Intelligent conference control method and system based on multi-modal perception and quantum encryption

Abstract

The application relates to the technical field of data transmission, in particular to an intelligent conference control method and system based on multi-modal perception and quantum encryption. The method determines motion performance by the current change of image data information of a conference multi-attendee terminal, analyzes the encryption priority of the attendee terminal according to the correlation degree of the motion performance and the conference and the correlation degree of audio text, evaluates the necessity of key update of different attendees according to the change degree of the encryption priority of the attendee terminal in the previous sequence and the time span performance of the key update, and performs encrypted data transmission according to the encryption priority. The application analyzes multi-modal data of the attendee terminal and the close relationship between conferences, preferentially encrypts data with high correlation conference content information, analyzes the key update condition according to the encryption priority change, and ensures the safe transmission of conference data.

Description

Technical Field

[0001] The present invention relates to the field of data transmission technology, and in particular to an intelligent conference control method and system based on multimodal perception and quantum encryption. Background Art

[0002] With the acceleration of digitalization and the widespread adoption of hybrid conferencing models, practical quantum computing poses a greater threat to traditional encryption systems, exposing high-value meeting content to security risks. Therefore, integrating multimodal perception to achieve understanding and autonomous control of the environment, behavior, and content, and integrating quantum key distribution (QKD) and post-quantum cryptography (PQC) to build an information-theoretic security protection system, have become key measures to resist quantum attacks and meet the rigid needs of conferencing.

[0003] In large-scale smart conference scenarios, when quantum encryption technology is used to ensure the security of information in the meeting, due to the large amount of real-time data transmission between multiple devices, the demand for keys during the transmission process is large, and the limited real-time coding capability of quantum key distribution equipment is insufficient to support the demand for keys. As a result, some critical data may be forced to use strategies such as key reuse, which makes it easy to be stolen or attacked, affecting the encryption transmission security of conference data. Summary of the Invention

[0004] In order to solve the technical problems in the above-mentioned prior art, the purpose of the present invention is to provide an intelligent conference control method and system based on multimodal perception and quantum encryption. The technical solutions adopted are as follows:

[0005] The present invention provides an intelligent conference control method based on multimodal perception and quantum encryption, the method comprising:

[0006] Obtaining image data and audio text data from each participant terminal;

[0007] Based on the characteristic displacement between the image data of each participant at the current moment and the previous image data, the current motion performance index of each participant is obtained; based on the correlation between the image data before the current moment and the meeting behavior, combined with the motion performance index, the current meeting relevance of each participant is obtained; based on the meeting relevance of words in the audio text data before the current moment, combined with the meeting relevance, the current encryption priority of each participant is obtained; the participant terminals are sorted from largest to smallest according to the encryption priority to obtain the current encryption priority sequence;

[0008] Based on the encryption priority deviation of each participant in the current encryption priority sequence and the order deviation between the previous encryption priority sequence and the previous encryption priority sequence, the priority change index of each participant is obtained; based on the order and encryption priority of each participant in the current encryption priority sequence, combined with the previous key update duration and priority change index, the current key update judgment index of each participant is obtained;

[0009] The key of the participating end is updated based on the key update judgment index, and encrypted transmission is performed according to the current encryption priority sequence.

[0010] Furthermore, the method for obtaining the sports performance index includes:

[0011] For any participant, feature point matching is performed between the current image data of the participant and the image data at the previous sampling moment to obtain a matching group. The positional distance between two pixels in each matching group is used as the displacement change degree of each matching group. The average of the displacement change degrees of all the current matching groups of the participant is used as the current displacement index of the participant.

[0012] Perform negative mapping on the number of all matching groups currently on the participant to obtain the current matching index of the participant;

[0013] The product of the current displacement index of the participant terminal and the matching index is normalized to obtain the current motion performance index of the participant terminal.

