Short video confusion encryption method and device

Abstract

The embodiment of the invention provides a short video confusion encryption method and device, and the method comprises the steps: carrying out the semantic clustering of original video subtitles and supplementary video subtitles, determining the time axis of each video clip, carrying out the lower-layer main track playing setting of each video clip according to the time axis, and carrying out the lower-layer main track playing setting. Performing upper-layer auxiliary track opaque playing setting on each corresponding video clip according to the confused time axis, and obtaining an encrypted short video file according to the video clip of the lower-layer main track playing setting and the video clip of the upper-layer auxiliary track opaque playing setting; according to the invention, short video subtitle processing can be accurately and efficiently carried out.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and specifically to a short video obfuscation encryption method and device. Background Art

[0002] With the booming development of short video platforms, the security and privacy protection of short video content has become an increasingly important issue. Traditional video encryption methods mainly focus on encrypting the entire video file. Although this method can protect the video content, it often causes the encrypted video to be unable to be played in regular players, affecting the user experience. At the same time, this full-file encryption method is also difficult to meet the flexible needs of partial content being visible and partial content being encrypted.

[0003] At present, some researchers have begun to explore more flexible and intelligent short video encryption methods. Among them, selective encryption based on video content has become an important research direction. This method analyzes the video content and adopts different encryption strategies for different parts, which can protect sensitive information while maintaining the accessibility of some content. However, how to accurately identify and locate the video content that needs to be encrypted remains a challenge.

[0004] The development of computer vision technology has provided new tools for video content analysis. By using computer vision algorithms, automatic video segmentation and scene recognition can be achieved. This lays the foundation for more sophisticated video content encryption. However, how to effectively apply these technologies to the short video encryption process still requires further research.

[0005] Speech recognition technology also plays an important role in video processing. By extracting the speech content in the video and converting it into text, it can provide additional information for understanding the video content. However, existing speech recognition technology still has the problem of insufficient accuracy when dealing with scenes such as complex background sounds and multi-person conversations. In addition, for video clips that are silent or dominated by background music, how to effectively supplement the missing subtitle information is also a difficult problem.

[0006] The rapid development of deep learning technology has provided new possibilities for solving the above problems. Deep learning models such as multi-layer perceptrons and autoregressive generative models have performed well in feature extraction and content generation. However, how to effectively apply these models to the short video encryption process, especially in processing the spatiotemporal features and semantic understanding of videos, still requires further exploration and optimization.

[0007] In the field of video encryption, obfuscation technology, as an important supplementary means, has received increasing attention in recent years. Traditional obfuscation technology mainly targets static data or program code. How to effectively apply obfuscation technology to dynamic video content, especially how to achieve effective content protection while ensuring video quality, remains a challenge.

[0008] In addition, existing video encryption methods often ignore the timeline characteristics of the video. How to use the timeline information of the video to achieve more flexible and effective encryption while maintaining the basic playability of the video is a direction worth studying.

[0009] To sum up, although short video encryption technology has made certain progress, there are still many challenges in content identification, selective encryption, application of obfuscation technology, and timeline utilization, which require further research and innovation. Summary of the invention

[0010] In response to the problems in the prior art, the present application provides a short video obfuscation encryption method and device, which can effectively perform short video security encryption through multi-track superposition.

[0011] In order to solve at least one of the above problems, the present application provides the following technical solutions:

[0012] In a first aspect, the present application provides a short video obfuscation encryption method, comprising:

[0013] Receive a short video file to be encrypted uploaded by a user, and divide the short video file to be encrypted into multiple video clips by using a preset computer vision algorithm;

[0014] Extracting subtitles from the video clips by using a preset speech recognition algorithm to obtain corresponding original video subtitles, and determining the video clips to be supplemented that lack corresponding video subtitles, extracting spatiotemporal features of the video clips to be supplemented by a frame division operation, obtaining low-order feature representations based on the spatiotemporal features, unifying the extracted low-order feature representations into a feature set and inputting them into a set deep learning multi-layer perceptron to obtain corresponding high-order feature representations, and determining the supplementary video subtitles corresponding to the video clips to be supplemented based on the high-order feature representations and a set autoregressive generation model;

[0015] Semantic clustering is performed on the original video subtitles and the supplementary video subtitles to determine the timeline of each of the video clips, and each of the video clips is set to be played on a lower main track according to the timeline, and each of the corresponding video clips is set to be played on an upper sub-track opaquely according to the timeline after the obfuscation process, and an encrypted short video file is obtained based on the video clips set to be played on the lower main track and the video clips set to be played on the upper sub-track opaquely.

[0016] Further, the video clips are set to be played on the lower main track according to the timeline, and the corresponding video clips are set to be played on the upper sub-track opaquely according to the timeline after the obfuscation process, and the encrypted short video file is obtained according to the video clips set to be played on the lower main track and the video clips set to be played on the upper sub-track opaquely, including:

[0017] According to the time axis of the video clips, each video clip is set to be played as the main track of the lower layer, and the video clips are randomly obfuscated on the time axis, and the obfuscated video clips are set to be played as the sub-track of the upper layer in an opaque manner;

[0018] The video clip played on the set lower main track and the obfuscated video clip played opaquely on the upper sub-track are fused and rendered to obtain an encrypted short video file.

[0019] Furthermore, the determining the supplementary video subtitles corresponding to the to-be-supplemented video segment according to the high-order feature representation and setting the autoregressive generation model includes:

[0020] The target high-order feature representations are weightedly fused through the attention mechanism to obtain the target feature representation of the video segment to be supplemented, and the target feature representation is semantically recognized through a preset knowledge graph inference algorithm. The result of the semantic recognition is input into a set autoregressive generation model, and the supplementary video subtitles corresponding to the video segment to be supplemented are determined according to the output of the autoregressive generation model.

[0021] Furthermore, the extracting subtitles from the video clips by using a preset speech recognition algorithm to obtain corresponding original video subtitles, and determining the video clips to be supplemented that lack corresponding video subtitles, includes:

[0022] Using a pre-trained speech recognition model to transcribe the audio content in the video clip, extracting corresponding original video subtitles, and performing a rationality check on the extracted original video subtitles;

[0023] The video segments to be supplemented that need to be supplemented with subtitles are determined based on the extracted original video subtitles, video duration, and picture change characteristics.

[0024] Furthermore, the extracting the spatiotemporal features of the video segment to be supplemented by the frame division operation and obtaining a low-order feature representation according to the spatiotemporal features includes:

[0025] Performing a frame division operation on the video segment to be supplemented that is determined to need to be supplemented with subtitles, dividing it into a continuous video frame sequence, and extracting spatiotemporal features including at least one of color, texture, and motion information for each video frame in the video frame sequence;

[0026] The extracted spatiotemporal features are compressed into a compact low-level feature representation, wherein the low-level feature representation includes at least basic visual and dynamic characteristics of video content.

[0027] Furthermore, the extracted low-order feature representations are unified into a feature set and input into a set deep learning multi-layer perceptron to obtain corresponding high-order feature representations, including:

[0028] Integrating the extracted low-level feature representations into a unified feature set, wherein the feature set contains rich visual information of the video clip;

[0029] The feature set is input into a set deep learning multi-layer perceptron model for multi-level feature transformation and combination, so as to obtain a high-order feature representation describing the video content output by the deep learning multi-layer perceptron model.

