An intelligent management system and method for a video fusion platform based on big data

Abstract

The present invention discloses an intelligent management system and method for a video fusion platform based on big data, which relates to the technical field of computer information processing. The video records in the video record storage repository are managed in chronological order, the historical records of the staff in video content traceability are obtained, the feature tags of each video record in the historical records are respectively extracted, the similarity of the image frames between the video records is calculated, several traceability records are collected, the corresponding relationship of the feature tags between the video records in the traceability records is used as the first matching condition, the similarity of the image frames in the video records is used as the second matching condition, and the video record to be deleted in a certain video record storage repository is used as the target record. The video records are matched from all video storage repositories through the feature tag corresponding to the first matching condition, the video records are matched from all video storage repositories through the second matching condition, and the control records are obtained by removing the target record from the video records.

Description

Technical Field

[0001] The present invention relates to the technical field of computer information processing, and in particular to an intelligent management system and method for a video fusion platform based on big data. Background Art

[0002] With the wide application of video terminals, a large amount of video recording data will be generated synchronously. In order to make video recordings play the roles of picture recording and event backtracking, it is necessary to deploy a data storage module with sufficient capacity. During the storage process of video recordings, it is necessary to clean the video recordings in the data storage module to free up storage space. In the operation of deleting historical records, there may be a situation where historical video records that may be used are accidentally deleted, resulting in the discovery of missing parts of historical records when querying historical records. In the current technical solutions, it is usually adopted to temporarily transfer the locally deleted data to another data storage module, such as a file recycle bin or cloud storage, and return it to the local when needed. Such a practice essentially still requires increasing the storage space to increase the capacity of video data, and cannot improve the accommodation efficiency of the existing storage space. Summary of the Invention

[0003] The purpose of the present invention is to provide an intelligent management system and method for a video fusion platform based on big data to solve the problems raised in the prior art.

[0004] To achieve the above purpose, the present invention provides the following technical solution: An intelligent management method for a video fusion platform based on big data, the method includes:

[0005] Step S100: Connect a number of video acquisition terminals to the video management platform. Each terminal is associated with a video record storage repository, manage the video records in the video record storage repository in chronological order, obtain the historical records of the staff in video content traceability, and record the video records with an associated relationship as video record pairs;

[0006] Step S200: For any one of the video record pairs, respectively extract the feature tags of each video record in the historical record, calculate the similarity of the image frames between the video records, and collect the feature tags and the similarity of the image frames to obtain a traceability record corresponding to a traceability process;

[0007] Step S300: Collect a number of traceability records, use the corresponding relationship of the feature tags between the video records in the traceability record as the first matching condition, and use the similarity of the image frames in the video records as the second matching condition;

[0008] Step S400: Take the video records to be deleted in a certain video record repository as target records, obtain the feature tags of the target records, match the video records from all video storage repositories according to the feature tags corresponding to the first matching condition, record the video records as associated records, and match the video records from all video storage repositories according to the second matching condition. The video records obtained by removing the target records are used as control records;

[0009] Step S500: Extract the image features from the target records, match the image features in the control records, record the image features that cannot be matched in the control records as target image features, obtain the video segments in the target records that include the target image features, and collect and send them to the relevant management personnel.

[0010] Further, step S100 includes:

[0011] Step S101: Manage a number of video acquisition terminals through a certain video management platform. Among them, each video acquisition terminal corresponds to a video record storage repository;

[0012] Step S102: Set a unit period, extract the time axes of the videos in all video record storage repositories, divide the time axes according to the unit period, and align the time axes of all videos in chronological order;

[0013] Step S103: Take the video records obtained by the staff in video record storage repository A1 in a certain unit period as the first record KA1, obtain the matching video records of the personnel for the first record KA1. When the unit period corresponding to the matching record is earlier than the first record KA1, take the matching record as the second record KA2, and record the video record storage repository corresponding to the second record KA2 as video record storage repository A2;

[0014] Step S104: Combine KA1 and KA2 into a traceability pair, denoted as d(KA1, KA2), and obtain the number of unit periods between KA1 and KA2 in the traceability record, denoted as NT.