[0014] Furthermore, the method for obtaining the conference relevance includes:

[0015] For any participating terminal, a preset number of image data before the current moment is recorded as analysis image data;

[0016] When the motion performance index in the analyzed image data is greater than the preset motion threshold, the video stream of the preset period before the analyzed image data is input into the trained neural network model, and the behavior relevance probability value of the analyzed image data is output; the average of all current behavior relevance probability values ​​is used as the current behavior relevance of the participating terminal;

[0017] The product of the mean value of the motion performance index of all currently analyzed image data of the participant terminal and the behavior correlation is used as the conference relevance of the participant terminal.

[0018] Furthermore, the method for obtaining the encryption priority includes:

[0019] For any participant, obtain each word vector in the current audio text data of the participant;

[0020] Calculate the maximum similarity between each word vector and the word vector in the preset conference vocabulary as the text relevance of each word vector; normalize the product of the mean of all current text relevances of the participant and the conference relevance to obtain the encryption priority of the participant.

[0021] Furthermore, the method for obtaining the priority change indicator includes:

[0022] For any participating terminal, in the current encryption priority sequence, calculate the difference between the mean of all encryption priorities and the encryption priority of the participating terminal to obtain the level deviation degree of the participating terminal;

[0023] Calculate the difference between the sequence number of the participating terminal in the previous encryption priority sequence and the sequence number in the current encryption priority sequence to obtain the ranking change degree of the participating terminal;

[0024] The priority change index of the participant terminal is obtained by combining the level deviation degree and the ranking change degree of the participant terminal.

[0025] Furthermore, the method for obtaining the key update determination indicator includes:

[0026] For any participating terminal, the product of the negative correlation mapping value of the participating terminal in the current encryption priority sequence and the encryption priority is normalized and used as the important performance weight of the participating terminal;

[0027] Based on the important performance weight, the time since the last key update of the participant and the priority change index are weighted and summed to obtain the key update judgment index of the participant.

[0028] Furthermore, the weighted summation of the time from the last key update of the participant terminal and the priority change index based on the important performance weight to obtain the key update determination index of the participant terminal includes:

[0029] The product of the importance performance weight and the time since the last key update is used as the time weighted value; the difference between the value 1 and the importance performance weight is used as the change performance weight; the product of the change performance weight and the priority change index is used as the change weighted value;

[0030] The sum of the duration weighted value and the change weighted value is used as the key update judgment indicator for the participating terminal.

[0031] Furthermore, the updating of the key of the participant terminal based on the key update determination indicator and the encrypted transmission according to the current encryption priority sequence include:

[0032] When the key update judgment index is greater than the preset update threshold, the key of the corresponding participating end will be updated, and all participating ends will be encrypted and transmitted according to the current encryption priority sequence.

[0033] Furthermore, the method for obtaining the matching group includes:

[0034] The feature points in the current image data are obtained through SIFT, and the optical flow method is used to match the image data at the previous sampling moment to obtain a matching group.

[0035] The present invention also provides an intelligent conference control system based on multimodal perception and quantum encryption, comprising a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, the steps of the above-mentioned intelligent conference control method based on multimodal perception and quantum encryption are implemented.

[0036] The present invention has the following beneficial effects:

[0037] The present invention analyzes the priority of different participating terminals in encrypted transmission by analyzing the current changes in multimodal data information under multiple participating terminals in a conference, combining the relevance of the change information with the conference. Furthermore, the present invention evaluates the key update judgment index of different participating terminals by combining the degree of change of the participating terminals under the previous encryption priority and the time span of the key update. This ensures the security of data transmission in the conference on the basis of ensuring the priority encryption transmission of important data. Based on the close relationship between the multimodal data analysis of the participating terminals and the conference, the present invention prioritizes the encryption of data with higher content information related to the conference, and analyzes the key update situation in combination with the encryption priority changes to ensure the secure transmission of conference data. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1 A flow chart of an intelligent conference control method based on multimodal perception and quantum encryption provided by one embodiment of the present invention;

[0040] Figure 2 A schematic diagram of a video screen change provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0041] To further illustrate the technical means and effectiveness of the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of an intelligent conference control method and system based on multimodal sensing and quantum encryption proposed in accordance with the present invention. In the following description, different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0042] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0043] The following describes in detail a specific solution of an intelligent conference control method and system based on multimodal perception and quantum encryption provided by the present invention with reference to the accompanying drawings.