[0030] Furthermore, the semantic recognition of the target feature representation is performed by a preset knowledge graph inference algorithm, the result of the semantic recognition is input into a set autoregressive generation model, and the supplementary video subtitle corresponding to the to-be-supplemented video segment is determined according to the output of the autoregressive generation model, including:

[0031] Use the pre-built knowledge graph to semantically match and associate the obtained high-order feature representation with the concept entities in the graph to determine the semantic concepts of product functions, attributes, and scenarios corresponding to the high-order feature representation;

[0032] The identified semantic concepts are input as conditions into a pre-trained autoregressive text generation model to obtain supplementary video subtitles that match the semantic concepts and are output by the autoregressive text generation model.

[0033] In a second aspect, the present application provides a short video obfuscation encryption device, comprising:

[0034] A video segment determination module is used to receive a short video file to be encrypted uploaded by a user, and divide the short video file to be encrypted into segments using a preset computer vision algorithm to obtain multiple video segments;

[0035] A video subtitle determination module is used to extract subtitles from the video segment by a preset speech recognition algorithm to obtain corresponding original video subtitles, and determine the video segment to be supplemented that lacks corresponding video subtitles, extract the spatiotemporal features of the video segment to be supplemented by a frame division operation, obtain a low-order feature representation based on the spatiotemporal features, unify the extracted low-order feature representation into a feature set and input it into a set deep learning multi-layer perceptron to obtain a corresponding high-order feature representation, and determine the supplementary video subtitles corresponding to the video segment to be supplemented based on the high-order feature representation and a set autoregressive generation model;

[0036] The video obfuscation encryption module is used to perform semantic clustering on the original video subtitles and the supplementary video subtitles, determine the timeline of each of the video clips, set each of the video clips to play on the lower main track according to the timeline, and set each of the corresponding video clips to play on the upper sub-track opaquely according to the obfuscated timeline, and obtain an encrypted short video file based on the video clips set to play on the lower main track and the video clips set to play on the upper sub-track opaquely.

[0037] In a third aspect, the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the short video obfuscation encryption method when executing the program.

[0038] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the short video obfuscation encryption method.

[0039] In a fifth aspect, the present application provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the short video obfuscation encryption method.

[0040] It can be seen from the above technical scheme that the present application provides a short video obfuscation encryption method and device, which performs semantic clustering on the original video subtitles and the supplementary video subtitles, determines the timeline of each video clip, sets each video clip for lower-level main track playback according to the timeline, and sets each corresponding video clip for upper-level sub-track opaque playback according to the obfuscated timeline, obtains an encrypted short video file based on the video clips set for lower-level main track playback and the video clips set for upper-level sub-track opaque playback, thereby enabling accurate and efficient short video subtitle processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0042] Figure 1 This is one of the flow charts of the short video obfuscation encryption method in the embodiment of the present application;

[0043] Figure 2 This is the second flow chart of the short video obfuscation encryption method in the embodiment of the present application;

[0044] Figure 3 This is the third flow chart of the short video obfuscation encryption method in the embodiment of the present application;

[0045] Figure 4 This is a fourth flow chart of the short video obfuscation encryption method in an embodiment of the present application;

[0046] Figure 5 This is the fifth flow chart of the short video obfuscation encryption method in the embodiment of the present application;

[0047] Figure 6 This is the sixth flow chart of the short video obfuscation encryption method in the embodiment of the present application;

[0048] Figure 7 This is the seventh flow chart of the short video obfuscation encryption method in the embodiment of the present application;

[0049] Figure 8 This is a structural diagram of a short video obfuscation encryption device in an embodiment of the present application;

[0050] Fig. 9 It is a schematic diagram of the structure of an electronic device in an embodiment of the present application.

[0051] Reference numerals:

[0052] Electronic device 9600, central processing unit 9100, memory 9140, communication module 9110, input unit 9120, audio processor 9130, display 9160, power supply 9170, buffer memory 9141, application / function storage unit 9142, data storage unit 9143, driver program storage unit 9144, antenna 9111, speaker 9131, microphone 9132. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0054] The acquisition, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of national laws and regulations.

[0055] In view of the problems existing in the prior art, the present application provides a short video obfuscation encryption method and device, which semantically clusters the original video subtitles and the supplementary video subtitles, determines the timeline of each video clip, sets the lower main track playback of each video clip according to the timeline, and sets the upper sub-track opaque playback of each corresponding video clip according to the obfuscated timeline, obtains an encrypted short video file based on the video clips set for the lower main track playback and the video clips set for the upper sub-track opaque playback, thereby enabling accurate and efficient short video subtitle processing.

[0056] In order to accurately and efficiently process short video subtitles, this application provides an embodiment of a short video obfuscation encryption method, see Figure 1 The short video obfuscation encryption method specifically includes the following contents:

[0057] Step S101: receiving a short video file to be encrypted uploaded by a user, and dividing the short video file to be encrypted into segments using a preset computer vision algorithm to obtain multiple video clips;

[0058] Optionally, in this embodiment, the system first receives the short video file to be encrypted uploaded by the user. The upload interface is designed as a user-friendly interface that supports multiple video formats to ensure that users can upload smoothly on different devices. After the upload is completed, the system automatically starts the preset computer vision algorithm to process the short video.

[0059] The core function of computer vision algorithms is to segment videos into multiple segments. The purpose of segmentation is to identify scene change points in the video, thereby dividing the video into logically complete segments. The system detects scene changes by analyzing the visual features of video frames, such as color histograms, edge information, and motion vectors.

[0060] In the specific implementation process, the system first analyzes the video frames frame by frame and extracts the visual features of each frame. These features include color, texture and motion information, which are obtained through feature extraction methods such as SIFT, SURF and optical flow. The system inputs these features into the scene detection model, and the model uses the trained parameters to judge the similarity between frames.

[0061] When the difference between the detected frames exceeds the set threshold, the system determines it as a scene switching point and performs storyboard segmentation at that point. This method ensures the accuracy of segmentation, so that each video segment has relatively independent content. At the same time, the system also takes into account the continuity of the video and the user's viewing experience, avoiding too frequent segmentation that results in too short segments.

[0062] Storyboard segmentation solves a key problem in video processing, namely how to accurately identify and segment scenes without affecting the smoothness of the video. This method not only improves the efficiency of video processing, but also facilitates subsequent encryption and editing operations.

[0063] In terms of application, this technology can be widely used in video editing, content review, copyright protection and other fields. For example, in video editing, editors can quickly locate important scenes through automatic storyboarding, saving a lot of time. In content review, the system can automatically identify and mark sensitive scenes in the video, improving review efficiency and accuracy.

[0064] In addition, this technology also plays an important role in the field of video encryption. When encrypting a video, the system can adopt different encryption strategies for different segments through segmentation, which increases the flexibility and security of encryption. This method is particularly suitable for video content that needs to be protected, such as movies, TV series, and educational videos.