[0015] Further, step S200 includes:

[0016] Step S201: Analyze the video content of the first record KA1 and the second record KA2 through a video analysis tool, output a text-based video content analysis, extract the feature tags respectively, record the feature tags corresponding to the first record KA1 into the first label combination U, and record the feature tags corresponding to the second record KA2 into the second label set U2;

[0017] Step S202: Obtain n1 image frames from the first record KA1, obtain n2 image frames from the second record KA2, and calculate the image similarity between the i1-th image frame in KA1 and the i2-th image frame in the second record KA2, denoted as σ i1，i2 ;

[0018] Step S203: Calculate the similarity reference value AS of the first record KA1 and the second record KA2

[0019]

[0020] Step S204: Pool the traceability pairs, the first label combination, the second label combination, the number of unit cycles, and the similarity reference value to obtain a traceability record denoted as Trace, Trace: (KA1, KA2, U1, U2, NT, AS).

[0021] Further, step S300 includes:

[0022] Step S301: Collect j traceability records, pool all the first label sets to obtain the first target set R1, R1 = U11 ∩ U12 ∩ U13 ∩ …… ∩ U1j, pool all the second label sets to obtain the second target set R2, R2 = U21 ∩ U22 ∩ U23 ∩ …… ∩ U2j, where U11, U12, U13, …… and U1j respectively represent the first label sets corresponding to the 1st, 2nd, 3rd, …… and j-th traceability records, and U21, U22, U23, …… and U2j respectively represent the second label sets corresponding to the 1st, 2nd, 3rd, …… and j-th traceability records, and form the first matching condition with R1 and R2;

[0023] By parsing the video content, obtain the feature labels of the video record, obtain the feature labels of the two videos in an association pair, and form an association relationship pair with the feature labels as a matching condition for a type of video record. When a certain type of label of the video record is known, obtain the corresponding label to achieve the matching relationship of the video record;

[0024] The first matching condition is related to the first target set, and different first target sets correspond to different first matching conditions;

[0025] Step S302: Obtain the average value of the number of unit cycles of j traceability records, denoted as JNT, obtain the average value of the similarity reference values of j traceability records, denoted as JAS, and form the second matching condition with JNT and JAS;

[0026] By analyzing the content of the video recording, according to the similarity of the images in the image frames and the query time range, obtain the time correlation range and the image correlation features of the relevant videos, and form a second matching condition, which is applied globally to the video management system.

[0027] Further, step S400 includes:

[0028] Step S401: Take the video record to be deleted in a certain video record repository as the target record L1. Through the video analysis tool, obtain the feature tags of the target record L1, gather the feature tags to obtain the target tag set G1, and obtain the first matching condition (Q1, Q2) corresponding to the target tag set G1, where Q1 satisfies the condition

[0029] Step S402: Match the associated record L2 of the target record L1 according to the first matching condition, and obtain the tag set G2 of the associated record, satisfying the condition

[0030] Step S403: Obtain the second matching condition (JNT, JAS) of the associated record L2. In all video record repositories, match the video records with an image similarity of JAS to the associated record L2 within the time range corresponding to the first JNT unit periods before L2, and gather the video records to obtain the reference record L3;

[0031] Step S404: Take out the target record L1 from the reference record L3 to obtain the comparison record L4 of the target record L1.

[0032] Further, step S500 includes:

[0033] Step S501: Extract the image features in the target record L1 to form an image feature library, and establish a feature template for each image feature.

[0034] Step S502: Match the images corresponding to the feature templates in the comparison record L4, gather the feature templates that cannot be matched in the comparison record L4 into the target feature templates, and gather the image features corresponding to the target feature templates to obtain the target image features.

[0035] First, according to the feature tags of the target record, obtain the video records that may trace back to the target record. This video record is the associated record. Then analyze the associated record to obtain the possible traceability range of the associated record, denoted as the reference record. Remove the target record from the reference record to obtain the comparison record;

[0036] Compare the target record with the comparison record, analyze whether the comparison record includes all the features in the target record, extract the image features in the target record that are not in the comparison record, and use these image features as the target image features.

