A video matrix batch generation method, system and storage medium

Abstract

The present application relates to the technical field of video clip, especially relates to a kind of video matrix batch generation method, system and storage medium, comprising the following steps: S1, constructs video editing template, S2, original media material pretreatment, S3, constructs multidimensional material matrix, S4, intelligent sequence filling, S5, sequence duplicate detection, S6, closed loop weight optimization, S7, video suppression and release, S8, title automatic generation.The present application can improve video batch generation efficiency and content rationality, effectively control video repetition, improve content diversity, and can form closed loop optimization mechanism, continuously improve video blockbuster potential, realize whole-process automation and traceable management.

Description

Technical Field

[0001] This invention relates to the field of video editing technology, and in particular to a method, system, and storage medium for batch generation of video matrices. Background Technology

[0002] With the rapid development of e-commerce and short video platforms, the market demand for content such as short videos promoting products and local lifestyles has exploded. Merchants and content creators need to quickly edit large amounts of raw footage into high-quality, frequently released videos. Currently, traditional video editing methods rely primarily on manual operation, resulting in low efficiency, long production cycles, and inconsistent editing quality. Furthermore, the high turnover rate and difficulty in recruiting professional video editors further restricts the large-scale production of video content.

[0003] To alleviate these problems, template-based automated editing tools have emerged in the existing technology. However, these tools typically employ a "single template fixed-fill" model, meaning that a single operation can only generate a video with a fixed structure, fixed music, and fixed order. If it is necessary to change the music, sound effects, or adjust the order of segments, it is often necessary to re-edit to match the timeline and rhythm, making it difficult to achieve flexible and diverse batch production.

[0004] Therefore, how to efficiently utilize original materials to achieve matrix-based and batch video generation while ensuring reasonable video content and controllable repetition, and how to dynamically optimize the generation strategy based on platform feedback data, has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] The main objective of this invention is to overcome the shortcomings of the prior art and provide a method, system and storage medium for batch generation of video matrix, which ensures that the video repetition rate does not exceed the selectable enthalpy value and conforms to common sense, maximizes the use of materials and produces high-quality and high-volume finished products; and allows all used content, including but not limited to video, text, music, sound effects and images, to be tracked in order to verify their effect.

[0006] The technical solution adopted by this invention to achieve its technical objective is: a method for batch generation of video matrix, comprising the following steps: S1. Build video editing templates: Obtain video drafts or import popular external videos as templates, identify the arrangement order of media materials, and mark specific timeline positions as slots to be filled; S2. Preprocessing of raw media footage: Denoising is performed on the raw media footage. Invalid segments, including but not limited to hesitation, shooting start delay, and failure to stop recording after the action is completed, are identified and removed by manual or AI model, and valid positive footage segments are extracted. S3. Construct a multi-dimensional material matrix: Arrange the effective positive clips in logical order and map them to a multi-dimensional candidate material matrix; wherein, the same logical position corresponds to one or more optional candidate clips; S4. Intelligent Sequence Filling: In response to the generation command, the material order is dynamically adjusted according to the actual situation of historical video data and materials, and candidate segments are randomly selected from the material matrix to fill the slots to be filled. S5. Sequence Deduplication Detection: Record the generated video segment combination sequence and calculate the similarity index between the current candidate video and the historical video library; if the duplication reaches different preset thresholds, the corresponding modification or reselection mechanism is triggered. S6. Closed-loop weight optimization: Obtain video performance data from external platforms, use a normalization algorithm to calculate the popularity index of the material clips, and update the call priority weight of each clip in the material matrix accordingly.

[0007] Preferably, in step S1, the slot to be filled includes at least one or more combinations of video track, subtitle track, audio track, and special effects track.

[0008] Preferably, the denoising process in step S2 includes at least one of image denoising, audio denoising, and timeline redundancy cropping.