[0065] In summary, this embodiment demonstrates a video segmentation method based on computer vision algorithm. By combining advanced visual feature extraction and scene detection technology, the system realizes intelligent segmentation of video content and provides an efficient and accurate solution for multimedia processing.

[0066] Step S102: extracting subtitles from the video clip by a preset speech recognition algorithm to obtain corresponding original video subtitles, and determining the video clip to be supplemented that lacks corresponding video subtitles, extracting spatiotemporal features of the video clip to be supplemented by a frame division operation, obtaining a low-order feature representation based on the spatiotemporal features, unifying the extracted low-order feature representation into a feature set and inputting it into a set deep learning multi-layer perceptron to obtain a corresponding high-order feature representation, and determining the supplementary video subtitles corresponding to the video clip to be supplemented based on the high-order feature representation and a set autoregressive generation model;

[0067] Optionally, in this embodiment, the system first processes the video clips through a preset speech recognition algorithm to extract the original video subtitles. The speech recognition algorithm is based on deep learning technology and can accurately transcribe audio signals into text. This step identifies each word by analyzing the spectral characteristics of the speech in the audio and generates the corresponding subtitle text.

[0068] After completing the subtitle extraction, the system needs to determine which video segments are missing the corresponding subtitles. This is usually achieved by comparing the audio content with the generated subtitles to identify the parts that are incomplete or missing. After identifying the video segments to be supplemented, the system performs frame operations on these segments.

[0069] Framing is the process of breaking a video into a continuous sequence of images. Through this process, the system can extract the spatiotemporal features of each frame. Spatiotemporal features include color, texture, and motion information, which are extracted using computer vision techniques. Specifically, color features can be obtained through color histograms, texture features can be extracted through methods such as gray-level co-occurrence matrices, and motion information is usually obtained through optical flow analysis.

[0070] From these spatiotemporal features, the system further extracts low-order feature representations. Low-order feature representations are a simplification and compression of spatiotemporal features, retaining the basic visual and dynamic characteristics of the video content. To achieve this, the system may use dimensionality reduction techniques, such as principal component analysis (PCA), to reduce the dimensionality of the features and improve computational efficiency.

[0071] Next, the system integrates the extracted low-level feature representations into a unified feature set and inputs it into the set deep learning multi-layer perceptron (MLP) model. The MLP model uses a multi-layer neural network structure to perform complex transformations and combinations on the input features to generate high-level feature representations. These high-level features reveal the deep semantic information of the video and provide the necessary foundation for subsequent subtitle generation.

[0072] Finally, the system uses the set autoregressive generation model to determine the supplementary subtitles corresponding to the video clip to be supplemented based on the obtained high-order feature representation. The autoregressive model gradually generates text based on contextual information to ensure that the generated subtitles are fluent and logically coherent. This generation method can fill in the blanks in the original subtitles and provide complete subtitle content by understanding high-order features.

[0073] This technical process solves several key problems in video subtitle generation. It not only improves the speed and accuracy of subtitle generation, but also ensures the semantic consistency of subtitles and video content. In practical applications, this technology can be used for automatic subtitle generation of movies, TV series, and online educational videos, improving video production efficiency and user experience.

[0074] For example, in film production, the system can quickly identify and supplement missing subtitles, reducing the workload of manual editing. In educational videos, teachers can use the system to generate accurate subtitles to help students better understand the video content. In addition, the technology can also be extended to real-time subtitle generation and multilingual translation to facilitate global audiences.

[0075] In summary, this embodiment demonstrates a comprehensive technical solution combining speech recognition, computer vision and deep learning. Through the application of multi-level feature analysis and generation models, the system realizes intelligent supplementation of video subtitles and provides an innovative solution for multimedia content processing.

[0076] Step S103: semantically clustering the original video subtitles and the supplementary video subtitles, determining the timeline of each of the video clips, setting each of the video clips for lower-layer main track playback according to the timeline, and setting each of the corresponding video clips for upper-layer sub-track opaque playback according to the obfuscated timeline, and obtaining an encrypted short video file based on the video clips set for lower-layer main track playback and the video clips set for upper-layer sub-track opaque playback.

[0077] Optionally, semantic clustering is first performed on the original video subtitles and the supplementary video subtitles. The purpose of semantic clustering is to group subtitles according to the semantic similarity of the subtitle content in order to determine the timeline of each video segment. Through natural language processing technology, the system analyzes the semantic features of the subtitles and clusters similar subtitle content together. This step ensures the logical consistency of the subtitle content and the video segment, providing a basis for the subsequent construction of the timeline.

[0078] After completing the semantic clustering, the system determines the exact timeline of each video clip. The timeline is determined based on the semantic flow of the subtitle content and the actual playback order of the video. In this way, the system can ensure that the playback time of each video clip matches its corresponding subtitle content, improving the overall logical fluency of the video.

[0079] Next, the system sets the lower-level main track playback settings for each video clip according to the determined timeline. The lower-level main track playback setting is the basic video playback track, ensuring that the video is displayed in a normal order and logic. This setting provides a guarantee for the basic playback effect of the video, allowing the audience to watch the video content according to the expected timeline.

[0080] At the same time, the system processes the obfuscated timeline and sets the corresponding video clips to play opaquely on the upper sub-track. The obfuscated timeline increases the complexity of the video content by disrupting the original order, thereby improving the security and encryption of the video. The upper sub-track opaque playback setting ensures that these obfuscated video clips can cover the lower main track video, thereby achieving encryption of the video content.

[0081] This dual-track playback strategy solves a core problem in video encryption, that is, how to enhance the security and anti-cracking of video content without affecting the overall viewing quality of the video. By combining the lower main track and the upper sub-track, the system can provide a complete and secure playback experience.

[0082] In practical applications, this technology can be used for copyright protection and encryption of sensitive content. For example, film production companies can use this technology to encrypt trailers of new movies to prevent unauthorized distribution. In the field of education, online courses can use this technology to encrypt important teaching videos to ensure that only authorized users can access them.

[0083] In addition, the encryption technology can also be applied to the commercial release and live broadcast of videos. Through real-time encryption and decryption, the system can ensure the security of live broadcast content and prevent the content from being intercepted and illegally disseminated. This capability is of great significance for maintaining the exclusivity and commercial value of content.

[0084] In summary, this embodiment demonstrates a video encryption method implemented through semantic clustering and dual-track playback settings. By combining advanced natural language processing and video processing technologies, the system implements intelligent encryption of video content, providing an innovative solution for copyright protection and secure transmission of multimedia content. This technology has broad application prospects in many fields and provides new possibilities and directions for the development of video technology.

[0085] From the above description, it can be seen that the short video obfuscation encryption method provided in the embodiment of the present application can determine the timeline of each video clip by semantically clustering the original video subtitles and the supplementary video subtitles, set each video clip to play on the lower main track according to the timeline, and set each corresponding video clip to play on the upper sub-track opaquely according to the obfuscated timeline, and obtain the encrypted short video file based on the video clips set to play on the lower main track and the video clips set to play on the upper sub-track opaquely, thereby accurately and efficiently processing short video subtitles.