[0037] Extract the image information including the target features from the target record for relevant staff to identify or retain;

[0038] Step S503: Obtain all the images including the target image features in the target record L1, gather them into the target image set, and send them to the relevant management personnel of the video management platform.

[0039] To better implement the above method, an intelligent management system for a video fusion platform based on big data is also proposed. The system includes:

[0040] A matching management module, a traceability record management module, a matching condition management module, a matching retrieval module, and a video extraction module. Among them, the matching management module is used to obtain the historical records of the staff's matching videos and obtain the video record pairs composed of video records. The traceability record management module is used to obtain the feature tags and image similarities in the video record pairs and manage the traceability records. The matching condition management module is used to gather the traceability records and manage the first matching condition and the second matching condition. The matching retrieval module is used to obtain the target record and obtain the control record according to the matching conditions. The video extraction module is used to compare the target record with the control record to obtain the target image set;

[0041] Furthermore, the matching management module includes: a terminal management unit, a time management unit, and a traceability collection unit. Among them, the terminal management unit is used to manage the video acquisition terminals. The time management unit is used to collect the time information of the video records and manage the time information of the video records through a unit cycle. The traceability collection unit is used to collect the historical records associated with the videos;

[0042] Furthermore, the traceability record management module includes: a label management unit, an image comparison unit, and a traceability record management unit. Among them, the label management unit is used to manage the feature tags of the video records. The image comparison unit is used to calculate the image similarities between the video records. The traceability record management unit is used to gather the video record features and manage the traceability records;

[0043] Furthermore, the matching condition management module includes: a first matching condition management unit and a second matching condition management unit. Among them, the first matching condition management unit is used to gather the corresponding relationships of the feature tags and manage the first matching condition. The second matching condition management unit is used to gather the image similarity features and manage the second matching condition;

[0044] Further, the matching and retrieval module includes a target record management unit, a first matching unit, a second matching unit, and a control record management unit. Among them, the target record management unit is used to store the video records to be deleted in a certain video record repository. The first matching unit is used to obtain associated records through the target records and the first matching condition. The second matching unit is used to obtain reference records through the associated set and the second matching condition. The control record management unit is used to remove the target records from the reference records to obtain control records;

[0045] Further, the video extraction module includes an image feature management unit, a feature screening unit, and an image aggregation unit. Among them, the image feature management unit is used to obtain the image features in the target records. The feature screening unit is used to compare and screen the image features in the control records to obtain the target image features. The image aggregation unit is used to aggregate the image information including the target image features in the target records to obtain the target image set.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows: By analyzing the video records, performing two rounds of matching and screening, obtaining the control video records of the videos to be deleted, further comparing the videos to be deleted with the control videos, and obtaining the unique image features in the videos to be deleted, so as to extract the unique parts of the videos to be deleted. After deleting these unique parts, it is difficult to obtain them from other databases, so special attention needs to be paid. For the parts that are not unique, even if they are deleted, relevant data can be found in other databases for supplementation, which has little impact on relevant staff to obtain historical records. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a schematic structural diagram of an intelligent management system for a video fusion platform based on big data according to the present invention;

[0048] Figure 2 It is a schematic flowchart of an intelligent management method for a video fusion platform based on big data according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0050] Embodiment: As Figure 1 and Figure 2 shown, the present invention provides a technical solution, an intelligent management method for a video fusion platform based on big data:

[0051] Step S100: Connect several video capture terminals to the video management platform. Each terminal is associated with a video record storage repository. Manage the video records in the video record storage repository in chronological order, obtain the historical records of the staff in video content traceability, and mark the video records with an associated relationship as a video record pair;

[0052] Among them, step S100 includes:

[0053] Step S101: Manage several video capture terminals through a certain video management platform. Among them, each video capture terminal corresponds to a video record storage repository;