[0009] Preferably, in step S3, the logical order includes chronological order, narrative order, action order, or scene switching order; The logical order is arranged as follows: multiple materials are marked with serial numbers on the timeline, and then the multiple materials are arranged sequentially on the timeline according to the serial numbers to form a material matrix.

[0010] Preferably, the nonlinear constraint criterion in step S4 further includes: when the system determines that the total duration of the remaining valid positive segments is less than the duration of the remaining slots, or the number of remaining segments is insufficient to support the subsequent filling requirements, the system executes a backtracking algorithm, returns to the current filling path, and re-executes the filling logic.

[0011] Preferably, the similarity index calculation method in step S5 includes at least one method based on n-gram sequence matching, hash fingerprint matching, and edit distance matching.

[0012] Preferably, the retry mechanism in step S5 includes: rejecting the current candidate fragment combination and re-extracting at least one candidate fragment from the material matrix to replace the original combination until the similarity is lower than the threshold or the maximum number of retries is reached.

[0013] Preferably, step S5 further includes: in response to the user's adoption operation for a specific piece, the system increases the duplication flag of the corresponding segment combination in the release library in real time, and marks it as a high duplication risk combination for subsequent filling to avoid.

[0014] Preferably, the external platform video performance data in step S6 includes at least one or more of the following: number of views, number of likes, number of comments, number of shares, completion rate, and follower gain rate.

[0015] Preferably, step S3 may include or exclude an automatic subtitle generation step: after completing the intelligent sequence filling, based on the filled segment content and combined with speech recognition or template mapping relationship, automatically generate or not generate subtitle content for the corresponding timeline.

[0016] Preferably, the process also includes video compression and publishing steps: after the filling and subtitle generation are completed, the video track, audio track, and subtitle track are automatically merged and compressed into a complete video file, and optionally automatically published to the target platform.

[0017] Preferably, the method also includes an automatic title generation step: based on the filled fragment combination content, a video title is automatically generated and corresponds one-to-one with its content through a text generation model or template matching method.

[0018] The present invention also provides a video matrix batch generation system, which is used in the aforementioned video matrix batch generation method, comprising: The template building module is used to obtain video drafts or import external videos, identify the arrangement order of media materials, and mark the slots to be filled. The material preprocessing module is used to perform noise reduction and invalid segment removal on the original media materials to extract valid positive segments; The material matrix construction module is used to arrange valid positive clips in logical order and construct a multi-dimensional candidate material matrix. The intelligent fill module is used to respond to the generation command, randomly select candidate segments from the material matrix to fill the slots, and follow the non-linear constraint of "allowing skipping and prohibiting reverse order"; The plagiarism detection module is used to record the generated fragment combination sequences, calculate the similarity index, and trigger the return and reselection mechanism; The closed-loop optimization module is used to obtain feedback data from external platforms, calculate the popularity index of creative materials, and update the weight of creative materials.

[0019] Preferably, the intelligent filling module further includes a backtracking submodule, which is used to backtrack and re-execute the filling logic when the remaining segment duration or quantity is insufficient.

[0020] Preferably, the plagiarism detection module further includes a repetition identifier update submodule, which is used to increase the repetition identifier of the corresponding segment combination in response to the user's acceptance operation.

[0021] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned video matrix batch generation method.

[0022] The computer program is also used to perform the following operations: perform weighting operations on all segments contained in videos with an average value above the average value of the same account, perform weighting operations on segments with an average value below the average value, and prioritize calling high-weight segments in subsequent batch generation.