[0086] In one embodiment of the short video obfuscation encryption method of the present application, see Figure 2 , and can also include the following:

[0087] Step S201: setting each video clip to be played as a lower-layer main track according to the time axis of the video clip, performing random time axis obfuscation processing on the video clip, and setting the obfuscated video clip to be played as an upper-layer secondary track in an opaque manner;

[0088] Step S202: The video clip played on the set lower main track and the confused video clip played opaquely on the upper sub-track are merged and rendered to obtain an encrypted short video file.

[0089] Optionally, in this embodiment, the system first sets each segment as the lower master track for playback according to the timeline of the video segment. The playback order of the lower master track is arranged in the logical order of the original timeline to ensure the basic playback effect of the video content. This step lays the foundation for the normal playback of the video, allowing the audience to watch the video content in the expected order.

[0090] Next, the system performs random timeline obfuscation on these video clips. The purpose of timeline obfuscation is to increase the security of the video by disrupting the order of the video clips. This obfuscation is achieved through a random algorithm to ensure that the obfuscation results are different each time, increasing the difficulty of decryption. The obfuscated clips are set to play opaquely on the upper secondary track. This dual-track playback method means that the upper obfuscated track will cover the lower main track, but not completely cover it, thereby achieving an encryption effect.

[0091] In step S202, the system performs fusion rendering on the set video clip played on the lower main track and the obfuscated video clip played opaquely on the upper sub-track. Fusion rendering refers to combining the video content of the two tracks into a whole video file. The system synchronizes the content of the upper and lower tracks through frame-by-frame rendering technology to ensure the smoothness and integrity of video playback. The encrypted short video file finally generated not only contains the original content, but also adds security protection through the opaque obfuscation layer.

[0092] This encryption method solves the problem of traditional video encryption that may affect viewing quality. Through the dual-track setting, the system can provide a playback solution that is both safe and does not affect the user experience. The audience will not be aware of the encryption process when watching, and it is difficult for unauthorized decryptors to crack the video content.

[0093] In practical applications, this technology can be used to protect sensitive video content. For example, movie production companies can use this method to encrypt movie trailers to prevent pirated copies from being distributed. In the education field, online course content can be encrypted through this technology to ensure that only authorized users can access it.

[0094] In addition, the encryption technology can also be used in live broadcasts of commercial activities. Through real-time encryption and decryption, the system can ensure the security of live broadcast content and prevent it from being intercepted and illegally disseminated. This capability is crucial to maintaining the exclusivity and commercial value of content.

[0095] In summary, this embodiment demonstrates a video encryption method implemented by timeline obfuscation and dual-track rendering. By combining advanced video processing technology, the system implements intelligent encryption of video content and provides an innovative solution for copyright protection and secure transmission of multimedia content. This technology has broad application prospects in many fields and provides new possibilities and directions for the development of video technology.

[0096] In one embodiment of the short video obfuscation encryption method of the present application, see Figure 3 , and can also include the following:

[0097] Step S301: performing feature weighted fusion on each of the target high-order feature representations through an attention mechanism to obtain a target feature representation of the video clip to be supplemented, and performing semantic recognition on the target feature representation through a preset knowledge graph inference algorithm;

[0098] Step S302: inputting the result of the semantic recognition into a set autoregressive generation model, and determining the supplementary video subtitles corresponding to the to-be-supplemented video segment according to the output of the autoregressive generation model.

[0099] Optionally, in this embodiment, the system first uses the attention mechanism to perform feature weighted fusion on the target high-order feature representation. The attention mechanism is a deep learning technology that can effectively identify and emphasize the most important parts of the data. By assigning different weights to each high-order feature, the system can highlight the features that are most critical to the semantic understanding of the video. This weighted fusion process can integrate multi-dimensional feature information and generate a comprehensive target feature representation of the video segment to be supplemented.

[0100] Next, the system uses a preset knowledge graph reasoning algorithm to perform semantic recognition on the target feature representation. The knowledge graph is a semantic network that contains a large number of conceptual entities and their relationships. The system uses these known semantic relationships to infer the specific semantic content in the video clip. Through this reasoning, the system is able to match the feature representation with the concepts in the knowledge graph to identify the semantic information involved in the video clip. This process not only improves the accuracy of semantic recognition, but also provides the necessary semantic basis for subsequent subtitle generation.

[0101] In step S302, the system inputs the result of semantic recognition into the set autoregressive generative model. The autoregressive generative model is a deep learning model for sequence generation, which realizes text generation by gradually predicting the next element in the sequence. Here, the model uses the semantic recognition result as a conditional input to generate supplementary subtitles that match the video content. This process ensures that the generated subtitle content is logically coherent and natural by utilizing contextual information.

[0102] This technical process solves the problem of lack of semantic depth in traditional subtitle generation. By combining attention mechanism, knowledge graph reasoning and autoregressive generation model, the system can not only generate accurate subtitles, but also ensure the semantic consistency of subtitles with video content. This provides users with a more vivid and accurate viewing experience, especially in teaching and demonstration videos that need to convey information accurately.

[0103] In practical applications, this technology can significantly improve the efficiency and quality of subtitle generation. For example, in film production, the system can quickly identify and supplement missing subtitles, reducing the workload of manual editing. In educational videos, teachers can use the system to automatically generate accurate subtitles to help students better understand complex knowledge points.

[0104] In addition, this technology can also be used for multilingual subtitle generation. By combining knowledge graphs in different languages, the system can perform semantic matching and text generation between different languages, providing multilingual support for global users. This multilingual support is particularly important for multinational companies and international education projects.

[0105] In summary, this embodiment demonstrates an intelligent subtitle generation method that combines attention mechanism, knowledge graph reasoning and autoregressive text generation. Through the application of multi-level feature analysis and generation models, the system realizes intelligent supplementation of video subtitles and provides an innovative solution for multimedia content processing.

[0106] In one embodiment of the short video obfuscation encryption method of the present application, see Figure 4 , and can also include the following:

[0107] Step S401: using a pre-trained speech recognition model to transcribe the audio content in the video clip, extracting corresponding original video subtitles, and performing a rationality check on the extracted original video subtitles;

[0108] Step S402: Determine the video segment to be supplemented with subtitles according to the extracted original video subtitles, video duration, and picture change characteristics.

[0109] Optionally, in this embodiment, the system first uses a pre-trained speech recognition model to transcribe the audio content in the video clip and extract the corresponding original video subtitles from it. The pre-trained speech recognition model is usually based on deep learning technology and is trained with data from a large corpus, and can efficiently recognize multiple languages ​​and accents. Through this process, the system can quickly and accurately convert the audio content into subtitles in text form.

[0110] After subtitle extraction, the system performs a plausibility check on these raw subtitles. The purpose of this step is to ensure the accuracy and completeness of the subtitle content. Plausibility checks can be performed in a variety of ways, including spelling checks, grammatical analysis, and comparative analysis with audio content. The system uses natural language processing technology to identify and correct errors in subtitles to make them more consistent with semantic and grammatical specifications.

[0111] In step S402, the system determines the video segments to be supplemented with subtitles based on the extracted original video subtitles, video duration, and screen change characteristics. Video duration and screen change characteristics are important analysis factors. Video duration is used to measure the time proportion covered by subtitles, while screen change characteristics help identify the matching degree between subtitles and screen content.

