Video encryption method, decryption method and device based on large model

Abstract

The invention provides a video encryption method, decryption method and device based on a large model, and relates to the technical field of artificial intelligence, in particular to the technical fields of encryption transmission, video processing, large models and the like. According to the specific implementation scheme, video features are extracted from a target video; inputting the video feature and the seed value into a multi-modal large model to generate a key; and encrypting the target video by using the key to obtain an encrypted video.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to technical fields such as encrypted transmission, video processing, and large models. Background Art

[0002] Video encryption methods usually rely on classic encryption algorithms. The encryption key can be used to encrypt the video at the encryption end, and the decryption key can be used to decrypt it at the decryption end. Strict security measures are required to generate, store, and transmit keys. Key leakage may lead to illegal access to video and other data. Summary of the invention

[0003] The present invention provides a video encryption method, a decryption method and a device based on a large model.

[0004] According to one aspect of the present disclosure, a large model-based video encryption method is provided, comprising:

[0005] Extracting video features from the target video;

[0006] Inputting the video feature and the seed value into a multimodal large model to generate a key;

[0007] The target video is encrypted using the key to obtain an encrypted video.

[0008] According to another aspect of the present disclosure, a large model-based video decryption method is provided, comprising:

[0009] Obtain video features and encrypted video of the target video;

[0010] Inputting the video feature and the seed value into a multimodal large model to generate a key;

[0011] The encrypted video is decrypted using the key to obtain the target video.

[0012] According to another aspect of the present disclosure, a large model-based video encryption device is provided, comprising:

[0013] An extraction module, used to extract video features from the target video;

[0014] A first input module, used for inputting the video feature and the seed value into the multimodal macro model to generate a key;

[0015] The encryption module is used to encrypt the target video using the key to obtain an encrypted video.

[0016] According to another aspect of the present disclosure, a large model-based video decryption device is provided, comprising:

[0017] An acquisition module, used to acquire video features and encrypted video of a target video;

[0018] A second input module is used to input the video feature and the seed value into the multimodal large model to generate a key;

[0019] The decryption module is used to use the key to decrypt the encrypted video to obtain the target video.

[0020] According to another aspect of the present disclosure, there is provided an electronic device, comprising:

[0021] at least one processor; and

[0022] a memory communicatively connected to the at least one processor; wherein,

[0023] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute any method in the embodiments of the present disclosure.

[0024] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute any method according to the embodiments of the present disclosure.

[0025] According to another aspect of the present disclosure, a computer program product is provided, including a computer program, which implements any method according to the embodiments of the present disclosure when executed by a processor.

[0026] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0028] Figure 1 is a flowchart of a video encryption method based on a large model according to an embodiment of the present disclosure;

[0029] Figure 2 is a flowchart of a video encryption method based on a large model according to another embodiment of the present disclosure;

[0030] Figure 3 is a flowchart of a video encryption method based on a large model according to another embodiment of the present disclosure;

[0031] Figure 4 is a flowchart of a video encryption method based on a large model according to another embodiment of the present disclosure;

[0032] Figure 5 is a flowchart of a video encryption method based on a large model according to another embodiment of the present disclosure;

[0033] Figure 6 is a flowchart of a video encryption method based on a large model according to another embodiment of the present disclosure;

[0034] Figure 7 is a flowchart of a video encryption method based on a large model according to another embodiment of the present disclosure;

[0035] Figure 8 is a flowchart of a video decryption method based on a large model according to an embodiment of the present disclosure;

[0036] Fig. 9 is a flowchart of a video decryption method based on a large model according to another embodiment of the present disclosure;

[0037] Fig.10 is a flowchart of a video decryption method based on a large model according to another embodiment of the present disclosure;

[0038] Fig.11 is a flowchart of a video decryption method based on a large model according to another embodiment of the present disclosure;

[0039] Fig.12 is a flowchart of a video decryption method based on a large model according to another embodiment of the present disclosure;

[0040] Fig.13 is a flowchart of a video decryption method based on a large model according to another embodiment of the present disclosure;

[0041] Fig.14 is a flowchart of a video decryption method based on a large model according to another embodiment of the present disclosure;

[0042] Fig.15 is a flowchart of an encryption process according to an embodiment of the present disclosure;

[0043] Fig.16 is a flowchart of a decryption process according to an embodiment of the present disclosure;

[0044] Fig.17 is a structural diagram of a video encryption device based on a large model according to an embodiment of the present disclosure;

[0045] Fig.18 is a structural schematic diagram of a video encryption device based on a large model according to another embodiment of the present disclosure;

[0046] Fig.19is a structural diagram of a video decryption device based on a large model according to an embodiment of the present disclosure;

[0047] Fig. 20 is a structural schematic diagram of a video decryption device based on a large model according to another embodiment of the present disclosure;

[0048] Fig.21 is a block diagram of an electronic device for implementing an embodiment of the present disclosure. DETAILED DESCRIPTION

[0049] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope of the present disclosure. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0050] Video encryption methods based on classic encryption algorithms include:

[0051] Advanced Encryption Standard (AES) is a symmetric encryption algorithm that can be used for video data, that is, encryption and decryption use the same key, which has high security and efficiency.

[0052] Rivest-Shamir-Adleman (RSA) is an asymmetric encryption algorithm, that is, encryption and decryption use different keys, called public key and private key respectively, but the computational complexity is relatively high.

[0053] The above encryption and decryption methods use fixed mathematical algorithms and preset key management strategies in the key generation and encryption process, which have certain limitations in key management and algorithm fixity. Specifically:

[0054] Complex key management: Strict security measures are required to generate, store, and transmit keys. Any key leakage may lead to illegal access to data. An additional key management system is required to ensure the security of the keys.

