwise, if the data timestamp partially overlaps, it means that the data is partially duplicated. Then, the time length, file size and resolution are verified and analyzed. If the time length, file size and resolution are completely consistent, it means that the data is duplicated. Otherwise, if any of them is different, it means that the data is differentially duplicated. Then, the corresponding data content is identified. If the data content is completely duplicated, it means that the data is duplicated. In this way, completely duplicated, partially duplicated and differentially duplicated data are identified, and classified according to the type of duplication to obtain a duplicate classification set. The classification here specifically means that all data with complete duplication are classified together, and data with partial duplication are classified together, and the classified data includes the duplicated data and the corresponding original data.

[0020] Step three, processing the obtained duplicate classification set. For the completely duplicated data category, the association between the completely duplicated data and the abnormal event is judged. If there is an association between them, a user secondary confirmation popup window is triggered. If there is no association between them, it is directly moved to the duplicate data recycle bin, and a default buffer period of t is retained, and the value of t is set by the operator. After the recycle bin is transferred, the indexes of all storage nodes are updated synchronously, and an index verification report is generated after synchronization is completed, showing the index update state of each node. For the data category of fragment repetition, the original file is marked in red, the repeated fragment is marked in blue, the overlapping area is highlighted in yellow, a timeline view is generated, and 3-5 key frames of the overlapping area are automatically extracted. According to the data integrity and keyness, the corresponding retention suggestion is generated. If the fragment is the core frame of the accident, the original file + repeated fragment is recommended to be retained, and the associated ID is marked. If the fragment is a routine commuting ordinary fragment, only the original file is recommended to be retained, and the repeated fragment is deleted. The association relationship between the original file and the repeated fragment is stored in the local database + cloud backup, and the associated ID, original file path, fragment path, overlapping information and user retention selection are recorded. For the data category of difference repetition, the video key frame is converted to 8x8 grayscale image by Python script, dHash value is generated, similarity is calculated by Hamming distance, and for the video with dHash similarity ≥70%, 1 second 1 frame is sampled. Use OpenCV to extract the unique features corresponding to the road marking and time watermark of each frame, and count the proportion of overlapping frames. If the overlapping frames ≥75% and the time watermark is completely consistent, further confirm the content repetition, then the difference repetition data is regarded as repeated data, and the merged processing is carried out, and one complete data is retained, and the merging log is recorded, including merging time, merging reason, original data path and retained data path. If the overlapping frames are less than 75% or the time watermark is inconsistent, it is determined as non-repeated data, and two data are retained, and the difference is added to the data description, which is convenient for subsequent retrieval and utilization.

[0021] Step four, after completing the repeated data processing, the storage space is optimized and managed, the remaining space of each storage node is checked regularly, and the remaining space is compared with the preset threshold value. When the remaining space is lower than the preset threshold value, the space release mechanism is automatically triggered, and the space release analysis signal is generated. According to the data access frequency, the data is divided into high-frequency access data, medium-frequency access data and low-frequency access data, and the specific quantitative standard is that the access is greater than or equal to 3 times in the last 30 days, which is classified as high-frequency access data, the access is 1-2 times in the last 30 days, which is classified as medium-frequency access data, the access is 0 times in the last 30 days, and the access is less than or equal to 1 time in the last 90 days, which is classified as low-frequency access data. For low-frequency access data, if its storage time exceeds the preset period, and the value of the preset period is set by the operator, and it is not marked as important data, and the quantitative standard of important data is that the core important data includes accident, scratch, violation related data or user manually marked important data, the general important data includes special scene record, the non-important data includes daily commuting record without special scene, the redundant data backed up to 3 or more nodes, then it is transferred to the cold storage device or the low-price storage layer of the cloud storage to release the local storage space. At the same time, for high-frequency access data, check whether there is a more efficient compression algorithm or encoding format, if there is, then perform data conversion, and the specific video data preferentially uses H.265 / HEVC encoding instead of H.264 / AVC, the old device adapts H.264 / AVC encoding, and the log / text data uses 7-Zip (LZMA2 algorithm) or ZIP64 compression.

[0022] Step five, establish a data backup and recovery mechanism to ensure that data can be quickly recovered when data is lost or damaged, and periodically check the integrity of all stored driving record data, and use checksum or hash value comparison to ensure data integrity and consistency. If the data is found to be damaged, first try to recover the data from the local backup or cloud backup, if the local backup and cloud backup are not available, then according to the importance of the data, start the data reconstruction process, for critical data such as accident records, through the log file of the driving recorder, vehicle sensor data and associated external camera data, data reconstruction is performed to restore the original data content as much as possible, at the same time, record the process of data recovery and reconstruction, including recovery time, recovery data volume, recovery success rate, etc. for subsequent analysis and improvement.