[0023] The working principle of this video matrix batch generation method, system, and storage medium is as follows: First, an editing template with "slots to be filled" along a timeline is constructed by importing video drafts or popular videos. The original footage is then denoised and invalid segments are removed to extract valid main clips. These valid clips are then mapped into a multi-dimensional candidate material matrix according to logical order such as time and narrative, where one or more candidate clips correspond to the same logical position. Upon receiving a generation command, the system dynamically adjusts the order based on historical data and the actual situation of the materials, randomly selecting candidate clips from the matrix to fill the slots, adhering to a non-linear constraint of "allowing skipping, prohibiting reverse order." Simultaneously, a backtracking algorithm is used to refill the slots when the remaining time or number of clips is insufficient. A duplicate check is performed after each video combination is generated; if the duplication rate exceeds a threshold, a reselection or clip replacement is triggered. The system also obtains performance data such as playback and likes from external platforms, calculates the "viral index" of the materials using a normalization algorithm, and updates the call weight of each clip to achieve closed-loop optimization. Furthermore, it supports automatic generation of subtitles, titles, and compressed publishing, and marks the duplication rate of user-adopted finished products, thereby continuously improving the efficiency and quality of batch production and controlling the risk of duplication.

[0024] Compared with the prior art, the beneficial effects of the present invention are: This invention improves the efficiency and content rationality of batch video generation: by combining a strategy of "extracting effective main clips + constructing a multi-dimensional material matrix in logical order + intelligent sequence filling," it solves the problems of cumbersome batch production steps and illogical arrangement of clips in traditional template-based editing (such as chaotic time order, narrative order, action order, or scene transition order). Simultaneously, through non-linear constraints that "allow skipping and prohibit reverse order" and a backtracking algorithm, it ensures that the generated videos are logically coherent and clearly organized.

[0025] This invention effectively controls video repetition and enhances content diversity: by recording the generated video segment combination sequence and combining similarity calculation methods such as n-gram matching, hash fingerprint matching, or edit distance matching, each generated result is checked for duplication. When the repetition exceeds a preset threshold, a mechanism for reselection or segment replacement is automatically triggered until qualified content is generated. Furthermore, the system can respond to user adoption actions, marking and avoiding combinations with high repetition risk, thereby significantly reducing the risk of content homogenization during batch generation.

[0026] This invention establishes a closed-loop optimization mechanism to continuously enhance the potential of videos to go viral: By acquiring performance data from external platforms (such as short video platforms) including views, likes, comments, shares, and completion rates, a normalization algorithm is used to calculate the "viral index" of each video clip, and its priority weight in the video clip matrix is ​​dynamically updated. Clips in videos from the same account that perform above average receive bonus points, while clips that perform below average receive deductions. This ensures that subsequent batch generation prioritizes high-weight clips, achieving a data-driven positive cycle and continuously improving the overall quality and dissemination effect of generated videos.

[0027] This invention enables fully automated and traceable management of the entire process: It supports the automatic generation of subtitles, titles, video compression, and publishing, and uses a method of generating MD5 fingerprints based on "combined template ID + segment file name + ingress point + duration" to uniquely identify each generated video record. This fingerprint is not only used for deduplication detection, but also for tracking the frequency of use and performance of each material segment in different videos, providing data support for subsequent material selection, template optimization, and weight adjustment, thus establishing a complete feedback mechanism for video production. Attached Figure Description

[0028] Figure 1 This is a step-by-step logic diagram of the video matrix batch generation method.

[0029] Figure 2 This is a flowchart illustrating the steps of a method for batch generation of video matrices.

[0030] in: Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. However, it should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0032] In the description of this invention, it should be noted that when an element is referred to as being "fixed to" or "set on" another element, it can be directly on or indirectly on the other element. When an element is referred to as being "connected to" another element, it can be directly connected to or indirectly connected to the other element.

[0033] In the description of this invention, it should be noted that the terms "center," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. "Several" means one or more, unless otherwise explicitly specified.

[0034] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. Example 1:

[0035] Please see Figures 1-2 A method for batch generation of video matrices includes the following steps: S1. Build video editing templates: Obtain video drafts or import popular external videos as templates, identify the arrangement order of media materials, and mark specific timeline positions as slots to be filled.

[0036] In this embodiment, the slot to be filled includes at least one or more combinations of video track, subtitle track, audio track, and special effects track.