[0054] In the embodiment, the video record storage repository includes a virtual storage repository and a physical storage repository. The physical storage repository is an actual storage unit such as a USB storage device, a hard disk, and a file server. The virtual storage repository is, for example, a virtualized storage unit formed by dividing or combining one or several physical storage repositories through a management device;

[0055] Step S102: Set a unit period, extract the time axes of the videos in all video record storage repositories, divide the time axes according to the unit period, and align the time axes of all videos in chronological order;

[0056] Step S103: Mark the video records obtained by the staff in video record storage repository A1 in a certain unit period as the first record KA1, obtain the matching video records of the personnel for the first record KA1. When the unit period corresponding to the matching record is earlier than the first record KA1, mark the matching record as the second record KA2, and mark the video record storage repository corresponding to the second record KA2 as video record storage repository A2;

[0057] Step S104: Combine KA1 and KA2 into a traceability pair, denoted as d(KA1, KA2), and obtain the number of unit periods between KA1 and KA2 in the traceability record, denoted as NT.

[0058] Step S200: For any one video record pair, respectively extract the feature tags of each video record in the historical record, calculate the similarity of the image frames between the video records, and collect the feature tags and the similarity of the image frames to obtain a traceability record corresponding to a traceability process;

[0059] Among them, step S200 includes:

[0060] Step S201: Through a video analysis tool, analyze the video content of the first record KA1 and the second record KA2, output a text-based video content analysis, respectively extract the feature tags, record the feature tags corresponding to the first record KA1 into the first label combination U1, and record the feature tags corresponding to the second record KA2 into the second label set U2;

[0061] Step S202: Obtain n1 image frames from the first record KA1, obtain n2 image frames from the second record KA2, and calculate the image similarity between the i1-th image frame in KA1 and the i2-th image frame in the second record KA2, denoted as σ i1，i2 ;

[0062] Step S203: Calculate the similarity reference value AS of the first record KA1 and the second record KA2

[0063]

[0064] Step S204: Aggregate the traceability pair, the first label combination, the second label combination, the number of unit cycles, and the similarity reference value to obtain a traceability record denoted as Trace, Trace: (KA1, KA2, U1, U2, NT, AS).

[0065] Step S300: Aggregate several traceability records, use the corresponding relationship of the feature labels between the video records in the traceability records as the first matching condition, and use the similarity of the image frames in the video records as the second matching condition;

[0066] Among them, Step S300 includes:

[0067] Step S301: Collect j traceability records, aggregate all the first label sets to obtain the first target set R1, R1 = U11 ∩ U12 ∩ U13 ∩... ∩ U1j, aggregate all the second label sets to obtain the second target set R2, R2 = U21 ∩ U22 ∩ U23 ∩... ∩ U2j, where U11, U12, U13,... and U1j respectively represent the first label sets corresponding to the 1st, 2nd, 3rd,... and j-th traceability records, and U21, U22, U23,... and U2j respectively represent the second label sets corresponding to the 1st, 2nd, 3rd,... and j-th traceability records, and form the first matching condition with R1 and R2;

[0068] Step S302: Obtain the average value of the number of unit cycles of j traceability records, denoted as JNT, obtain the average value of the similarity reference values of j traceability records, denoted as JAS, and form the second matching condition with JNT and JAS.

[0069] Step S400: Take the video record to be deleted in a certain video record repository as the target record, obtain the feature label of the target record, match the video record from all video repositories through the feature label corresponding to the first matching condition, denote the video record as the associated record, and match the video record from all video repositories through the second matching condition, and remove the target record from the video record to obtain the control record;

[0070] Among them, step S400 includes:

[0071] Step S401: Take the video record to be deleted in a certain video record repository as the target record L1. Through a video analysis tool, obtain the feature tags of the target record L1, gather the feature tags to obtain the target tag set G1, and obtain the first matching conditions (Q1, Q2) corresponding to the target tag set G1, where Q1 satisfies the condition

[0072] Step S402: Match the associated record L2 of the target record L1 according to the first matching condition, obtain the tag set G2 of the associated record, and satisfy the condition

[0073] Step S403: Obtain the second matching conditions (JNT, JAS) of the associated record L2. In all video record repositories, match the video records with an image similarity of JAS to the associated record L2 within the time range corresponding to the first JNT unit periods of L2, and gather the video records to obtain the reference record L3;

[0074] Step S404: Take out the target record L1 from the reference record L3 to obtain the control record L4 of the target record L1.