[0112] The system analyzes the logical relationship between subtitles and the screen, as well as the distribution of subtitles in the video, to identify segments where subtitles are missing or incomplete. This process may involve semantic understanding of the video content and scene recognition to ensure that the subtitles accurately reflect the actual content of the video.

[0113] The process of determining the video segments to be supplemented solves the common problems of missing subtitles and inaccurate subtitles in video subtitle processing. Through systematic analysis and recognition, the system can automatically find the parts that need to be supplemented with subtitles, reducing the workload and errors of manual detection.

[0114] In practical applications, this technology is of great significance for improving the accuracy and completeness of video subtitles. For example, in film and television production, producers can use the system to quickly identify missing parts of subtitles to ensure that the content released meets the expectations of the audience. In the field of online education, teachers can use the system to generate comprehensive subtitles for course videos to help students better understand and learn.

[0115] In addition, this technology can also be applied to multilingual subtitle generation. By combining speech recognition models in different languages, the system is able to extract and supplement subtitles between different languages, providing multilingual support for international users. This multilingual capability is particularly important for multinational companies and international education projects, helping to expand the global audience of video content.

[0116] In summary, this embodiment demonstrates an intelligent subtitle generation method that combines speech recognition, natural language processing, and video content analysis. Through multi-level analysis and processing, the system realizes the automatic extraction and supplementation of video subtitles, providing an innovative solution for the production and release of multimedia content.

[0117] In one embodiment of the short video obfuscation encryption method of the present application, see Figure 5 , and can also include the following:

[0118] Step S501: performing a frame division operation on the video segment to be supplemented that is determined to need to supplement subtitles, dividing it into a continuous video frame sequence, and extracting spatiotemporal features including at least one of color, texture and motion information for each video frame in the video frame sequence;

[0119] Step S502: compressing the extracted spatiotemporal features into a compact low-order feature representation, wherein the low-order feature representation at least includes basic visual and dynamic characteristics of video content.

[0120] Optionally, in this embodiment, the system first performs a frame division operation on the video segment to be supplemented with subtitles, and divides it into a continuous video frame sequence. The purpose of this step is to disassemble the dynamic video content into a static image sequence to facilitate subsequent feature extraction. The frame division operation is usually implemented through video decoding technology, extracting each frame from the video stream to form a set of images arranged in time order.

[0121] Next, for each video frame, the system extracts spatiotemporal features including color, texture, and motion information. Color features are usually obtained through color histograms or color space conversion (such as RGB to HSV) to describe the color distribution of pixels in the frame. Texture features are extracted using texture analysis algorithms (such as grayscale co-occurrence matrices or Gabor filters) to help identify repetitive patterns and structural information in the image. The extraction of motion information relies on methods such as optical flow analysis or frame difference to capture the movement and changes of objects in the video.

[0122] In step S502, the system compresses the extracted spatiotemporal features into a compact low-order feature representation. This process usually involves feature dimensionality reduction techniques such as principal component analysis (PCA) or t-SNE. The low-order feature representation retains the basic visual and dynamic characteristics of the video content while reducing the dimensionality and complexity of the feature data, making subsequent data processing and analysis more efficient.

[0123] This feature compression method solves the computational burden problem in video processing. By reducing the feature dimension, the system can improve data processing speed and storage efficiency without losing important information. This is especially important for real-time video analysis and large-scale video data processing.

[0124] In practical applications, this technology can be used in video content analysis, automatic subtitle generation, and video retrieval. For example, in video retrieval, the system can quickly compare low-level feature representations of different video clips to achieve fast video matching and retrieval. In automatic subtitle generation, these feature representations provide basic data for semantic analysis and help identify important scenes and dialogues in the video.

[0125] In addition, this technology can also be applied to video compression and transmission. By representing video content as a compact feature vector, the system can reduce the size of video files and improve transmission efficiency. This capability is particularly important for video playback in network environments with limited bandwidth and on mobile devices.

[0126] In summary, this embodiment demonstrates a video analysis method implemented by frame segmentation and feature compression. Through multi-level feature extraction and dimensionality reduction technology, the system achieves efficient representation and processing of video content, providing an innovative solution for the development of multimedia technology.

[0127] In one embodiment of the short video obfuscation encryption method of the present application, see Figure 6 , and can also include the following:

[0128] Step S601: Integrate the extracted low-level feature representations into a unified feature set, wherein the feature set contains rich visual information of the video clip;

[0129] Step S602: Input the feature set into a set deep learning multi-layer perceptron model for multi-level feature transformation and combination, and obtain a high-order feature representation describing the video content output by the deep learning multi-layer perceptron model.

[0130] Optionally, in this embodiment, the system first integrates the extracted low-level feature representations into a unified feature set. The low-level feature representations include information such as color, texture, and motion, which are extracted from video frames to describe the basic visual and dynamic characteristics of the video clip. The integration process is achieved through feature fusion technology, which combines features from different sources and types into a comprehensive feature set to ensure that it can fully reflect the visual information of the video clip.

[0131] This feature set plays a key role in video analysis. By integrating the rich visual information in the video clips, the system can provide a detailed data foundation for subsequent deep learning processing. This integration not only improves the integrity of the data, but also lays the foundation for complex pattern recognition and semantic analysis.

[0132] Next, the system inputs the integrated feature set into the set deep learning multi-layer perceptron (MLP) model. MLP is a commonly used neural network structure that is suitable for multi-level feature transformation and combination. Through the nonlinear activation and connection of multiple layers of neurons, MLP can establish complex mapping relationships between input features, thereby extracting more abstract and advanced feature representations.

[0133] During the model processing, the input low-level features are transformed layer by layer by the multi-layer perceptron, and each layer is processed by weighted summation and activation function to combine the features nonlinearly. This multi-level feature transformation can effectively capture the deep patterns and semantic relationships in the video content, and finally generate high-level feature representations. These high-level feature representations not only contain the basic visual characteristics of the video, but also reflect higher-level semantic information, such as scene structure, object relationships, and action dynamics.

[0134] Through this processing method, the system solves the key problem in video analysis: how to extract representative and recognizable high-level features from a large number of features. Compared with traditional methods, deep learning technology can autonomously learn and optimize the feature extraction process without relying on artificial rules, thereby improving the accuracy and efficiency of analysis.

[0135] In practical applications, this technology has broad application prospects in the fields of video content recognition, classification and retrieval. For example, in video surveillance, the system can analyze video streams in real time, identify abnormal behaviors and events, and improve the response speed of security monitoring. In film and television production, automated video content analysis can help editors quickly locate and classify materials, improving production efficiency.

[0136] In addition, this technology can also be used in augmented reality and virtual reality applications. Through the deep understanding of video content, the system can achieve a more natural and intelligent interactive experience. This is of great significance for improving user experience and promoting the development of emerging technologies.

[0137] In summary, this embodiment demonstrates a method for video content analysis through feature fusion and deep learning. By integrating low-order features and multi-level processing of deep neural networks, the system achieves efficient recognition and analysis of video content, providing an innovative solution for the development of multimedia technology. This technology not only improves the intelligent level of video processing, but also provides a new development direction for future multimedia applications.