[0055] Fixed encryption algorithms: Many encryption algorithms use fixed encryption processes and key generation methods, which makes the encryption scheme easy to crack, especially when the key is reused. The encryption algorithm and key generation are pre-set and vulnerable to pattern attacks or repeated keys.

[0056] Lack of flexibility: Traditional algorithms lack flexibility when dealing with complex encryption scenarios, especially in application scenarios that require high security.

[0057] Figure 1 1 is a flow chart of a large model-based video encryption method 100 according to an embodiment of the present disclosure, the method comprising:

[0058] S101, extracting video features from the target video;

[0059] S102, inputting the video feature and the seed value into a multimodal large model to generate a key;

[0060] S103: Use the key to encrypt the target video to obtain an encrypted video.

[0061] In the disclosed embodiment, the target video is a video that needs to be encrypted for transmission. A neural network model can be used to extract video features from the target video. The seed value can also be called a key seed, and the seed value can be used to generate an encryption key at the encryption end and a decryption key at the decryption end. The encryption end can generate an encryption key based on video features and a seed value. The seed value can be a fixed value or dynamically generated. The encryption end can send the seed value to the decryption end, or send information used to dynamically generate the seed value to the decryption end. If it is symmetric encryption, the encryption key and the decryption key can be the same. The multimodal models deployed at the encryption end and the decryption end can be the same.

[0062] In the disclosed embodiment, the multimodal big model can process multimodal data. Multimodal data may include: a data set consisting of two or more different types of data, which may come from different information sources or perception methods, such as images, text, voice, video, etc. Among them, the multimodal big model can be obtained by fine-tuning the basic big model using multimodal training data. The multimodal training data may include text features, voice features, video features, and annotated output information, etc.

[0063] In the disclosed embodiment, a multimodal large model is used to process multimodal data such as input video features and seed values, and output information for video features and seed values ​​can be dynamically generated. The output information can be used as an encryption key, or the output information can be calculated and processed by a formula, and the calculated value can be used as an encryption key. Since the video features extracted from different videos are usually different, the keys for different videos generated based on video features and seed values ​​using a multimodal large model are usually different.

[0064] In the disclosed embodiment, the key output by the multimodal large model can be used to encrypt the video data at the encryption end, and there can be multiple specific encryption algorithms. For example, symmetric encryption algorithms such as AES, Data Encryption Standard (DES), 3DES, and International Data Encryption Algorithm (IDEA) can be used. After the video is encrypted using a dynamically generated encryption key at the encryption end, an encrypted video can be obtained. Then, the encrypted video can be sent to the decryption end. The encryption end can send the video features and seed values ​​for calculating the encryption key to the decryption end. The decryption end and the encryption end can deploy the same multimodal large model. The decryption end can input the video features and the seed value into the multimodal large model to generate a decryption key. Then, the encrypted video is decrypted using the decryption algorithm and decryption key corresponding to the encryption end to restore the target video.

[0065] According to the embodiments of the present disclosure, a key can be generated based on a seed value and video features using a multimodal large model to encrypt a video. Since it is difficult to obtain the multimodal large model of the encryption end and the decryption end, it is difficult to reversely crack the key, thereby improving the security of the key. Furthermore, a video encrypted using the key cannot be cracked, thereby also improving the security of the video.

[0066] Figure 2 2 is a flow chart of a large model-based video encryption method 200 according to another embodiment of the present disclosure. The method 200 can be used to implement the large model-based video encryption method 100. In one implementation, the method 200 further includes:

[0067] S201: Dynamically generate the seed value according to one or more items of metadata, identification information and time information of the video.

[0068] In the disclosed embodiment, the encryption end can dynamically generate a seed value based on the metadata, identification information, and time information of the video using a seed value generation algorithm. The metadata of the video may include data such as file size, creation time, and encoding format. The identification information of the video may include an identity identifier and a device identifier associated with the video. The time information of the video may include the upload time, production time, and duration of the video. At the encryption end, one or more of the metadata, identification information, and time information of the target video may be obtained first. Since the metadata, identification information, and time information of videos with different contents may be different, and the time information of videos with the same content at different times is also different, the metadata, identification information, and time information of the target video may be processed based on a composite algorithm to obtain a dynamic seed value. For example, the composite algorithm may include multiple calculation formulas, one for converting metadata, one for converting identification information, and one for converting time information. The specific parameters of the formula are not limited, and may be simple addition, subtraction, multiplication, and division formulas, or may be complex calculation formulas. Encryption keys and decryption keys may be generated based on the seed value and the video features of the target video, thereby encrypting and decrypting the target video.

[0069] According to the embodiments of the present disclosure, a seed value can be dynamically generated, thereby dynamically generating an encryption key for a target video, making the encrypted video difficult to crack, thereby improving the security of the video.

[0070] Figure 3 3 is a flow chart of a video encryption method 300 based on a large model according to another embodiment of the present disclosure. The method 300 can be used to implement step S102 in the video encryption method 100 based on a large model. In one embodiment, the method 300 includes: inputting the video feature and the seed value into the multimodal large model to generate a key, and further includes:

[0071] S301, using the multimodal large model to obtain output information based on the video feature, the seed value and the temperature parameter of the multimodal large model; wherein the video feature includes one or more of the video frame, video content information and video description text in the target video;

[0072] S302: Perform a hash operation on the output information to obtain the key.

[0073] In the disclosed embodiment, one or more video frames can be extracted from the target video, and video features can be generated based on the video frames. Video content information may include color distribution, motion vectors, etc. of the video frame, and video features can be generated based on the video content information. The video description text may include a specific description of the video content, which may also be referred to as a video content introduction, a video introduction, etc. Video features can be generated based on the video description text. For example, embedding a video frame or a video description text may be performed to obtain vectorized video features.