[0075] Step S500: Extract the image features from the target record, match the image features in the control record, record the image features that cannot be matched in the control record as the target image features, obtain the video segments including the target image features in the target record, and gather and send them to the relevant management personnel;

[0076] Among them, step S500 includes:

[0077] Step S501: Extract the image features in the target record L1 to form an image feature library, and establish a feature template corresponding to each image feature;

[0078] In the embodiment, a slice of a feature image is collected and denoted as H, and the image feature obtained through the backbone neural network is denoted as η(H), and η(H) is used as a feature template;

[0079] Collect the image Q in the control record, and the image feature obtained through the backbone neural network is denoted as η(Q);

[0080] Compare η(H) with η(Q). When η(H) cannot be matched in the control record, it is recorded as a target image feature;

[0081] Step S502: Match the images corresponding to the feature templates in the control record L4, collect the feature templates that cannot be matched in the control record L4 into the target feature templates, and collect the image features corresponding to the target feature templates to obtain the target image features;

[0082] Step S503: Obtain all the images in the target record L1 that include the target image features, collect them into the target image set, and send them to the relevant management personnel of the video management platform.

[0083] An intelligent management system for a video fusion platform based on big data, the system includes: a matching management module, a traceability record management module, a matching condition management module, a matching retrieval module, and a video extraction module;

[0084] Among them, the matching management module is used to obtain the historical records of the staff matching videos, and obtain the video record pairs composed of video records. Among them, the matching management module includes: a terminal management unit, a time management unit, and a traceability collection unit. Among them, the terminal management unit is used to manage video capture terminals, the time management unit is used to collect the time information of video records, manage the time information of video records through a unit cycle, and the traceability collection unit is used to collect the historical records associated with the video;

[0085] Among them, the traceability record management module is used to obtain the feature tags and image similarities in the video record pairs, and manage the traceability records. The traceability record management module includes: a tag management unit, an image comparison unit, and a traceability record management unit. Among them, the tag management unit is used to manage the feature tags of video records, the image comparison unit is used to calculate the image similarities between video records, and the traceability record management unit is used to collect video record features and manage the traceability records;

[0086] Among them, the matching condition management module is used to collect traceability records and manage the first matching condition and the second matching condition. Among them, the matching condition management module includes: a first matching condition management unit and a second matching condition management unit. Among them, the first matching condition management unit is used to collect the corresponding relationships of feature tags and manage the first matching condition, and the second matching condition management unit is used to collect image similarity features and manage the second matching condition;

[0087] Among them, the matching retrieval module is used to obtain the target record, obtain the control record according to the matching conditions. The matching retrieval module includes a target record management unit, a first matching unit, a second matching unit, and a control record management unit. Among them, the target record management unit is used to store the video records to be deleted in a certain video record repository. The first matching unit is used to obtain the associated records through the target record and the first matching condition. The second matching unit is used to obtain the reference records through the associated set and the second matching condition. The control record management unit is used to remove the target record from the reference records to obtain the control record;

[0088] Among them, the video extraction module is used to compare the target record with the control record to obtain a target image set. Among them, the video extraction module includes: an image feature management unit, a feature screening unit, and an image aggregation unit. Among them, the image feature management unit is used to obtain the image features in the target record, the feature screening unit is used to compare and screen the image features in the control record to obtain target image features, and the image aggregation unit is used to aggregate the image information including the target image features in the target record to obtain a target image set.