[0044] During the meeting, each participant collects multimodal data from participants in real time and then transmits it synchronously. The data generated by participants during the smart meeting is analyzed to facilitate encrypted transmission of meeting data, thereby ensuring meeting security. Given the limited real-time encoding capabilities of the system's key distribution devices, encryption priority analysis is required for the data collected by each device, and keys are adjusted to ensure secure and reliable encrypted transmission.

[0045] See also Figure 1 , which shows a flow chart of an intelligent conference control method based on multimodal perception and quantum encryption provided by one embodiment of the present invention, the method comprising the following steps:

[0046] S1: Acquire image data and audio text data from each participant.

[0047] During a smart conference, local multimodal data is collected from each participant. Multiple integrated sensor components capture real-time multimodal data from participants, including image data and audio signal data at each moment. Image information can be used to analyze participant movement, while audio signals can be used to analyze participant speech. Comprehensive multimodal data evaluation prioritizes encrypted transmission, ensuring secure transmission of meeting data.

[0048] To facilitate subsequent speech analysis of the audio signal, speech recognition technology is used to convert the audio signal into text data, and analysis is performed based on the speech-to-text data. It should be noted that the settings for speech recognition conversion and acquisition frequency can be adjusted by the implementer based on the specific implementation scenario and are not limited or detailed here.

[0049] S2: Based on the characteristic displacement between the image data of each participant at the current moment and the previous image data, the current motion performance index of each participant is obtained; based on the degree of correlation between the image data before the current moment and the meeting behavior, combined with the motion performance index, the current meeting relevance of each participant is obtained; based on the meeting relevance of the words in the audio text data before the current moment, combined with the meeting relevance, the current encryption priority of each participant is obtained; the participating terminals are sorted from large to small according to the encryption priority to obtain the current encryption priority sequence.

[0050] During a meeting, not all data and information are important. Based on the multimodal data collected by each participant, further analysis is conducted to determine whether this information is closely related to the meeting content. Data carrying meeting content information needs to be encrypted first to ensure the secure transmission of meeting data.

[0051] During a meeting, to ensure the effectiveness of the meeting, there are often individual speakers giving speeches or other types of presentations at the same time, while the rest of the participants are generally in a listening state, that is, they remain in the same position. However, since the speaker needs to speak or present, his or her video image will change, so analysis is performed based on the image data collected by the participants. Figure 2 , which shows a schematic diagram of a video screen change provided by an embodiment of the present invention.

[0052] When a speaker is speaking, the image information on the screen generally changes. Since the position of the built-in camera in the device remains unchanged, when the coordinates of the corresponding feature points between the captured images change significantly, and when large changes occur, the proportion of matching feature points between the captured images in time sequence is smaller, indicating that the degree of change in the screen information of the participants at the current continuous moment is higher.

[0053] Preferably, in an embodiment of the present invention, the method for obtaining the sports performance index includes:

[0054] First, for any participating terminal, feature point matching is performed between the current image data of the participating terminal and the image data at the previous sampling moment to obtain a matching group. In an embodiment of the present invention, feature points in the current image data are obtained through SIFT, and optical flow is used to match them with the image data at the previous sampling moment to obtain matching groups. Each matching group corresponds to two pixels, and feature point matching can be used to screen out the presence of identical features in the image. It should be noted that the feature point screening and matching methods are technical means well known to those skilled in the art and are not elaborated or limited here.

[0055] The positional distance between two pixels in each matching group is then used as the displacement change degree of each matching group, representing the degree of positional change of the same feature point. The positional distance can be calculated using Euclidean distance. The average displacement change degree of all matching groups on the participating terminal is used as the current displacement index of the participating terminal. A smaller displacement index indicates fewer motion information features.