[0074] In the disclosed embodiment, the temperature parameter of the model is a hyperparameter used to adjust the diversity and randomness of the model generation results. Generally, the higher the temperature, the more random the generated text; the lower the temperature, the more stable the generated text. On the encryption end, the temperature parameter of the multimodal large model can be set, and prompt words can be generated according to a specific prompt word template based on the received video features and seed values. The prompt words are input into the multimodal large model to generate an encryption key. It is also possible to generate output information first, and then perform a hash operation on the output information to obtain an encryption key. It is also possible to use the seed value and temperature parameter to generate a piece of text, and convert the generated text into an encryption key through a hash function.

[0075] Then, the encryption end can use the encryption key and encryption algorithm such as AES to encrypt the target video to obtain the encrypted video. In addition, the encryption end can send one or more of the encrypted video, video features, seed value and temperature parameters of the multimodal large model to the decryption end directly or through other devices. It is also possible to send one or more of the metadata, identification information and time information of the target video for dynamically calculating the seed value to the decryption end. The seed value is dynamically calculated by the decryption end. Furthermore, the encryption end and the decryption end can also be pre-configured with the same hash algorithm.

[0076] According to the embodiments of the present disclosure, the multimodal large model can obtain output information based on dynamic seed values ​​and different video features, and then generate an encryption key through a hash operation. The generated encryption key is more difficult to crack, thereby improving the security of the encryption key and the video.

[0077] Figure 4 4 is a flow chart of a large model-based video encryption method 400 according to another embodiment of the present disclosure. The method 400 can be used to implement step S103 in the large model-based video encryption method 100. In one embodiment, the method 400 includes: encrypting the target video using the key to obtain an encrypted video, and further includes:

[0078] S401, dividing the target video into blocks to obtain one or more video blocks;

[0079] S402: Use the key to encrypt the one or more video blocks to obtain one or more encrypted video blocks.

[0080] In the disclosed embodiment, the video can be divided into blocks first. The video blocks are smaller than the complete video and can be transmitted in parallel, which is convenient for fast transmission. In addition, the decryption end needs to merge multiple video blocks to obtain the complete video. If any video block fails to be decrypted, it is difficult to obtain the complete video, which is more secure. For example, the video can be divided into blocks according to the video duration, video size, preset number of video blocks, etc. When the video blocks are obtained, the encryption key calculated in the above embodiment can be used to encrypt the video blocks to obtain encrypted video blocks.

[0081] According to the embodiments of the present disclosure, the video can be encrypted in blocks, which speeds up the encryption, reduces the occupation of device computing power by the encryption service, and improves the encryption efficiency; and the file size of the video block is small and the transmission speed is fast, which makes it easy to increase the video transmission speed through parallel transmission and other methods.

[0082] Figure 5 is a flow chart of a method 500 for video encryption based on a large model according to another embodiment of the present disclosure. The method 500 can be used to implement step S103 in the method 100 for video encryption based on a large model. In one embodiment, the method 500 includes: inputting the video feature and the seed value into the multimodal large model to generate a key, and further includes:

[0083] S501. Input the overall video features and seed value of the target video into the multimodal large model to generate the key.

[0084] In the disclosed embodiment, the overall video feature is not required to include features of all video frames of the target video. The overall video feature can be a video feature generated based on one or more video frames of the target video, or a video feature generated based on a video description text of the target video. After obtaining the overall video feature of the target video, the overall video feature and the seed value can be used to generate an overall encryption key, and the overall encryption key can be used to encrypt each video block of the target video to obtain each encrypted video block.

[0085] According to the embodiments of the present disclosure, the same key can be used to encrypt all video blocks of a target video, thereby reducing the computing power occupied during encryption and improving the encryption speed.

[0086] Figure 66 is a flow chart of a method 600 for video encryption based on a large model according to another embodiment of the present disclosure. The method 600 can be used to implement step S103 in the method 100 for video encryption based on a large model. In one embodiment, the method 600 includes: inputting the video feature and the seed value into the multimodal large model to generate a key, and further includes:

[0087] S601. Input the video features and seed values ​​of the one or more video blocks into the multimodal large model respectively to generate keys corresponding to the one or more video blocks; the key corresponding to each video block is used to encrypt the video block.

[0088] In the disclosed embodiment, the video features of each video block can be obtained respectively, and the video features and seed values ​​of each video block can be input into the multimodal large model respectively to generate an encryption key for each video block. For example, based on the video features and seed value of video block V1, the encryption key generated is S1. Based on the video features and seed value of video block V2, the encryption key generated is S2. After the encryption key corresponding to the video block is generated based on the multimodal large model, the encryption key corresponding to the video block is used to encrypt the video block to obtain an encrypted video block.

[0089] According to the embodiments of the present disclosure, different video blocks of a target video may be encrypted using their respective keys, thereby improving the security and transmission rate of the video.

[0090] Figure 7 is a flow chart of a large model-based video encryption method 700 according to another embodiment of the present disclosure. The method 700 can be used to implement the large model-based video encryption method 100. In one implementation, the method 700 further includes:

[0091] S701, using an adversarial algorithm to process the target video to generate false data;

[0092] S702, mixing the target video and the false data to obtain a mixed video;

[0093] S703, using the key to encrypt the target video to obtain an encrypted video, including: using the key to encrypt the mixed video to obtain the encrypted video.