[0037] S2. Preprocessing of raw media footage: Denoising is performed on the raw media footage. Invalid segments, including but not limited to hesitation, shooting start delay, and failure to stop recording after the action, are identified and removed by manual or AI models, and valid positive segments are extracted.

[0038] In this embodiment, the denoising process includes at least one of image denoising, audio denoising, and timeline redundancy cropping.

[0039] S3. Construct a multi-dimensional material matrix: Arrange the effective positive clips in logical order and map them into a multi-dimensional candidate material matrix; wherein, the same logical position corresponds to one or more optional candidate clips.

[0040] In this embodiment, the logical order includes chronological order, narrative order, action order, or scene transition order. The logical order is arranged as follows: multiple materials are marked with serial numbers on the timeline, and then the multiple materials are arranged sequentially on the timeline according to the serial numbers to form a material matrix.

[0041] In this embodiment, the automatic subtitle generation step may or may not be included: after completing the intelligent sequence filling, based on the filled segment content and combined with speech recognition or template mapping relationship, the corresponding timeline subtitle content may or may not be generated automatically.

[0042] S4. Intelligent Sequence Filling: In response to the generation command, the material order is dynamically adjusted according to the actual situation of historical video data and materials, and candidate segments are randomly selected from the material matrix to fill the slots to be filled.

[0043] In this embodiment, the nonlinear constraint criterion also includes: when the system determines that the total duration of the remaining valid positive segments is less than the duration of the remaining slots, or the number of remaining segments is insufficient to support the subsequent filling requirements, the system executes a backtracking algorithm, returns to the current filling path, and re-executes the filling logic.

[0044] S5. Sequence Deduplication Detection: Record the generated video segment combination sequence and calculate the similarity index between the current candidate video and the historical video library; if the duplication reaches different preset thresholds, the corresponding modification or reselection mechanism is triggered. It also includes responding to a user's adoption of a specific piece of content, the system in real time increasing the duplication flag of the corresponding segment combination in the release library, and marking it as a high-risk duplication combination for subsequent filling to avoid.

[0045] In this embodiment, the similarity index is calculated using at least one of the following methods: n-gram sequence matching, hash fingerprint matching, and edit distance matching.

[0046] The fallback and reselection mechanism includes: rejecting the current candidate fragment combination and re-extracting at least one candidate fragment from the material matrix to replace the original combination, until the similarity is lower than the threshold or the maximum number of retries is reached.

[0047] S6. Closed-loop weight optimization: Obtain video performance data from external platforms, use a normalization algorithm to calculate the popularity index of the material clips, and update the call priority weight of each clip in the material matrix accordingly.

[0048] In this embodiment, the video performance data from the external platform includes at least one or more of the following: number of plays, number of likes, number of comments, number of shares, completion rate, and average viewing time.

[0049] S7. Video Compression and Publishing: After completing the filling and subtitle generation, the video track, audio track, and subtitle track are automatically merged and compressed into a complete video file, and can optionally be automatically published to the target platform.

[0050] S8. Automatic title generation: Based on the combined content of the filled segments, the video title is automatically generated and corresponds one-to-one with the content through text generation model or template matching.

[0051] Specifically, in use, the system first extracts the arrangement order of media materials from existing video drafts or external viral videos, and marks key positions on the timeline as replaceable "slots" (such as video tracks, subtitle tracks, sound effects tracks, special effects tracks, etc.).

[0052] Subsequently, the original footage is preprocessed, and invalid segments such as hesitant actions, start delays, and failure to stop recording at the end are identified and removed by manual or AI models, retaining only the valid main footage segments.

[0053] Next, these effective fragments are organized into a multi-dimensional candidate material matrix according to logical relationships such as time order, narrative order, or action order, where the same logical position can correspond to multiple replaceable candidate fragments.

[0054] After receiving the generation instruction, the system randomly selects candidate segments from the matrix to fill the template slots, and strictly follows the filling logic of "allowing skipping and prohibiting reverse order".