[0056] Further considering that changes in image information may reduce the number of matchable feature points, the number of all current matching groups on the participating terminal is negatively mapped to obtain the current matching index of the participating terminal. The smaller the number, the higher the likelihood of information change. It should be noted that negative correlation mapping is a technical means well known to those skilled in the art, and can be implemented in the form of negative exponential power or inverse proportional value based on a natural constant, etc., and is not further described or limited here.

[0057] Finally, the product of the participant's current displacement index and the matching index is normalized to obtain the participant's current motion performance index. This is combined with the distance between the midpoints of the matching groups and the number of matching groups to obtain the participant's current motion performance index. A larger motion performance index indicates a possible change in the image information. It should be noted that normalization is a well-known technique for those skilled in the art, and can be performed using linear normalization or standard normalization. The specific normalization method is not limited here.

[0058] Even if there are changes in the movement of participants, when this change is not related to the meeting, the importance will be lower and the need for priority encryption will be lower. Therefore, the image data is further combined with the correlation between the meeting behavior to obtain the correlation between the picture change and the meeting. Preferably, in an embodiment of the present invention, the method for obtaining the meeting relevance includes:

[0059] For any participating terminal, a preset number of image data before the current moment is recorded as analysis image data. To reduce misjudgment, the image information in the continuous time sequence before the current moment is further analyzed based on the comprehensive motion situation. When the motion performance index in the analyzed image data is greater than the preset motion threshold, it indicates significant motion performance, and at this time, a person with a large behavioral amplitude change appears. In this embodiment of the present invention, the preset number can be set to 10, and the preset motion threshold can be set to 0.6. The specific values ​​can be adjusted by the implementer.

[0060] The video stream for a preset period of time before the image data analysis is then fed into the trained neural network model, which then outputs the probability value associated with the behavior in the analyzed image data. The probability is then derived by combining the video stream with the neural network model. The greater the probability, the more relevant the behavior is to the meeting content and the greater the difference from meaningless behavior. The preset period can be set to the first 2 seconds, and the specific setting is adjustable by the implementer.

[0061] It should be noted that the neural network model can select the training set as labeled video images of people, and the labels are the behaviors of the people in the video images. After the training is completed, the model is tested using the validation set. The specific training of the neural network model is a technical means well known to those skilled in the art and will not be elaborated or limited here.

[0062] The correlation levels of multiple motion situations are comprehensively analyzed, and the average of all current behavior correlation probability values ​​is taken as the current behavior correlation of the participating terminal, reflecting the relevant analysis of the screen performance before the current moment.

[0063] The product of the mean of all the motion performance indicators of all the currently analyzed image data of the participant and the behavioral correlation is combined as the conference relevance of the participant, reflecting the conference relevance of the picture changes. The greater the conference relevance, the more important the facial changes of the corresponding participant are, and encrypted transmission needs to be given priority.

[0064] Further analysis is needed in conjunction with the collected audio and text data. Normally, while the speaker is speaking, other participants remain mute. If the text data is highly relevant to the meeting topic, it should also be prioritized for encryption. Therefore, further analysis is conducted in conjunction with the collected audio and text data to more accurately assess priorities.

[0065] Preferably, in an embodiment of the present invention, the method for obtaining the encryption priority includes:

[0066] For any participating terminal, each word vector in the current audio text data of the participating terminal is obtained. In an embodiment of the present invention, the word vector can be obtained using the Word2Vec word embedding model. Word vector extraction is a technical means well known to those skilled in the art and is not limited or elaborated here.

[0067] By extracting word vectors, the relevant situation of the meeting can be further analyzed. The maximum similarity between each word vector and the word vector in the preset meeting vocabulary is calculated as the text relevance of each word vector. The similarity between each word vector and the preset meeting vocabulary related to the extracted meeting is calculated. The higher the similarity, the more significant the conference relevance represented by the corresponding word vector. In an embodiment of the present invention, the similarity calculation between word vectors can be obtained using cosine similarity. The implementer of the preset meeting vocabulary can set it according to the specific implementation scenario, and there is no limitation here.