[0094] In the disclosed embodiment, the adversarial algorithm may include a generative adversarial algorithm, or a generative adversarial network (GANs). The implementation of GANs includes letting two networks compete with each other. One of them is a generator network, which can continuously capture data in the training library to generate new samples. The other is a discriminator network, which can judge whether the data provided by the generator is true based on relevant data. The two networks compete with each other during the training process and improve together. The ultimate goal is to enable the generator to generate new data that is difficult to distinguish from real data. False data can be generated based on the adversarial algorithm according to the target video. The false data may include false video frames, false masks, etc. The target video and the false data can be mixed together by a set mixing method to obtain a mixed video. For example, one mixing method can be to splice the false data at the head of the target video. Another mixing method can be to splice the false data at the tail of the target video. Another mixing method can be to insert the false data group into multiple positions in the middle of the target video.

[0095] In the disclosed embodiment, the target video as a whole can be encrypted using the dynamically calculated encryption key according to the encryption method of the above embodiment to obtain an encrypted video. The mixed video can be divided into blocks, and then the target video blocks can be encrypted using the dynamically calculated encryption key according to the encryption method of the above embodiment to obtain each encrypted video block.

[0096] According to the embodiments of the present disclosure, the interference of false data can be used. Even if the decryption is successful at the decryption end, it is difficult to obtain the target video if the false data cannot be removed, thereby further improving the security of the video.

[0097] In one implementation, the information used to generate the seed value also includes location information of the false data in the target video.

[0098] In the disclosed embodiment, the position of the false data in the target video or the interference mode of the false data on the target video can be used as information for generating a seed value, and the seed value can be dynamically generated based on the above information and one or more of the metadata, identification information and time information of the video. The encryption end can also send the above information to the decryption end to dynamically calculate the seed value and the decryption key.

[0099] According to the embodiments of the present disclosure, the influence of false data on the target video can be recorded, so that the decryption end can remove the false data after decryption, so that the decryption end can obtain the correct video.

[0100] In one embodiment, blockchain can be used to record encryption records. For example, the seed value generation process required for video encryption can be implemented through smart contracts. Smart contracts will generate seed values ​​based on specific inputs such as metadata and identity information of the video, and record this process on the blockchain to ensure that the generation of seed values ​​is transparent and cannot be tampered with. Blockchain is also used for key management. For example, blockchain technology can be used to manage and record the generation and use of keys. This can track the use of keys and ensure the security and transparency of key management.

[0101] Figure 8 FIG. 8 is a flow chart of a large model-based video decryption method 800 according to an embodiment of the present disclosure, the method comprising:

[0102] S801, obtaining video features and encrypted video of the target video;

[0103] S802, inputting the video feature and the seed value into a multimodal large model to generate a key;

[0104] S803: Use the key to decrypt the encrypted video to obtain the target video.

[0105] In the embodiments of the present disclosure, the same words as the encryption end have the same meanings as the decryption end, and the relevant description of the embodiments of the decryption end can be referred to. The decryption end can receive the encrypted video from the encryption end or other devices, and can receive the video features of the target video corresponding to the encrypted video from the encryption end or other devices.

[0106] In the disclosed embodiment, the decryption end can generate a decryption key based on the video features and the seed value, and decrypt the encrypted video at the decryption end to obtain the target video. The seed value can be a fixed value or dynamically generated. The encryption end can send the seed value to the decryption end, or send the information used to dynamically generate the seed value to the decryption end. If it is symmetric encryption, the encryption key and the decryption key can be the same. The multimodal models deployed by the encryption end and the decryption end can be the same.

[0107] According to the embodiment of the present disclosure, a corresponding decryption key can be generated based on the same large model as the encryption end to decrypt the encrypted video. Since it is difficult to obtain the multimodal large model of the encryption end and the decryption end, it is difficult to reversely crack the key, thereby improving the security of the key and the video. There is no need to transmit the key between the encryption end and the decryption end, which can prevent the key from being intercepted or cracked, and can improve the security of encryption.

[0108] Fig. 9 is a flow chart of a large model-based video decryption method 900 according to another embodiment of the present disclosure. The method 900 can be used to implement the large model-based video decryption method 800. In one implementation, the method further includes:

[0109] S901: Dynamically generate the seed value according to one or more items of metadata, identification information and time information of the video.

[0110] In the disclosed embodiment, the decryption end may also receive metadata, identification information, and time information of the target video corresponding to the encrypted video, and generate a seed value using the same algorithm as the encryption end. Then, based on the seed value and video features, a decryption key is generated based on the same multimodal large model as the encryption end. When downloading or receiving an encrypted video, one or more of the metadata, identification information, and time information of the target video corresponding to the encrypted video may be obtained, and the above data may be processed based on the same algorithm as the encryption end to obtain a dynamic seed value. The dynamic seed value is the same as the seed value of the encryption end.

[0111] According to the embodiments of the present disclosure, there is no need to transmit seed values ​​between the encryption end and the decryption end, and seed values ​​and decryption keys can be more securely generated dynamically based on a large model, thereby improving the security of keys and videos.

[0112] Fig.10 1 is a flow chart of a method 1000 for video decryption based on a large model according to another embodiment of the present disclosure. The method 1000 can be used to implement step S802 in the method 800 for video decryption based on a large model. In one embodiment, the method 1000 includes: inputting the video feature and the seed value into the multimodal large model to generate a key, and further includes:

[0113] S1001, using the multimodal large model to obtain output information based on the video feature, the seed value and the temperature parameter of the multimodal large model; wherein the video feature includes one or more of a video frame, video content information and a video description text in the target video;

[0114] S1002: Perform a hash operation on the output information to obtain the key.

[0115] In the disclosed embodiment, the decryption end can receive video features, temperature parameters, etc. from the decryption end, and use the same multimodal large model as the encryption end to generate prompt words according to a specific prompt word template based on the received video features and seed values. The prompt words are input into the multimodal large model to generate a decryption key. Output information can also be generated first, and then a hash operation is performed on the output information to obtain a decryption key. Then, the decryption end can use the decryption key and a decryption algorithm such as AES to decrypt the encrypted video to obtain the target video. In addition, the decryption end can receive one or more of the encrypted video, video features, seed values, and temperature parameters of the multimodal large model from the encryption end directly or through other devices. One or more of the metadata, identification information, and time information of the target video for dynamically calculating the seed value can also be received from the encryption end. The decryption end can dynamically calculate the seed value using the same algorithm as the encryption end. Further, the encryption end and the decryption end can also be pre-configured with the same hash algorithm.