[0138] In one embodiment of the short video obfuscation encryption method of the present application, see Figure 7 , and can also include the following:

[0139] Step S701: Using the pre-built knowledge graph, the obtained high-order feature representation is semantically matched and associated with the concept entities in the graph to determine the semantic concepts of product functions, attributes, and scenarios corresponding to the high-order feature representation;

[0140] Step S702: Input the identified semantic concept as a condition into a pre-trained autoregressive text generation model to obtain a supplementary video subtitle output by the autoregressive text generation model that matches the semantic concept.

[0141] Optionally, in this embodiment, a pre-built knowledge graph is first used to semantically match and associate high-order feature representations with conceptual entities in the graph. A knowledge graph is a structured semantic network that contains a large number of entities and their relationships. By comparing high-order features with concepts in the graph, the system can identify the association between features and product functions, attributes, and scenarios. This process uses a semantic matching algorithm to reason about the information in the graph to determine the specific semantic concepts involved in the video clip.

[0142] The key to semantic matching is to make full use of the relationships and contextual information in the graph. By identifying the similarities and associations between features, the system can accurately map the video content to the corresponding semantic concepts. This approach not only improves the accuracy of recognition, but also provides a deeper understanding of the video content, ensuring that the extracted concepts are consistent with the actual scenes and functions.

[0143] In step S702, the identified semantic concepts are input as conditions into a pre-trained autoregressive text generation model. An autoregressive model is a deep learning model used to generate sequence data, which can generate corresponding text outputs based on input conditions. In this example, the model generates supplementary video subtitles that match the identified semantic concepts by gradually predicting the next word.

[0144] This text generation process relies on a large amount of known data for training, which enables the model to generate natural and fluent text in different contexts. When generating subtitles, the autoregressive model can combine semantic concepts and video context information to ensure that the subtitle content is not only accurate but also coherent.

[0145] This technology solves the problem of semantic inconsistency in the process of automatically generating subtitles. By combining knowledge graphs and autoregressive models, the system is able to generate subtitles that are highly matched to the video content, reducing the workload of manual editing and improving the efficiency and accuracy of subtitle generation.

[0146] In practical applications, this method can be used in a variety of scenarios. For example, in film and television production, the system can automatically generate subtitles for videos to reduce manual intervention. In educational videos, automatically generated subtitles can help students better understand the course content and improve learning outcomes.

[0147] In addition, this technology can also be used to generate multilingual subtitles. By performing semantic matching in knowledge graphs in different languages, the system can generate corresponding subtitle texts between different languages, providing multilingual support for global users. This is particularly important for multinational companies and international education projects, and can significantly expand the audience range of video content.

[0148] In summary, this embodiment demonstrates an intelligent subtitle generation method that combines knowledge graph and autoregressive text generation model. Through multi-level semantic analysis and text generation, the system realizes the automatic supplement of video subtitles and provides an innovative solution for the production and release of multimedia content. This technology has broad application prospects in many fields and provides new possibilities and directions for the development of video technology.

[0149] In order to accurately and efficiently process short video subtitles, the present application provides an embodiment of a short video obfuscation encryption device for implementing all or part of the content of the short video obfuscation encryption method, see Figure 8 The short video obfuscation encryption device specifically includes the following contents:

[0150] The video segment determination module 10 is used to receive a short video file to be encrypted uploaded by a user, and to segment the short video file to be encrypted by using a preset computer vision algorithm to obtain a plurality of video segments;

[0151] The video subtitle determination module 20 is used to extract subtitles from the video segment by a preset speech recognition algorithm to obtain corresponding original video subtitles, and determine the video segment to be supplemented that lacks corresponding video subtitles, extract the spatiotemporal features of the video segment to be supplemented by a frame division operation, obtain a low-order feature representation according to the spatiotemporal features, unify the extracted low-order feature representation into a feature set and input it into a set deep learning multi-layer perceptron to obtain a corresponding high-order feature representation, and determine the supplementary video subtitles corresponding to the video segment to be supplemented according to the high-order feature representation and a set autoregressive generation model;

[0152] The video obfuscation encryption module 30 is used to perform semantic clustering on the original video subtitles and the supplementary video subtitles, determine the timeline of each of the video clips, set each of the video clips to play on the lower main track according to the timeline, and set each of the corresponding video clips to play on the upper sub-track opaquely according to the obfuscated timeline, and obtain an encrypted short video file based on the video clips set to play on the lower main track and the video clips set to play on the upper sub-track opaquely.

[0153] From the above description, it can be seen that the short video obfuscation encryption device provided in the embodiment of the present application can determine the timeline of each video clip by semantically clustering the original video subtitles and the supplementary video subtitles, set each video clip to play on the lower main track according to the timeline, and set each corresponding video clip to play on the upper sub-track opaquely according to the obfuscated timeline, and obtain the encrypted short video file based on the video clips set to play on the lower main track and the video clips set to play on the upper sub-track opaquely, thereby accurately and efficiently processing short video subtitles.

[0154] From the hardware level, in order to accurately and efficiently process short video subtitles, the present application provides an embodiment of an electronic device for implementing all or part of the content in the short video obfuscation encryption method, and the electronic device specifically includes the following content:

[0155] Processor, memory, communication interface and bus; wherein the processor, memory and communication interface communicate with each other through the bus; the communication interface is used to realize information transmission between the short video obfuscation encryption device and the core business system, user terminal and related databases and other related devices; the logic controller can be a desktop computer, a tablet computer and a mobile terminal, etc., but the present embodiment is not limited thereto. In the present embodiment, the logic controller can be implemented with reference to the embodiment of the short video obfuscation encryption method and the embodiment of the short video obfuscation encryption device in the embodiment, and the contents thereof are incorporated herein, and the repeated parts are not repeated.

[0156] It is understandable that the user terminal may include a smart phone, a tablet electronic device, a network set-top box, a portable computer, a desktop computer, a personal digital assistant (PDA), a vehicle-mounted device, a smart wearable device, etc. Among them, the smart wearable device may include smart glasses, a smart watch, a smart bracelet, etc.

[0157] In practical applications, part of the short video obfuscation encryption method can be executed on the electronic device side as described above, or all operations can be completed in the client device. The specific selection can be based on the processing capability of the client device and the limitations of the user's usage scenario. This application does not limit this. If all operations are completed in the client device, the client device may also include a processor.

[0158] The client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission with the server. The server may include a server on the task scheduling center side, and other implementation scenarios may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, or a server cluster consisting of multiple servers, or a server structure of a distributed device.

[0159] Fig. 9 FIG. 9 is a schematic block diagram of the system structure of the electronic device 9600 according to an embodiment of the present application. Fig. 9As shown, the electronic device 9600 may include a central processor 9100 and a memory 9140; the memory 9140 is coupled to the central processor 9100. It is worth noting that Fig. 9 is exemplary; other types of structures may also be used to supplement or replace this structure to implement telecommunication functions or other functions.