[0068] Combined with the conference relevance of text and picture, the product of the mean of all current text relevance of the participant and the conference relevance is normalized to obtain the encryption priority of the participant. When the relevance of the speech text to the conference is high and the information relevance of the picture change is greater, the necessity of encrypted transmission first is greater, so the encryption priority is greater.

[0069] Taking into account the coding capability of the key distribution device in the system, the key is distributed to the data with high encryption priority first. Therefore, after obtaining the encryption priority, the encryption priority sequence can be obtained by sorting from large to small according to the encryption priority. The earlier the participating end is in the sequence, the more important it is and the higher the priority encryption demand is.

[0070] In a meeting, a speaker's speech generally lasts for a period of time. During this period, the identity of the speaker will not change. However, considering that there may be questions and answers, in an embodiment of the present invention, the time interval between each two analyses of the encrypted priority sequence is set to 10 seconds, which can be adjusted by the implementer.

[0071] S3: Based on the encryption priority deviation of each participating terminal in the current encryption priority sequence and the order deviation between the previous encryption priority sequences, the priority change index of each participating terminal is obtained; through the order and encryption priority of each participating terminal in the current encryption priority sequence, combined with the previous key update duration and priority change index, the current key update judgment index of each participating terminal is obtained.

[0072] During quantum encryption, data transmitted with fixed keys can be easily cracked, potentially leaking meeting content and compromising security. Therefore, key updates are necessary during data transmission. For data with a higher encryption priority, which typically carries important meeting information, key updates must be performed more frequently to ensure secure transmission. For data with a lower encryption priority, which typically carries less important information, key updates can be performed less frequently. Furthermore, given the high volume of data generated during a meeting, the capabilities of the key distribution equipment in the system must also be considered.

[0073] When the encryption priority of the generated multimodal data remains unchanged, the key update frequency for data with a higher encryption priority is greater than that for data with a lower encryption priority. When the encryption priority of the generated multimodal data changes, no key update is required for data with a lower encryption priority. Otherwise, a key update is required.

[0074] Therefore, the priority change situation is analyzed and then adjusted and updated. Preferably, in an embodiment of the present invention, the method for obtaining the priority change indicator includes:

[0075] For any participating terminal, in the current encryption priority sequence, the difference between the mean of all encryption priorities and the encryption priority of the participating terminal is calculated to obtain the level deviation of the participating terminal. The higher the overall deviation of the encryption priority of the participating terminal, the more significant the change in the information characteristics of the participating terminal is on the overall participating terminal. The larger the deviation, the more likely the identity of the speaker in the meeting is to change, and the greater the necessity of key update.

[0076] Then, the difference between the serial number of the participant in the previous encryption priority sequence and the serial number in the current encryption priority sequence is calculated to obtain the ranking change degree of the participant. By comparing with the previous encryption priority analysis, when the encryption priority order changes greatly, it means that the data transmission encryption process may adopt a less secure strategy such as key reuse, and there is a higher demand for key update.

[0077] Finally, the level deviation and ranking change of the participating end are combined to obtain the priority change index of the participating end. In an embodiment of the present invention, the product of the level deviation and the ranking change is normalized to obtain the priority change index of the participating end, reflecting the priority change situation.

[0078] Further considering the update time span to analyze the necessity of key updates, since the data generated by participants with lower encryption priorities is less important, and considering the capacity limitations of the key distribution device in the system, key updates are prioritized for data generated by participants with higher encryption priorities. In other words, for participants with lower encryption priorities, the need for key updates is less with a longer time span than for participants with higher encryption priorities with the same time span.