[0116] According to the embodiments of the present disclosure, the multimodal large model can obtain output information based on dynamic seed values ​​and different video features, and then generate a decryption key through a hash operation, thereby improving the decryption success rate and improving the security of the video.

[0117] Fig.11 1 is a flow chart of a large model-based video decryption method 1100 according to another embodiment of the present disclosure. The method 1100 can be used to implement step S803 in the large model-based video decryption method 800. In one embodiment, the method 1100 includes: using the key to decrypt the encrypted video to obtain the target video, and further includes:

[0118] S1101, using the key to decrypt the one or more encrypted video blocks to obtain the one or more decrypted video blocks;

[0119] S1102: Merge the one or more decrypted video blocks to obtain the target video.

[0120] In the disclosed embodiment, the decryption end can receive multiple encrypted video blocks from the encryption end or other devices. Based on the decryption key calculated in the above embodiment, the encrypted video blocks can be decrypted separately to obtain decrypted video blocks, and then the decrypted video blocks are merged in the order of blocks to obtain the target video. If any video block is not successfully decrypted, it is difficult to obtain the complete video, and the security is higher.

[0121] According to the disclosed embodiment, each encrypted video block can be decrypted based on the key to obtain the target video, which reduces the computing power occupied during decryption and improves the decryption efficiency. In addition, the file size of the video block is small and the transmission speed is fast, which facilitates the improvement of the video transmission speed through parallel transmission and other methods.

[0122] Fig.12 1 is a flow chart of a method 1200 for video decryption based on a large model according to another embodiment of the present disclosure. The method 1200 can be used to implement step S802 in the method 800 for video decryption based on a large model. In one embodiment, the method 1200 includes: inputting the video feature and the seed value into the multimodal large model to generate a key, and further includes:

[0123] S1201. Input the overall video features and seed value of the target video into the multimodal large model to generate the key.

[0124] In the disclosed embodiment, after obtaining the overall video features of the target video corresponding to the encrypted video, the same multimodal macro model as the encryption end can be used to generate an overall decryption key based on the overall video features and the seed value. The overall decryption key is used to decrypt and merge each encrypted video block to obtain the target video.

[0125] According to the embodiments of the present disclosure, the same key can be used to decrypt each encrypted video block to obtain the target video, thereby reducing the computing power occupied during decryption and improving the decryption efficiency.

[0126] Fig.13 1 is a flow chart of a method 1300 for video decryption based on a large model according to another embodiment of the present disclosure. The method 1300 can be used to implement step S802 in the method 800 for video decryption based on a large model. In one embodiment, the method 1300 includes: inputting the video feature and the seed value into the multimodal large model to generate a key, and further includes:

[0127] S1301. Input the video features and seed values ​​of the one or more video blocks into the multimodal large model respectively to generate keys corresponding to the one or more video blocks; the key corresponding to each video block is used to encrypt the video block.

[0128] In the disclosed embodiment, the video features corresponding to each encrypted video block can be obtained respectively, for example, the video features corresponding to each encrypted video block are received from the decryption end or other devices. And the video features and seed values ​​of each video block are respectively input into the multimodal large model to generate the decryption key of each encrypted video block. For example, based on the video features and seed value of the encrypted video block Vs1, the decryption key generated is S1. Based on the video features and seed value of the encrypted video block Vs2, the decryption key generated is S2. After the decryption key of the video block is generated by the multimodal large model same as the encryption end, the decryption key of the video block is used to decrypt each encrypted video block, and then merged to obtain the target video.

[0129] According to the embodiments of the present disclosure, different video blocks corresponding to the encrypted video blocks can be decrypted using their respective keys, which can improve the security and transmission rate of the video.

[0130] Fig.14 1 is a flow chart of a large model-based video decryption method 1400 according to another embodiment of the present disclosure. The method 1400 can be used to implement the large model-based video decryption method 800. In one implementation, the method further includes:

[0131] S1401, using the key to decrypt the encrypted video to obtain the target video, including: using the key to decrypt the encrypted video to obtain a mixed video;

[0132] S1402. Separate the false data from the mixed video according to the position information of the false data in the target video to obtain the target video.

[0133] In the disclosed embodiment, the decryption key obtained by the above embodiment can be used to decrypt the encrypted video to obtain a mixed video. Then, the position of the false data in the target video or the interference mode of the false data on the target video is obtained, and based on the position of the false data in the target video, the false data is removed from the mixed video to obtain the target video. For example, if the position of the false data in the target video is the first 10 frames, the first 10 frames of the mixed video can be removed to obtain the target data. If the position of the false data in the target video is the last 5 frames, the last 5 frames of the mixed video can be removed to obtain the target data. If the position of the false data in the target video is the 1st, 5th, 7th, and 10th frames, the 1st, 5th, 7th, and 10th frames of the mixed video can be removed to obtain the target data.

[0134] According to the embodiments of the present disclosure, the target video can be obtained by removing the interference of false data, thereby improving the security of the video.

[0135] In one implementation, the location information of the false data in the target video is used to generate the seed value.

[0136] In the disclosed embodiment, the decryption end may receive the position of the false data in the target video or the interference mode of the false data on the target video from the encryption end or other devices, and based on this information, the seed value and the decryption key may be dynamically calculated and generated.