[0160] In one embodiment, the short video obfuscation encryption method function may be integrated into the central processing unit 9100. The central processing unit 9100 may be configured to perform the following control:

[0161] Step S101: receiving a short video file to be encrypted uploaded by a user, and dividing the short video file to be encrypted into segments using a preset computer vision algorithm to obtain multiple video clips;

[0162] Step S102: extracting subtitles from the video clip by a preset speech recognition algorithm to obtain corresponding original video subtitles, and determining the video clip to be supplemented that lacks corresponding video subtitles, extracting spatiotemporal features of the video clip to be supplemented by a frame division operation, obtaining a low-order feature representation based on the spatiotemporal features, unifying the extracted low-order feature representation into a feature set and inputting it into a set deep learning multi-layer perceptron to obtain a corresponding high-order feature representation, and determining the supplementary video subtitles corresponding to the video clip to be supplemented based on the high-order feature representation and a set autoregressive generation model;

[0163] Step S103: semantically clustering the original video subtitles and the supplementary video subtitles, determining the timeline of each of the video clips, setting each of the video clips for lower-layer main track playback according to the timeline, and setting each of the corresponding video clips for upper-layer sub-track opaque playback according to the obfuscated timeline, and obtaining an encrypted short video file based on the video clips set for lower-layer main track playback and the video clips set for upper-layer sub-track opaque playback.

[0164] From the above description, it can be seen that the electronic device provided by the embodiment of the present application determines the timeline of each video clip by semantically clustering the original video subtitles and the supplementary video subtitles, sets each video clip to be played on the lower main track according to the timeline, and sets each corresponding video clip to be played on the upper sub-track opaquely according to the timeline after the obfuscation process, and obtains an encrypted short video file based on the video clips set to be played on the lower main track and the video clips set to be played on the upper sub-track opaquely, thereby accurately and efficiently processing short video subtitles.

[0165] In another embodiment, the short video obfuscation encryption device can be configured separately from the central processing unit 9100. For example, the short video obfuscation encryption device can be configured as a chip connected to the central processing unit 9100, and the short video obfuscation encryption method function can be implemented through the control of the central processing unit.

[0166] like Fig. 9 As shown, the electronic device 9600 may also include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily have to include Fig. 9 In addition, the electronic device 9600 may also include Fig. 9 For components not shown, reference may be made to the prior art.

[0167] like Fig. 9 As shown, the central processing unit 9100 is sometimes also referred to as a controller or an operation control, and may include a microprocessor or other processor device and / or logic device. The central processing unit 9100 receives input and controls the operation of various components of the electronic device 9600.

[0168] The memory 9140 may be, for example, one or more of a cache, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory or other suitable devices. The above-mentioned information related to the failure may be stored, and a program for executing the relevant information may also be stored. The CPU 9100 may execute the program stored in the memory 9140 to implement information storage or processing, etc.

[0169] The input unit 9120 provides input to the central processing unit 9100. The input unit 9120 is, for example, a key or a touch input device. The power supply 9170 is used to provide power to the electronic device 9600. The display 9160 is used to display display objects such as images and texts. The display may be, for example, an LCD display, but is not limited thereto.

[0170] The memory 9140 may be a solid-state memory, such as a read-only memory (ROM), a random access memory (RAM), a SIM card, etc. It may also be a memory that saves information even when the power is off, can be selectively erased, and is provided with more data, examples of which are sometimes referred to as EPROMs, etc. The memory 9140 may also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142, which is used to store application programs and function programs or processes for executing the operation of the electronic device 9600 through the central processor 9100.

[0171] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various drivers for communication functions of the electronic device and / or for executing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0172] The communication module 9110 is a transmitter / receiver that sends and receives signals via the antenna 9111. The communication module 9110 (transmitter / receiver) is coupled to the central processor 9100 to provide input signals and receive output signals, which may be the same as the case of a conventional mobile communication terminal.

[0173] Based on different communication technologies, multiple communication modules 9110 may be provided in the same electronic device, such as a cellular network module, a Bluetooth module and / or a wireless LAN module, etc. The communication module 9110 (transmitter / receiver) is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide an audio output via the speaker 9131 and receive an audio input from the microphone 9132, thereby realizing a common telecommunication function. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. In addition, the audio processor 9130 is also coupled to the central processor 9100, so that recording can be performed on the local machine through the microphone 9132, and the sound stored on the local machine can be played through the speaker 9131.

[0174] The embodiments of the present application also provide a computer-readable storage medium capable of implementing all the steps of the short video obfuscation encryption method in the above embodiment, where the execution subject is a server or a client. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, all the steps of the short video obfuscation encryption method in the above embodiment are implemented, where the execution subject is a server or a client. For example, when the processor executes the computer program, the following steps are implemented:

[0175] Step S101: receiving a short video file to be encrypted uploaded by a user, and dividing the short video file to be encrypted into segments using a preset computer vision algorithm to obtain multiple video clips;

[0176] Step S102: extracting subtitles from the video clip by a preset speech recognition algorithm to obtain corresponding original video subtitles, and determining the video clip to be supplemented that lacks corresponding video subtitles, extracting spatiotemporal features of the video clip to be supplemented by a frame division operation, obtaining a low-order feature representation based on the spatiotemporal features, unifying the extracted low-order feature representation into a feature set and inputting it into a set deep learning multi-layer perceptron to obtain a corresponding high-order feature representation, and determining the supplementary video subtitles corresponding to the video clip to be supplemented based on the high-order feature representation and a set autoregressive generation model;

[0177] Step S103: semantically clustering the original video subtitles and the supplementary video subtitles, determining the timeline of each of the video clips, setting each of the video clips for lower-layer main track playback according to the timeline, and setting each of the corresponding video clips for upper-layer sub-track opaque playback according to the obfuscated timeline, and obtaining an encrypted short video file based on the video clips set for lower-layer main track playback and the video clips set for upper-layer sub-track opaque playback.

[0178] From the above description, it can be seen that the computer-readable storage medium provided in the embodiment of the present application determines the timeline of each video clip by semantically clustering the original video subtitles and the supplementary video subtitles, sets each video clip to be played on the lower main track according to the timeline, and sets each corresponding video clip to be played on the upper sub-track opaquely according to the timeline after the obfuscation process, and obtains an encrypted short video file based on the video clips set to be played on the lower main track and the video clips set to be played on the upper sub-track opaquely, thereby enabling accurate and efficient short video subtitle processing.

[0179] The embodiments of the present application also provide a computer program product capable of implementing all the steps of the short video obfuscation encryption method in the above embodiments, where the execution subject is a server or a client. When the computer program / instruction is executed by a processor, the steps of the short video obfuscation encryption method are implemented. For example, the computer program / instruction implements the following steps:

[0180] Step S101: receiving a short video file to be encrypted uploaded by a user, and dividing the short video file to be encrypted into segments using a preset computer vision algorithm to obtain multiple video clips;

[0181] Step S102: extracting subtitles from the video clip by a preset speech recognition algorithm to obtain corresponding original video subtitles, and determining the video clip to be supplemented that lacks corresponding video subtitles, extracting spatiotemporal features of the video clip to be supplemented by a frame division operation, obtaining a low-order feature representation based on the spatiotemporal features, unifying the extracted low-order feature representation into a feature set and inputting it into a set deep learning multi-layer perceptron to obtain a corresponding high-order feature representation, and determining the supplementary video subtitles corresponding to the video clip to be supplemented based on the high-order feature representation and a set autoregressive generation model;

[0182] Step S103: semantically clustering the original video subtitles and the supplementary video subtitles, determining the timeline of each of the video clips, setting each of the video clips for lower-layer main track playback according to the timeline, and setting each of the corresponding video clips for upper-layer sub-track opaque playback according to the obfuscated timeline, and obtaining an encrypted short video file based on the video clips set for lower-layer main track playback and the video clips set for upper-layer sub-track opaque playback.