[0055] After each generation, the system records the sequence of video clips and compares them with the historical video library for similarity (such as n-gram matching, hash fingerprinting, etc.). If the repetition exceeds the preset threshold, it will trigger a reselection or clip replacement.

[0056] Finally, the system automatically collects video performance data (views, likes, completion rate, etc.) from external platforms (such as Douyin, Kuaishou, etc.), calculates the "viral index" of each material segment through a normalization algorithm, and dynamically updates its call weight in the matrix to form a positive closed loop. Implementation: 2:

[0057] Based on the above embodiments, the present invention also provides a video matrix batch generation system, which is used in the video matrix batch generation method and includes a template construction module, a material preprocessing module, a material matrix construction module, an intelligent filling module, a deduplication detection module, and a closed-loop optimization module.

[0058] In this embodiment, the template building module is used to obtain video drafts or import external videos, identify the arrangement order of media materials, and mark the slots to be filled.

[0059] In this embodiment, the material preprocessing module is used to perform noise reduction and invalid segment removal on the original media material to extract valid positive segments.

[0060] In this embodiment, the material matrix construction module is used to arrange the valid positive clips in logical order to construct a multi-dimensional candidate material matrix.

[0061] In this embodiment, the intelligent fill module is used to respond to the generation command, randomly select candidate segments from the material matrix to fill the slots, and follow the non-linear constraint of "allowing skipping and prohibiting reverse order".

[0062] Furthermore, the intelligent filling module also includes a backtracking submodule, which is used to backtrack and re-execute the filling logic when the remaining segment duration or quantity is insufficient.

[0063] In this embodiment, the deduplication detection module is used to record the generated fragment combination sequence, calculate the similarity index, and trigger the back-to-back reselection mechanism.

[0064] Furthermore, the plagiarism detection module also includes a repetition identifier update submodule, which is used to increase the repetition identifier of the corresponding segment combination in response to the user's acceptance operation.

[0065] In this embodiment, the closed-loop optimization module is used to obtain feedback data from external platforms, calculate the popularity index of the content, and update the weight of the content.

[0066] Specifically, in use, firstly, the template building module obtains the video draft imported by the user or the external popular video, automatically identifies the arrangement order of the media materials, and marks the key positions on the timeline as slots to be filled; Subsequently, the material preprocessing module performs noise reduction on the original media material and removes invalid segments such as hesitant actions and start delays, thereby extracting valid main clips; Next, the material matrix construction module arranges these valid fragments according to logical order such as time sequence and narrative order to construct a multi-dimensional candidate material matrix, in which the same logical position can correspond to multiple optional fragments; Afterwards, the intelligent fill module responds to the generation command, randomly selects candidate segments from the material matrix to fill each slot, and strictly follows the non-linear constraint of "allowing skipping and prohibiting reverse order". At the same time, its backtracking submodule will automatically return to the current fill path and re-execute the fill logic when the remaining segment duration or number is insufficient to complete the fill. Each time a fill is completed, the deduplication detection module records the generated segment combination sequence, calculates its similarity index with the historical video library, and triggers a back-to-select mechanism when the repetition exceeds the limit. The repetition label update submodule will increase the repetition label of the corresponding segment combination when the user adopts a video, so that subsequent fills can avoid it. Finally, the closed-loop optimization module obtains feedback data such as play counts and likes from external platforms, calculates the popularity index of each material segment, and dynamically updates the call priority weight in the material matrix, thereby continuously optimizing the quality and effect of subsequent batch-generated videos.

[0067] The solution in this embodiment can be selectively combined with solutions in other embodiments. Implementation: 3:

[0068] Based on the above embodiments, the present invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the video matrix batch generation method described above.

[0069] The computer program is also used to perform the following operations: perform weighting operations on all segments contained in videos with an average value above the average value of the same account, perform weighting operations on segments with an average value below the average value, and prioritize calling high-weight segments in subsequent batch generation.