[0137] According to the embodiment of the present disclosure, based on the position information of the false data in the target video, the decryption end can remove the false data after decryption, so that the decryption end can obtain the correct target video.

[0138] In the encryption method of the embodiment of the present disclosure, a random text generated based on a language model can be used as an encryption key. The specific steps are as follows:

[0139] 1. Language model selection: Select a large-scale pre-trained language model (referred to as big language model or big model) that can generate random and diverse text based on specific input.

[0140] 2. Seed value generation: (1) Use a specific seed value as input to ensure that the generated key is unique and random each time it is encrypted. (2) Dynamically generate a seed value based on timestamp, user identity, video metadata (such as file size, creation time, encoding format, etc.) and a composite algorithm. The seed value can be automatically adjusted each time encryption is performed to enhance the security of encryption.

[0141] 3. Temperature setting: Adjust the temperature parameter of the language model to control the randomness and creativity of the text generated by the model. The higher the temperature, the more random the generated text; the lower the temperature, the more stable the generated text.

[0142] 4. Encryption key generation:

[0143] (1) Generate a piece of text using the above seed value and temperature parameters, and convert the text generated by the model into an encryption key through a hash function.

[0144] (2) The model for generating the key can be a multimodal large model. The model can use the images and text of the video frames as data. It can not only generate encryption keys based on the seed value, but also integrate the features of the video content, such as the color distribution and motion vector of the video frame. Inputting the video features and seed values ​​into the model can make the generated key have higher randomness and complexity to ensure the encryption effect.

[0145] 5. Encryption process: Fig.15 As shown, a seed value is input into the language model, and a language model is selected to generate a key. The generated key is used to encrypt the video data. A symmetric encryption algorithm (such as AES) can be used, but the key is a dynamic key generated by the language model. In addition, the video data can be pre-processed (such as split into multiple data blocks) and then encrypted block by block. The encryption keys of different video blocks can be the same or different.

[0146] 6. Decryption process: Fig.16 As shown in the figure, enter the same seed value as the encryption process, and select the same language model as the encryption process to generate the key. Use the generated key to decrypt the encrypted video data and restore the original video content. If the video is divided into blocks during encryption, after receiving the encrypted video blocks, use the decryption key to decrypt the data block by block. The decrypted data blocks will be reassembled into the original video content to ensure that the restored content is consistent with the original video content.

[0147] 6. Forged video data: In the process of video encryption, in addition to using traditional encryption algorithms, generative adversarial networks can be used to generate false data (streams). These false data (streams) are mixed with real data, making the encrypted data not only difficult to decrypt, but also difficult to identify the real content contained in it before decryption. The decryption end can help identify which parts of the data are real through the features in the decryption key generation process (such as certain specific patterns of seed values).

[0148] In addition, encryption complexity can be further enhanced: the pseudo-encryption layer generated by the generative adversarial network can increase the difficulty of cracking, because the cracker needs to identify and separate the false data stream from the real data stream, which greatly increases the complexity of decryption.

[0149] According to the embodiments of the present disclosure, since the seed value and the encryption key can be dynamically generated, the randomness and diversity of the large language model are used to make cracking difficult, thereby improving the security of encryption. The key is generated by the language model, which reduces the dependence on the traditional key management system. One or more of the information used to generate the seed value, the seed value, the temperature parameter, the block, the information of the forged false data, etc. can be adjusted according to actual needs to adapt to different security requirements and application scenarios.

[0150] Fig.17 1 is a schematic diagram of a large model-based video encryption device 1700 according to an embodiment of the present disclosure. In one implementation, the device includes:

[0151] An extraction module 1710, for extracting video features from a target video;

[0152] A first input module 1720, for inputting the video feature and the seed value into the multimodal macro model to generate a key;

[0153] The encryption module 1730 is used to encrypt the target video using the key to obtain an encrypted video.

[0154] Fig.18 1 is a schematic diagram of a large model-based video encryption device 1800 according to another embodiment of the present disclosure. The device 1800 may include an extraction module 1810, a first input module 1820, and an encryption module 1830. The functions of the above modules can refer to the functions of the modules of the large model-based video encryption device in the above embodiment. In one embodiment, the device further includes:

[0155] The first seed module 1840 is used to dynamically generate the seed value according to one or more items of metadata, identification information and time information of the video.

[0156] In one embodiment, Fig.18As shown, the first input module 1820 includes:

[0157] The first output submodule 1821 is used to obtain output information using the multimodal large model based on the video feature, the seed value and the temperature parameter of the multimodal large model; wherein the video feature includes one or more of the video frame, video content information and video description text in the target video;

[0158] The first operation submodule 1822 is used to perform a hash operation on the output information to obtain the key.

[0159] In one embodiment, Fig.18 As shown, the encryption module 1830 includes:

[0160] The block division submodule 1831 is used to divide the target video into blocks to obtain one or more video blocks;

[0161] The video block encryption submodule 1832 is used to encrypt the one or more video blocks using the key to obtain one or more encrypted video blocks.

[0162] In one embodiment, the first input module 1820 is further used to input the overall video features and seed value of the target video into the multimodal large model to generate the key.

[0163] In one embodiment, the first input module 1820 is further used to input the video features and seed values ​​of the one or more video blocks into the multimodal large model respectively to generate keys corresponding to the one or more video blocks; the key corresponding to each video block is used to encrypt the video block.

[0164] In one embodiment, Fig.18 As shown, the device also includes:

[0165] The adversarial module 1850 is used to process the target video using an adversarial algorithm to generate false data;

[0166] A mixing module 1860, configured to mix the target video and the false data to obtain a mixed video;

[0167] In one implementation, the encryption module 1830 is further configured to use the key to encrypt the mixed video to obtain the encrypted video.

[0168] In one implementation, the information used to generate the seed value also includes location information of the false data in the target video.