[0023] Second embodiment Please refer to Figure 2 The application provides a data backup management system for a driving recorder, which comprises a scene classification calculation module, a data backup processing module, a hierarchical storage module, a duplicate data processing module and a space optimization module, and combines Figure 1 It can be known that the information between the above function modules is one-way transmission; The scene classification calculation module is configured to automatically count the number of effective vehicle uses in a time period through the dashboard APP, and calculate the use frequency according to the formula use frequency = effective vehicle use number / time period, match the use frequency with a preset comparison threshold, divide the use frequency into high-frequency use, medium-frequency use and low-frequency use, and transmit the classified use scene to the data backup processing module; The data backup processing module is configured to perform corresponding backup processing according to different use scenes, adopt weekly fixed backup + emergency scene real-time backup for high-frequency use, adopt backup within 12 hours after use + monthly overall arrangement for medium-frequency use, and adopt immediate backup after single trip and final backup before parking for low-frequency use, generate backup processing information corresponding to the scene, and transmit the backup processing information to the hierarchical storage module. The hierarchical storage module is configured to obtain the dashboard data based on the backup processing information, store the data in a three-level hierarchy of year-month-event type, set access permissions for each level, and identify complete repeated data, fragment repeated data and differential repeated data through a hierarchical process of time stamp comparison, time length / file size / resolution verification and data content identification, form a repeated classification set, and transmit the repeated classification set to the repeated data processing module. The repeated data processing module is configured to process the repeated classification set by type, perform secondary confirmation or recycle bin transfer for complete repeated data according to the relevance to abnormal events, generate a timeline view and key frame comparison information for fragment repeated data, generate a retention suggestion and persistent association, and determine whether the differential repeated data is repeated through dHash algorithm and OpenCV frame feature extraction, perform merging or retention processing, generate repeated data processing information, and transmit the repeated data processing information to the space optimization module. The space optimization module is configured to periodically check the remaining space of the storage node, trigger space release when the remaining space is lower than a preset threshold, transfer non-important data that is older than a preset period to cold storage, and use efficient compression algorithms / encoding format conversion for high-frequency access data.

[0024] Some data in the above formula are dimensionless numerical calculations, and the contents not described in detail in the specification are all prior art known to those skilled in the art.

[0025] The above embodiments are only used to illustrate the technical method of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical method of the present application.

Claims

1. A data backup management method for dashcams, characterized in that, The method specifically includes the following steps: Step 1: Count the number of valid vehicle uses within a preset period using the dashcam and calculate the usage frequency. Compare the frequency with a preset comparison threshold to classify the usage into three categories: high-frequency, medium-frequency, and low-frequency. Then, perform corresponding backup processing and generate backup processing information for each scenario. Step 2: Based on the backup processing information, obtain the backed-up driving record data, perform hierarchical storage, obtain the driving record data in each storage level, and identify completely duplicate data, fragment duplicate data, and differential duplicate data through multi-level deduplication methods, thereby forming a duplicate classification set; Step 3: Process the different duplicate types in the duplicate classification set separately. For completely duplicate data, determine its correlation with abnormal events, and perform secondary confirmation or recycle bin transfer based on the correlation results, while updating the storage node index synchronously. For fragmented duplicate data, generate a timeline view and keyframe comparison information, and generate retention suggestions based on data integrity and criticality. For differential duplicate data, determine whether it is duplicate data based on feature overlap and time watermark consistency, and perform merging or retention processing. Step 4: Regularly check the remaining space of each storage node. When the remaining space is lower than the preset threshold, trigger the space release mechanism. According to the access frequency, divide the data into high-frequency access data, medium-frequency access data, and low-frequency access data. For low-frequency access data, transfer it to cold storage devices or low-cost cloud storage tiers; for high-frequency access data, check and use more efficient compression algorithms to transform the data. Step 5: Regularly perform integrity checks on driving record data using checksum or hash value comparison. If data corruption is found, prioritize recovery from local or cloud backups. If backups are unavailable, initiate a data reconstruction process for critical data, record data recovery and reconstruction process information, and form a process log.

2. The data backup management method for a dashcam according to claim 1, characterized in that, The method for classifying usage scenarios into three categories—high-frequency, mid-frequency, and low-frequency—is as follows: If the frequency of use is greater than P max If the usage frequency is within the comparison threshold range, it is classified as high-frequency use; if the usage frequency is less than P, it is classified as medium-frequency use; if the usage frequency is less than P, it is classified as medium-frequency use. min If so, it is classified as low-frequency use.

3. The data backup management method for a dashcam according to claim 1, characterized in that, The method for generating backup processing information for the corresponding scenario is as follows: For high-frequency usage scenarios, a fixed backup is performed once a week, plus real-time backups for emergency scenarios. Emergency scenarios include scratches and accidents, generating high-frequency backup processing information. For mid-frequency usage scenarios, backups are performed within twelve hours after each vehicle use, plus a comprehensive monthly cleanup, generating mid-frequency backup processing information. For low-frequency usage scenarios, a backup is performed immediately after each trip, plus a final backup before parking, to generate low-frequency backup processing information.