[0070] The solution in this embodiment can be selectively combined with solutions in other embodiments.

[0071] The specific usage process of this video matrix batch generation method, system, and storage medium is as follows: (1) Template construction: Users import video drafts or select external popular videos as templates. The system automatically identifies or manually adjusts the arrangement order of media materials and marks key positions on the timeline (such as video track segments, subtitle positions, sound effect start points, special effect periods, etc.) as "slots to be filled".

[0072] (2) Original material preprocessing: The user uploads the original shooting or recording material, and the system performs image denoising, audio denoising and timeline redundancy cropping. Through manual annotation or AI model, invalid segments including but not limited to hesitant actions, start delays, and failure to stop recording after the action is completed are automatically removed, and several "valid positive film segments" are output.

[0073] (3) Constructing a multi-dimensional material matrix: The system marks all valid main clips with serial numbers according to their logical order (such as chronological order, narrative progression, action transitions, or scene changes), and arranges them sequentially to form a material matrix. Multiple candidate clips are allowed to correspond to the same logical position for subsequent random selection.

[0074] (4) Intelligent Sequence Filling: After the user issues the generation command, the system dynamically adjusts the filling order of the materials based on historical video data and the actual situation such as the remaining duration and quantity of the current materials. Candidate segments are randomly selected from the material matrix to fill the slots to be filled in the template, and the non-linear constraints of "allowing skipping (slots can be skipped) and prohibiting reverse order (logical order cannot be reversed)" are strictly followed. If the system determines that the total duration or quantity of the remaining valid segments is insufficient to support the subsequent slots, the backtracking algorithm is automatically executed to return to the current filling path and re-execute the filling logic.

[0075] (5) Sequence deduplication detection: For each video combination generated, the system records its segment combination sequence and calculates the similarity index between the current candidate video and the historical video library (using n-gram matching, hash fingerprint matching, or edit distance matching, etc.). If the repetition reaches the preset mild threshold, the system triggers local modification (such as replacing a segment in a slot); if it reaches the severe threshold, the combination is rejected and forced to reselect until the similarity is lower than the threshold or the maximum number of retries is reached. In addition, if a user performs an "adoption" operation on a video, the system will increase the repetition flag of the corresponding segment combination in real time, marking it as a high-risk repetition combination for subsequent filling.

[0076] (6) Closed-loop weight optimization: The video performance data (play count, likes, comments, reposts, completion rate, follower increase rate, etc.) uploaded by the client from external platforms (such as short video platforms) are used to calculate the "viral index" of each material segment using a normalization algorithm, and the priority weight of each segment in the material matrix is ​​updated accordingly. For segments contained in videos with performance above the average under the same account, weight is increased; for segments with performance below the average, weight is decreased, and high-weight segments are prioritized in subsequent batch generation.

[0077] (7) Automatic Composition and Publishing: After completing the filling and subtitle generation, the system merges and compresses the video track, audio track, and subtitle track into a single complete video file, and can optionally automatically publish it to the target platform. At the same time, based on the combined content of the filled segments, the system automatically generates a video title through a text generation model or template matching method, which corresponds one-to-one with the video.

[0078] Through the above process, this invention achieves semi-automatic intelligent production from raw materials to batch finished products, while ensuring the diversity and repetition of content are controllable, and continuously optimizing the generation strategy through data feedback.

[0079] It should be noted that although the above embodiments have been described herein, this does not limit the scope of patent protection of this invention. Therefore, any changes and modifications made to the embodiments described herein based on the innovative concept of this invention, or equivalent structural, procedural, or functional transformations made using the description and drawings of this invention, directly or indirectly applying the above technical solutions to other related technical fields, are all included within the scope of protection of this invention.