[0169] Fig.191 is a schematic diagram of a large model-based video decryption device 1900 according to an embodiment of the present disclosure. In one implementation, the device includes:

[0170] An acquisition module 1910 is used to acquire video features of a target video and an encrypted video;

[0171] A second input module 1920, for inputting the video feature and the seed value into the multimodal macro model to generate a key;

[0172] The decryption module 1930 is used to use the key to decrypt the encrypted video to obtain the target video.

[0173] Fig. 20 2 is a schematic diagram of a large model-based video decryption device 2000 according to another embodiment of the present disclosure. The device 2000 may include an acquisition module 2010, a second input module 2020, and a decryption module 2030. The functions of the above modules can refer to the functions of the modules of the large model-based video decryption device in the above embodiment. In one embodiment, the device further includes:

[0174] The second seed module 2040 is used to dynamically generate the seed value according to one or more items of metadata, identification information and time information of the video.

[0175] In one embodiment, Fig. 20 As shown, the second input module 2020 includes:

[0176] The second output submodule 2021 is used to obtain output information using the multimodal large model based on the video feature, the seed value and the temperature parameter of the multimodal large model; wherein the video feature includes one or more of the video frame, video content information and video description text in the target video;

[0177] The second operation submodule 2022 is used to perform a hash operation on the output information to obtain the key.

[0178] In one embodiment, Fig. 20 As shown, the decryption module 2030 includes:

[0179] The video block decryption submodule 2031 is used to decrypt the one or more encrypted video blocks using the key to obtain the one or more decrypted video blocks;

[0180] The merging submodule 2032 is used to merge the one or more decrypted video blocks to obtain the target video.

[0181] In one embodiment, the second input module 2020 is further used to input the overall video features and seed value of the target video into the multimodal large model to generate the key.

[0182] In one embodiment, the second input module 2020 is further used to input the video features and seed values ​​of the one or more video blocks into the multimodal large model respectively to generate keys corresponding to the one or more video blocks; the key corresponding to each video block is used to encrypt the video block.

[0183] In one embodiment, Fig. 20 As shown, the device also includes:

[0184] The decryption module 2030 is further used to decrypt the encrypted video using the key to obtain a mixed video;

[0185] The separation module 2050 is used to separate the false data from the mixed video according to the position information of the false data in the target video to obtain the target video.

[0186] In one implementation, the location information of the false data in the target video is used to generate the seed value.

[0187] For the description of specific functions and examples of each module and submodule of the device in the embodiment of the present disclosure, reference can be made to the relevant description of the corresponding steps in the above method embodiment, which will not be repeated here.

[0188] In the technical solution disclosed herein, the acquisition, storage and application of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0189] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0190] Fig.21 A schematic block diagram of an example electronic device 2100 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0191] like Fig.21As shown, the device 2100 includes a computing unit 2101, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 2102 or a computer program loaded from a storage unit 2108 into a random access memory (RAM) 2103. In the RAM 2103, various programs and data required for the operation of the device 2100 can also be stored. The computing unit 2101, the ROM 2102, and the RAM 2103 are connected to each other via a bus 2104. An input / output (I / O) interface 2105 is also connected to the bus 2104.

[0192] A number of components in the device 2100 are connected to the I / O interface 2105, including: an input unit 2106, such as a keyboard, a mouse, etc.; an output unit 2107, such as various types of displays, speakers, etc.; a storage unit 2108, such as a disk, an optical disk, etc.; and a communication unit 2109, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 2109 allows the device 2100 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0193] The computing unit 2101 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 2101 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 2101 performs the various methods and processes described above, such as a video encryption method based on a large model and / or a video decryption method based on a large model. For example, in some embodiments, a video encryption method based on a large model and / or a video decryption method based on a large model may be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as a storage unit 2108. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 2100 via the ROM 2102 and / or the communication unit 2109. When the computer program is loaded into the RAM 2103 and executed by the computing unit 2101, one or more steps of the video encryption method based on the large model and / or the video decryption method based on the large model described above may be performed. Alternatively, in other embodiments, the computing unit 2101 may be configured to execute the large model-based video encryption method and / or the large model-based video decryption method in any other appropriate manner (eg, by means of firmware).

[0194] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0195] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0196] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0197] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0198] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0199] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0200] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0201] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the principles of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A video encryption method based on a large model, comprising: Extracting video features from the target video; Inputting the video features and the seed value into a multimodal macro model to generate a key; The target video is encrypted using the key to obtain an encrypted video.

2. The method according to claim 1, further comprising: The seed value is dynamically generated according to one or more of metadata, identification information and time information of the video.

3. The method according to claim 1 or 2, wherein: The video features and the seed value are input into a multimodal large model to generate a key, including: Obtaining output information using the multimodal large model based on the video features, the seed value, and the temperature parameter of the multimodal large model; wherein the video features include one or more of a video frame, video content information, and a video description text in the target video; A hash operation is performed on the output information to obtain the key.

4. The method according to any one of claims 1 to 3, wherein: The target video is encrypted using the key to obtain an encrypted video, including: Divide the target video into blocks to obtain one or more video blocks; The one or more video blocks are encrypted using the key to obtain one or more encrypted video blocks.

5. The method according to claim 4, wherein: The video features and the seed value are input into a multimodal large model to generate a key, including: The overall video features and a seed value of the target video are input into a multimodal large model to generate the key.

6. The method according to claim 4, wherein: Inputting the video features and the seed value into a multimodal large model to generate a key also includes: The video features and seed values ​​of the one or more video blocks are respectively input into the multimodal large model to generate keys corresponding to the one or more video blocks; the key corresponding to each video block is used to encrypt the video block.