[0079] Preferably, in an embodiment of the present invention, the method for obtaining the key update determination indicator includes:

[0080] For multimodal data with a lower encryption priority, its importance is lower and the key can be left unrekeyed for a longer period of time. Therefore, the lower the participant's ranking in the encryption priority sequence, the smaller the weight. For any participant, the product of the negative correlation mapping value of the participant's sequence number in the current encryption priority sequence and the encryption priority is normalized to serve as the participant's importance weight. The higher the encryption priority and the higher the ranking among all participants, the more important the participant is.

[0081] Then, based on the important performance weight, the weighted sum of the time from the last key update of the participating terminal and the priority change index is performed to obtain the key update judgment index of the participating terminal. In the embodiment of the present invention, the product of the important performance weight and the time from the last key update is used as the time weighted value. For data with higher importance, the priority is higher considering the time. Therefore, the difference between the value 1 and the important performance weight is used as the change performance weight, and the product of the change performance weight and the priority change index is used as the change weighted value to reflect the adjustment of the weight due to the impact of the priority change. Finally, the sum of the time weighted value and the change weighted value is used as the key update judgment index of the participating terminal.

[0082] Thus, the necessity of performing key update for corresponding data transmission is obtained each time the encryption priority is calculated.

[0083] S4: Update the key of the participating end based on the key update judgment index, and encrypt the transmission according to the current encryption priority sequence.

[0084] Based on the necessity of key update during data transmission, data information generated by conference speakers and others can be transmitted preferentially and securely during the meeting to ensure that key data in the meeting is not leaked, thereby ensuring the normal progress of the meeting.

[0085] In this embodiment of the present invention, when the key update determination indicator exceeds a preset update threshold, it indicates that the data generated by the participating terminal at that moment requires a key update during encrypted transmission. The corresponding participating terminal's key is updated, and all participating terminals are encrypted for transmission according to the current encryption priority sequence. The system continuously analyzes key updates during data transmission during the conference to ensure the security of data transmission generated during the conference. The preset update threshold can be set to 0.7, and the specific value can be adjusted by the implementer and is not limited here.

[0086] In summary, the present invention analyzes the priority of different participating terminals in encrypted transmission by combining the current changes in multimodal data information under multiple participating terminals with the relevance of the change information to the conference. It further combines the degree of change of the participating terminals under the previous encryption priority and the time span performance of the key update to evaluate the key update judgment index of different participating terminals, ensuring the priority encryption transmission of important data and guaranteeing the security of data transmission in the conference. Based on the close relationship between the multimodal data analysis of the participating terminals and the conference, the present invention prioritizes the encryption of data with higher content information related to the conference, and analyzes the key update situation in combination with the encryption priority change to ensure the secure transmission of conference data.

[0087] The present invention also provides an intelligent conference control system based on multimodal perception and quantum encryption, comprising a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, the steps of the above-mentioned intelligent conference control method based on multimodal perception and quantum encryption are implemented.

[0088] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0089] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. An intelligent conference control method based on multimodal perception and quantum encryption, characterized in that: The method comprises: Obtaining image data and audio text data from each participant terminal; Based on the characteristic displacement between the image data of each participant at the current moment and the previous image data, the current motion performance index of each participant is obtained; based on the correlation between the image data before the current moment and the meeting behavior, combined with the motion performance index, the current meeting relevance of each participant is obtained; based on the meeting relevance of words in the audio text data before the current moment, combined with the meeting relevance, the current encryption priority of each participant is obtained; the participant terminals are sorted from largest to smallest according to the encryption priority to obtain the current encryption priority sequence; Based on the encryption priority deviation of each participant in the current encryption priority sequence and the order deviation between the previous encryption priority sequence and the previous encryption priority sequence, the priority change index of each participant is obtained; based on the order and encryption priority of each participant in the current encryption priority sequence, combined with the previous key update duration and priority change index, the current key update judgment index of each participant is obtained; The key of the participating end is updated based on the key update judgment index, and encrypted transmission is performed according to the current encryption priority sequence.