[0183] From the above description, it can be seen that the computer program product provided by the embodiment of the present application determines the timeline of each video clip by semantically clustering the original video subtitles and the supplementary video subtitles, sets each video clip to be played on the lower main track according to the timeline, and sets each corresponding video clip to be played on the upper sub-track opaquely according to the timeline after the obfuscation process, and obtains an encrypted short video file based on the video clips set to be played on the lower main track and the video clips set to be played on the upper sub-track opaquely, thereby accurately and efficiently processing short video subtitles.

[0184] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0185] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (apparatus), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0186] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0187] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0188] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A short video obfuscation encryption method, characterized in that: The method comprises: Receive a short video file to be encrypted uploaded by a user, and divide the short video file to be encrypted into multiple video clips by using a preset computer vision algorithm; Extracting subtitles from the video clips by using a preset speech recognition algorithm to obtain corresponding original video subtitles, and determining the video clips to be supplemented that lack corresponding video subtitles, extracting spatiotemporal features of the video clips to be supplemented by a frame division operation, obtaining low-order feature representations based on the spatiotemporal features, unifying the extracted low-order feature representations into a feature set and inputting them into a set deep learning multi-layer perceptron to obtain corresponding high-order feature representations, and determining the supplementary video subtitles corresponding to the video clips to be supplemented based on the high-order feature representations and a set autoregressive generation model; Semantic clustering is performed on the original video subtitles and the supplementary video subtitles to determine the timeline of each of the video clips, and each of the video clips is set to be played on a lower main track according to the timeline, and each of the corresponding video clips is set to be played on an upper sub-track opaquely according to the timeline after the obfuscation process, and an encrypted short video file is obtained based on the video clips set to be played on the lower main track and the video clips set to be played on the upper sub-track opaquely.

2. The short video obfuscation encryption method according to claim 1, characterized in that: The method of setting the playback of each video clip on the lower main track according to the timeline, and setting the playback of each video clip on the upper sub-track opaquely according to the obfuscated timeline, and obtaining an encrypted short video file according to the video clips set to be played on the lower main track and the video clips set to be played on the upper sub-track opaquely, includes: According to the time axis of the video clips, each video clip is set to be played as the main track of the lower layer, and the video clips are randomly obfuscated on the time axis, and the obfuscated video clips are set to be played as the sub-track of the upper layer in an opaque manner; The video clip played on the set lower main track and the obfuscated video clip played opaquely on the upper sub-track are fused and rendered to obtain an encrypted short video file.

3. The short video obfuscation encryption method according to claim 1, characterized in that: The step of determining the supplementary video subtitles corresponding to the to-be-supplemented video segment according to the high-order feature representation and setting the autoregressive generation model includes: Performing feature weighted fusion on each of the target high-order feature representations through an attention mechanism to obtain the target feature representation of the video clip to be supplemented, and performing semantic recognition on the target feature representation through a preset knowledge graph inference algorithm; The result of the semantic recognition is input into a set autoregressive generation model, and the supplementary video subtitles corresponding to the to-be-supplemented video segment are determined according to the output of the autoregressive generation model.

4. The short video obfuscation encryption method according to claim 1, characterized in that: The extracting subtitles from the video clips by using a preset speech recognition algorithm to obtain corresponding original video subtitles and determining the video clips to be supplemented that lack corresponding video subtitles, includes: Using a pre-trained speech recognition model to transcribe the audio content in the video clip, extracting corresponding original video subtitles, and performing a rationality check on the extracted original video subtitles; The video segments to be supplemented that need to be supplemented with subtitles are determined based on the extracted original video subtitles, video duration, and picture change characteristics.

5. The short video obfuscation encryption method according to claim 1, characterized in that: The step of extracting the spatiotemporal features of the video segment to be supplemented by a frame division operation and obtaining a low-order feature representation according to the spatiotemporal features includes: Performing a frame division operation on the video segment to be supplemented that is determined to need to be supplemented with subtitles, dividing it into a continuous video frame sequence, and extracting spatiotemporal features including at least one of color, texture, and motion information for each video frame in the video frame sequence; The extracted spatiotemporal features are compressed into a compact low-level feature representation, wherein the low-level feature representation includes at least basic visual and dynamic characteristics of video content.

6. The short video obfuscation encryption method according to claim 1, characterized in that: The extracted low-order feature representations are unified into a feature set and input into a set deep learning multi-layer perceptron to obtain corresponding high-order feature representations, including: Integrating the extracted low-level feature representations into a unified feature set, wherein the feature set contains rich visual information of the video clip; The feature set is input into a set deep learning multi-layer perceptron model for multi-level feature transformation and combination, so as to obtain a high-order feature representation describing the video content output by the deep learning multi-layer perceptron model.

7. The short video obfuscation encryption method according to claim 1, characterized in that: The method of performing semantic recognition on the target feature representation by using a preset knowledge graph inference algorithm, inputting the result of the semantic recognition into a set autoregressive generation model, and determining the supplementary video subtitles corresponding to the to-be-supplemented video segment according to the output of the autoregressive generation model, includes: Use the pre-built knowledge graph to semantically match and associate the obtained high-order feature representation with the concept entities in the graph to determine the semantic concepts of product functions, attributes, and scenarios corresponding to the high-order feature representation; The identified semantic concepts are input as conditions into a pre-trained autoregressive text generation model to obtain supplementary video subtitles that match the semantic concepts and are output by the autoregressive text generation model.

8. A short video obfuscation encryption device, characterized in that: The device comprises: A video segment determination module is used to receive a short video file to be encrypted uploaded by a user, and divide the short video file to be encrypted into segments using a preset computer vision algorithm to obtain multiple video segments; A video subtitle determination module is used to extract subtitles from the video segment by a preset speech recognition algorithm to obtain corresponding original video subtitles, and determine the video segment to be supplemented that lacks corresponding video subtitles, extract the spatiotemporal features of the video segment to be supplemented by a frame division operation, obtain a low-order feature representation based on the spatiotemporal features, unify the extracted low-order feature representation into a feature set and input it into a set deep learning multi-layer perceptron to obtain a corresponding high-order feature representation, and determine the supplementary video subtitles corresponding to the video segment to be supplemented based on the high-order feature representation and a set autoregressive generation model; The video obfuscation encryption module is used to perform semantic clustering on the original video subtitles and the supplementary video subtitles, determine the timeline of each of the video clips, set each of the video clips to play on the lower main track according to the timeline, and set each of the corresponding video clips to play on the upper sub-track opaquely according to the obfuscated timeline, and obtain an encrypted short video file based on the video clips set to play on the lower main track and the video clips set to play on the upper sub-track opaquely.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the short video obfuscation encryption method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the short video obfuscation encryption method described in any one of claims 1 to 7 are implemented.

Images