[0089] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

Claims

1. An intelligent management method for a video fusion platform based on big data, characterized in that: The method includes the following steps: Step S100: Connect several video acquisition terminals to a video management platform. Each terminal is associated with a video record storage repository. Manage the video records in the video record storage repository in chronological order, obtain the historical records of the staff in video content traceability, and record the video records with an associated relationship as a video record pair; Step S200: For any video record pair, respectively extract the feature tags of each video record in the historical record, calculate the similarity of the image frames between the video records, gather the feature tags and the similarity of the image frames, and obtain a traceability record corresponding to a traceability process; Step S300: Gather several traceability records, use the corresponding relationship of the feature tags between the video records in the traceability record as the first matching condition, and use the similarity of the image frames in the video record as the second matching condition; Step S400: Use the video record to be deleted in a certain video record storage repository as the target record, obtain the feature tags of the target record, based on the first matching condition corresponding to the feature tags, match video records from all video storage repositories, record the video records as associated records, and through the second matching condition, match video records from all video storage repositories, and record the video records after removing the target record as the comparison record; Step S500: Extract the image features from the target record, match the image features in the comparison record, record the image features that cannot be matched in the comparison record as the target image features, obtain the video segments including the target image features in the target record, and gather and send them to relevant management personnel.

2. The intelligent management method of a video fusion platform based on big data according to claim 1, characterized in that: Step S100 includes: Step S101: Manage several video acquisition terminals through a certain video management platform, where each video acquisition terminal corresponds to a video record storage repository; Step S102: Set a unit period, extract the time axes of the videos in all video record storage repositories, divide the time axes according to the unit period, and align the time axes of all videos in chronological order; Step S103: Record the video record of a certain unit period obtained by the staff in the video record storage repository A1 as the first record KA1, obtain the matching record of the person for the first record KA1. When the unit period corresponding to the matching record is earlier than the first record KA1, use the matching record as the second record KA2, and record the video record storage repository corresponding to the second record KA2 as the video record storage repository A2; Step S104: Combine KA1 and KA2 into a traceability pair, denoted as d(KA1, KA2), and obtain the number of unit periods between KA1 and KA2 in the traceability record, denoted as NT.

3. The intelligent management method of a video fusion platform based on big data according to claim 2, characterized in that: Step S200 includes: Step S201: Through a video analysis tool, analyze the video content of the first record KA1 and the second record KA2, output a text-based video content analysis, respectively extract the feature tags, record the feature tags corresponding to the first record KA1 into the first label set U1, and record the feature tags corresponding to the second record KA2 into the second label set U2; Step S202: Obtain n1 image frames from the first record KA1, obtain n2 image frames from the second record KA2, and calculate the image similarity between the i1-th image frame in KA1 and the i2-th image frame in the second record KA2, denoted as σ i1，i2 ; Step S203: Calculate the similarity reference value AS of the first record KA1 and the second record KA2, ； Step S204: Aggregate the traceability pair, the first tag set, the second tag set, the number of unit periods, and the similarity reference value to obtain a traceability record denoted as Trace, Trace: (KA1, KA2, U1, U2, NT, AS).

4. An intelligent management method for a video fusion platform based on big data according to claim 3, characterized in that: Step S300 includes: Step S301: Collect j traceability records, aggregate all the first tag sets, calculate the common set of all the first tag sets to obtain the first target set R1, R1 = U11 ∩ U12 ∩ U13 ∩ …… ∩ U1j, aggregate all the second tag sets, calculate the common set of all the second tag sets to obtain the first target set R2, R2 = U21 ∩ U22 ∩ U23 ∩ …… ∩ U2j, where U11, U12, U13, ……, and U1j respectively represent the first tag sets corresponding to the 1st, 2nd, 3rd, ……, and jth traceability records, and U21, U22, U23, ……, and U2j respectively represent the second tag sets corresponding to the 1st, 2nd, 3rd, ……, and jth traceability records. Combine R1 and R2 to form the first matching condition; Step S302: Obtain the average value of the number of unit periods of the j traceability records, denoted as JNT, and obtain the average value of the similarity reference values of the j traceability records, denoted as JAS. Combine JNT and JAS to form the second matching condition.