4. The data backup management method for a dashcam according to claim 1, characterized in that, Multi-level plagiarism detection methods are as follows: At the first level, if two data sets have exactly the same timestamp, they are considered completely duplicate data; if the timestamps partially overlap, they proceed to the second level of verification. The second level is determined to be duplicate data if the duration, file size, and resolution of the two data are completely identical; if any parameter is different, the third level of identification is performed. The third level involves comparing image features or analyzing video content to determine whether the data content is completely duplicated, thereby distinguishing between fragmented duplicate data and differentially duplicated data.

5. The data backup management method for a dashcam according to claim 1, characterized in that, The details of handling completely duplicate data include: If completely duplicated data is associated with an abnormal event, a secondary confirmation pop-up will be triggered, and subsequent operations will be performed after the user confirms. If completely duplicated data is not related to an abnormal event, it is directly transferred to the duplicate data recycle bin. The default retention period t of the recycle bin is set by the operator. After the recycle bin is moved, the indexes of the local core node, local extended node, and cloud node are updated synchronously. After the synchronization is completed, an index verification report is generated.

6. The data backup management method for a dashcam according to claim 1, characterized in that, Details of handling duplicate fragment data include: In the timeline view, the original file is marked in red, repeated segments are marked in blue, and overlapping areas are highlighted in yellow. 3-5 keyframes of the overlapping area are automatically extracted, including the core frame of the accident or the road sign frame. If the fragment is a core frame of an accident, it is recommended to keep the original file and duplicate fragments and mark them with the associated ID; if the fragment is a normal fragment of daily commuting, it is recommended to keep only the original file.

7. The data backup management method for a dashcam according to claim 1, characterized in that, Details of handling duplicate data include: The video keyframes were converted into 8×8 grayscale images using a Python script, generating dHash values. Similarity was calculated using Hamming distance. A Hamming distance ≤ 5 corresponds to a similarity ≥ 80%. For videos with a dHash similarity ≥ 70%, samples were taken at 1 frame per second, and OpenCV was used to extract road markings and time watermark features from each frame. If the percentage of overlapping frames is ≥75% and the time watermarks are completely identical, it is determined to be duplicate data, merged and one complete copy is retained, and a merge log is recorded; if the percentage of overlapping frames is <75% or the time watermarks are inconsistent, it is determined to be non-duplicate data, two copies are retained and a difference description is added.

8. The data backup management method for a dashcam according to claim 1, characterized in that, The quantitative standard for data access frequency is: High-frequency access data, with ≥3 valid accesses in the past 30 days; Mid-frequency access data, with 1-2 valid accesses in the last 30 days; Low-frequency access data, with 0 valid accesses in the past 30 days and ≤1 valid access in the past 90 days.

9. The data backup management method for a dashcam according to claim 1, characterized in that, The quantitative standards for important data are: Core and important data, including accident, scratch, and traffic violation data, or data that users have manually marked as important; Important general data, and records for special scenarios; Non-critical data, daily commuting records without special scenarios, and redundant data that has been backed up to 3 or more nodes.

10. A data backup management system for a dashcam, used to execute the backup management method according to any one of claims 1-9, characterized in that, include: The scene classification calculation module is used to automatically count the number of valid vehicle uses within a time period through the dashcam's accompanying APP, and calculate the usage frequency according to the formula usage frequency = number of valid vehicle uses / time period. The usage frequency is matched with a preset comparison threshold and divided into high-frequency, medium-frequency, and low-frequency usage scenarios. At the same time, the classified usage scenarios are transmitted to the data backup processing module. The data backup processing module is used to perform corresponding backup processing according to different usage scenarios. High-frequency usage scenarios use weekly fixed backups + emergency scenario instant backups, medium-frequency usage scenarios use backups within twelve hours after the end of the trip + monthly comprehensive cleanup, and low-frequency usage scenarios use immediate backups after a single trip + final backups before parking. It generates backup processing information for the corresponding scenario and transmits it to the hierarchical storage module. The hierarchical storage module is used to obtain driving record data based on backup processing information, store the data in a three-level hierarchy of year-month-event type, and set access permissions for each level. It is also used to identify completely duplicate data, fragmented duplicate data, and differential duplicate data through a hierarchical process of timestamp comparison, duration / file size / resolution verification, and data content recognition, forming a duplicate classification set, and transmitting it to the duplicate data processing module. The duplicate data processing module is used to classify and process duplicate data sets. Completely duplicate data undergoes secondary confirmation or recycle bin transfer based on its correlation with abnormal events. Fragment duplicate data generates a timeline view and keyframe comparison information, generates retention suggestions, and persists the correlation. Differential duplicate data is determined to be duplicated or not through the dHash algorithm and OpenCV frame feature extraction, and merges or retains it. At the same time, duplicate data processing information is generated and transmitted to the space optimization module. The space optimization module is used to periodically check the remaining space of storage nodes and trigger space release when it falls below a preset threshold: it transfers low-frequency access data that has exceeded a preset period and is not important to cold storage, and uses efficient compression algorithms / encoding format conversion for high-frequency access data.

Images