7. The method according to any one of claims 1 to 3, further comprising: Process the target video using an adversarial algorithm to generate false data; Mixing the target video and the false data to obtain a mixed video; The target video is encrypted using the key to obtain an encrypted video, including: the mixed video is encrypted using the key to obtain the encrypted video.

8. The method according to claim 7, wherein: The information used to generate the seed value also includes location information of the false data in the target video.

9. A video decryption method based on a large model, comprising: Obtain video features and encrypted video of the target video; Inputting the video features and the seed value into a multimodal macro model to generate a key; The encrypted video is decrypted using the key to obtain the target video.

10. The method according to claim 9, further comprising: The seed value is dynamically generated according to one or more of metadata, identification information and time information of the video.

11. The method according to claim 9 or 10, wherein: The video features and the seed value are input into a multimodal large model to generate a key, including: Obtaining output information using the multimodal large model based on the video features, the seed value, and the temperature parameter of the multimodal large model; wherein the video features include one or more of a video frame, video content information, and a video description text in the target video; A hash operation is performed on the output information to obtain the key.

12. The method according to any one of claims 9 to 11, wherein: Decrypting the encrypted video using the key to obtain the target video includes: decrypting one or more encrypted video blocks using the key to obtain the one or more decrypted video blocks; The one or more decrypted video blocks are merged to obtain the target video.

13. The method according to claim 12, wherein: The video features and the seed value are input into a multimodal large model to generate a key, including: The overall video features and a seed value of the target video are input into a multimodal large model to generate the key.

14. The method according to claim 12, wherein: Inputting the video features and the seed value into a multimodal large model to generate a key also includes: The video features and seed values ​​of the one or more video blocks are respectively input into the multimodal large model to generate keys corresponding to the one or more video blocks; the key corresponding to each video block is used to encrypt the video block.

15. The method according to any one of claims 9 to 11, further comprising: Decrypting the encrypted video using the key to obtain the target video includes: decrypting the encrypted video using the key to obtain a mixed video; The false data is separated from the mixed video according to the position information of the false data in the target video to obtain the target video.

16. The method according to claim 15, wherein: The position information of the false data in the target video is used to generate the seed value.

17. A video encryption device based on a large model, comprising: An extraction module, used to extract video features from the target video; A first input module, configured to input the video features and the seed value into a multimodal macro model to generate a key; The encryption module is used to encrypt the target video using the key to obtain an encrypted video.

18. The apparatus according to claim 17, further comprising: The first seed module is used to dynamically generate the seed value according to one or more of the metadata, identification information and time information of the video.

19. The device according to claim 17 or 18, wherein: The first input module comprises: A first output submodule, configured to obtain output information using the multimodal large model based on the video feature, the seed value, and a temperature parameter of the multimodal large model; wherein the video feature includes one or more of a video frame, video content information, and a video description text in the target video; The first operation submodule is used to perform a hash operation on the output information to obtain the key.

20. The device according to any one of claims 17 to 19, wherein The encryption module comprises: A block submodule, used to block the target video to obtain one or more video blocks; The video block encryption submodule is used to encrypt the one or more video blocks using the key to obtain one or more encrypted video blocks.

21. The device according to claim 20, wherein: The first input module is also used to input the overall video features and seed value of the target video into the multimodal large model to generate the key.

22. The device according to claim 20, wherein: The first input module is also used to input the video features and seed values ​​of the one or more video blocks into the multimodal large model respectively to generate keys corresponding to the one or more video blocks; the key corresponding to each video block is used to encrypt the video block.

23. The apparatus according to any one of claims 17 to 19, further comprising: An adversarial module, used for processing the target video using an adversarial algorithm to generate false data; A mixing module, used for mixing the target video and the false data to obtain a mixed video; The encryption module is further used to encrypt the mixed video using the key to obtain the encrypted video.

24. The device according to claim 23, wherein: The information used to generate the seed value also includes location information of the false data in the target video.

25. A video decryption device based on a large model, comprising: An acquisition module, used to acquire video features and encrypted video of a target video; A second input module, used for inputting the video features and the seed value into the multimodal macro model to generate a key; The decryption module is used to decrypt the encrypted video using the key to obtain the target video.

26. The apparatus according to claim 25, further comprising: The second seed module is used to dynamically generate the seed value according to one or more of the metadata, identification information and time information of the video.

27. The device according to claim 25 or 26, wherein: The second input module comprises: A second output submodule, configured to obtain output information using the multimodal large model based on the video features, the seed value, and a temperature parameter of the multimodal large model; wherein the video features include one or more of a video frame, video content information, and a video description text in the target video; The second operation submodule is used to perform a hash operation on the output information to obtain the key.

28. The device according to any one of claims 25 to 27, wherein The decryption module comprises: a video block decryption submodule, configured to decrypt one or more encrypted video blocks using the key to obtain the one or more decrypted video blocks; The merging submodule is used to merge the one or more decrypted video blocks to obtain the target video.

29. The device according to claim 28, wherein The second input module is also used to input the overall video features and seed value of the target video into the multimodal large model to generate the key.

30. The device according to claim 28, wherein The second input module is also used to input the video features and seed values ​​of the one or more video blocks into the multimodal large model respectively to generate keys corresponding to the one or more video blocks; the key corresponding to each video block is used to encrypt the video block.

31. The apparatus according to any one of claims 25 to 27, further comprising: The decryption module is further used to decrypt the encrypted video using the key to obtain a mixed video; The separation module is used to separate the false data from the mixed video according to the position information of the false data in the target video to obtain the target video.

32. The device according to claim 31, wherein The position information of the false data in the target video is used to generate the seed value.

33. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-8 or any one of claims 9-16.

34. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-9 or any one of claims 9-16.

35. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 9 or any one of claims 9 to 16.

36. A large model comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 9 or any one of claims 9 to 16.

Images