5. The intelligent management method of a video fusion platform based on big data according to claim 4, characterized in that: Step S400 includes: Step S401: Take the video record to be deleted in a certain video record repository as the target record L1. Through a video analysis tool, obtain the feature tags of the target record L1, aggregate the feature tags to obtain the target tag set G1, and obtain the first matching condition (Q1, Q2) corresponding to the target tag set G1, where Q1 satisfies the condition Q1 ⊆ G1; Step S402: Match the associated record L2 of the target record L1 according to the first matching condition, and obtain the tag set G2 of the associated record, satisfying the condition Q2 ⊆ G2; Step S403: Obtain the second matching condition (JNT, JAS) of the associated record L2. In all video record repositories, match the video records with an image similarity of JAS within the time range corresponding to the first JNT unit periods of L2. Aggregate the video records to obtain the reference record L3; Step S404: Remove the target record L1 from the reference record L3 to obtain the control record L4 of the target record L1.

6. An intelligent management method for a video fusion platform based on big data according to claim 5, characterized in that: Step S500 includes: Step S501: Extract the image features in the target record L1 to form an image feature library, and establish a feature template for each image feature; Step S502: Match the images corresponding to the feature templates in the control record L4, aggregate the feature templates that cannot be matched in the control record L4 into the target feature templates, and aggregate the image features corresponding to the target feature templates to obtain the target image features; Step S503: Obtain all the images in the target record L1 that include the target image features, aggregate them into the target image set, and send them to the relevant management personnel of the video management platform.

7. An intelligent management system for a video fusion platform based on big data, which is used to execute an intelligent management method for a video fusion platform based on big data according to any one of claims 1-6, and is characterized in that: The system includes: A matching management module, a traceability record management module, a matching condition management module, a matching retrieval module, and a video extraction module. Among them, the matching management module is used to obtain the historical records of staff matching videos, obtain video record pairs composed of video records. The traceability record management module is used to obtain the feature tags and image similarity in the video record pairs and manage the traceability records. The matching condition management module is used to collect traceability records and manage the first matching condition and the second matching condition. The matching retrieval module is used to obtain target records and obtain comparison records according to the matching conditions. The video extraction module is used to compare the target records with the comparison records to obtain a set of target images.

8. An intelligent management system for a video fusion platform based on big data according to claim 7, characterized in that: The matching management module includes: a terminal management unit, a time management unit, and a traceability collection unit. Among them, the terminal management unit is used to manage video capture terminals. The time management unit is used to collect the time information of video records and manage the time information of video records through a unit cycle. The traceability collection unit is used to collect the historical records associated with the video. The traceability record management module includes: a tag management unit, an image comparison unit, and a traceability record management unit. Among them, the tag management unit is used to manage the feature tags of video records. The image comparison unit is used to calculate the image similarity between video records. The traceability record management unit is used to collect video record features and manage the traceability records.

9. The intelligent management system of a video fusion platform based on big data according to claim 7, characterized in that: The matching condition management module includes: a first matching condition management unit and a second matching condition management unit. Among them, the first matching condition management unit is used to collect the corresponding relationships of feature tags and manage the first matching condition. The second matching condition management unit is used to collect image similarity features and manage the second matching condition. The matching retrieval module includes a target record management unit, a first matching unit, a second matching unit, and a comparison record management unit. Among them, the target record management unit is used to manage the video records to be deleted in a certain video record repository. The first matching unit is used to obtain associated records through the target records and the first matching condition. The second matching unit is used to obtain reference records through the associated set and the second matching condition. The comparison record management unit is used to remove the target records from the reference records to obtain comparison records.

10. An intelligent management system for a video fusion platform based on big data according to claim 7, characterized in that: The video extraction module includes: an image feature management unit, a feature screening unit, and an image collection unit. Among them, the image feature management unit is used to obtain the image features in the target records. The feature screening unit is used to compare and screen image features in the comparison records to obtain target image features. The image collection unit is used to collect the image information including the target image features in the target records to obtain a set of target images.

Images