Claims

1. A method for batch generation of video matrix, characterized in that, Includes the following steps: S1. Build video editing templates: Obtain video drafts or import popular external videos as templates, identify the arrangement order of media materials, and mark specific timeline positions as slots to be filled; S2. Preprocessing of raw media footage: Denoising is performed on the raw media footage. Invalid segments, including but not limited to hesitation, shooting start delay, and failure to stop recording after the action is completed, are identified and removed by manual or AI model, and valid positive footage segments are extracted. S3. Construct a multi-dimensional material matrix: Arrange the effective positive clips in logical order and map them to a multi-dimensional candidate material matrix; wherein, the same logical position corresponds to one or more optional candidate clips; S4. Intelligent Sequence Filling: In response to the generation command, the material order is dynamically adjusted according to the actual situation of historical video data and materials, and candidate segments are randomly selected from the material matrix to fill the slots to be filled. S5. Sequence Deduplication Detection: Record the generated video segment combination sequence and calculate the similarity index between the current candidate video and the historical video library; if the duplication reaches different preset thresholds, the corresponding modification or reselection mechanism is triggered. S6. Closed-loop weight optimization: Obtain video performance data from external platforms, use a normalization algorithm to calculate the popularity index of the material clips, and update the call priority weight of each clip in the material matrix accordingly.

2. The video matrix batch generation method according to claim 1, characterized in that, In step S1, the slot to be filled includes at least one or more combinations of video track, subtitle track, audio track, and special effects track.

3. The video matrix batch generation method according to claim 1, characterized in that, The denoising process in step S2 includes at least one of image denoising, audio denoising, and timeline redundancy cropping.

4. The video matrix batch generation method according to claim 1, characterized in that, In step S3, the logical order includes chronological order, narrative order, action order, or scene switching order. The logical order is arranged as follows: multiple materials are marked with serial numbers on the timeline, and then the multiple materials are arranged sequentially on the timeline according to the serial numbers to form a material matrix.

5. The video matrix batch generation method according to claim 1, characterized in that, The nonlinear constraint criteria in step S4 also include: when the system determines that the total duration of the remaining valid positive segments is less than the duration of the remaining slots, or the number of remaining segments is insufficient to support the subsequent filling requirements, the system executes a backtracking algorithm, returns to the current filling path, and re-executes the filling logic.

6. The video matrix batch generation method according to claim 1, characterized in that, The similarity index calculation method in step S5 includes at least one of the following: n-gram sequence matching, hash fingerprint matching, and edit distance matching.

7. The video matrix batch generation method according to claim 1, characterized in that, The retry mechanism in step S5 includes: rejecting the current candidate fragment combination and re-extracting at least one candidate fragment from the material matrix to replace the original combination until the similarity is lower than the threshold or the maximum number of retries is reached.

8. The video matrix batch generation method according to claim 1, characterized in that, Step S5 further includes: in response to the user's adoption operation for a specific piece, the system increases the duplication flag of the corresponding segment combination in the release library in real time and marks it as a high duplication risk combination for subsequent filling to avoid.

9. The video matrix batch generation method according to claim 1, characterized in that, The external platform video performance data in step S6 includes at least one or more of the following: number of plays, number of likes, number of comments, number of shares, completion rate, and average viewing time.

10. A video matrix batch generation system, used in the video matrix batch generation method as described in any one of claims 1 to 9, characterized in that, include: The template building module is used to obtain video drafts or import external videos, identify the arrangement order of media materials, and mark the slots to be filled. The material preprocessing module is used to perform noise reduction and invalid segment removal on the original media materials to extract valid positive segments; The material matrix construction module is used to arrange valid positive clips in logical order and construct a multi-dimensional candidate material matrix. The intelligent fill module is used to respond to the generation command, randomly select candidate segments from the material matrix to fill the slots, and follow the non-linear constraint of "allowing skipping and prohibiting reverse order"; The plagiarism detection module is used to record the generated fragment combination sequences, calculate the similarity index, and trigger the return and reselection mechanism; The closed-loop optimization module is used to obtain feedback data from external platforms, calculate the popularity index of creative materials, and update the weight of creative materials.

Images