2. The intelligent conference control method based on multimodal perception and quantum encryption according to claim 1 is characterized in that: The method for obtaining the sports performance index includes: For any participant, feature point matching is performed between the current image data of the participant and the image data at the previous sampling moment to obtain a matching group. The positional distance between two pixels in each matching group is used as the displacement change degree of each matching group. The average of the displacement change degrees of all the current matching groups of the participant is used as the current displacement index of the participant. Perform negative mapping on the number of all matching groups currently on the participant to obtain the current matching index of the participant; The product of the current displacement index of the participant terminal and the matching index is normalized to obtain the current motion performance index of the participant terminal.

3. The intelligent conference control method based on multimodal perception and quantum encryption according to claim 1 is characterized in that: The method for obtaining the conference relevance includes: For any participating terminal, a preset number of image data before the current moment is recorded as analysis image data; When the motion performance index in the analyzed image data is greater than the preset motion threshold, the video stream of the preset period before the analyzed image data is input into the trained neural network model, and the behavior relevance probability value of the analyzed image data is output; the average of all current behavior relevance probability values ​​is used as the current behavior relevance of the participating terminal; The product of the mean value of the motion performance index of all currently analyzed image data of the participant terminal and the behavior correlation is used as the conference relevance of the participant terminal.

4. The intelligent conference control method based on multimodal perception and quantum encryption according to claim 1 is characterized in that: The method for obtaining the encryption priority includes: For any participant, obtain each word vector in the current audio text data of the participant; Calculate the maximum similarity between each word vector and the word vector in the preset conference vocabulary as the text relevance of each word vector; normalize the product of the mean of all current text relevances of the participant and the conference relevance to obtain the encryption priority of the participant.

5. The intelligent conference control method based on multimodal perception and quantum encryption according to claim 1 is characterized in that: The method for obtaining the priority change indicator includes: For any participating terminal, in the current encryption priority sequence, calculate the difference between the mean of all encryption priorities and the encryption priority of the participating terminal to obtain the level deviation degree of the participating terminal; Calculate the difference between the sequence number of the participating terminal in the previous encryption priority sequence and the sequence number in the current encryption priority sequence to obtain the ranking change degree of the participating terminal; The priority change index of the participant terminal is obtained by combining the level deviation degree and the ranking change degree of the participant terminal.

6. The intelligent conference control method based on multimodal perception and quantum encryption according to claim 1 is characterized in that: The method for obtaining the key update determination indicator includes: For any participating terminal, the product of the negative correlation mapping value of the participating terminal in the current encryption priority sequence and the encryption priority is normalized and used as the important performance weight of the participating terminal; Based on the important performance weight, the time since the last key update of the participant and the priority change index are weighted and summed to obtain the key update judgment index of the participant.

7. The intelligent conference control method based on multimodal perception and quantum encryption according to claim 6 is characterized in that: The method of performing weighted summation of the time from the last key update of the participant terminal and the priority change index based on the important performance weight to obtain the key update determination index of the participant terminal includes: The product of the importance performance weight and the time since the last key update is used as the time weighted value; the difference between the value 1 and the importance performance weight is used as the change performance weight; the product of the change performance weight and the priority change index is used as the change weighted value; The sum of the duration weighted value and the change weighted value is used as the key update judgment indicator for the participating terminal.

8. The intelligent conference control method based on multimodal perception and quantum encryption according to claim 1 is characterized in that: The method of updating the key of the participating terminal based on the key update determination indicator and performing encrypted transmission according to the current encryption priority sequence includes: When the key update judgment index is greater than the preset update threshold, the key of the corresponding participating end will be updated, and all participating ends will be encrypted and transmitted according to the current encryption priority sequence.

9. The intelligent conference control method based on multimodal perception and quantum encryption according to claim 2 is characterized in that: The method for obtaining the matching group includes: The feature points in the current image data are obtained through SIFT, and the optical flow method is used to match the image data at the previous sampling moment to obtain a matching group.

10. An intelligent conference control system based on multimodal perception and quantum encryption, 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 computer program, the processor implements the steps of an intelligent conference control method based on multimodal perception and quantum encryption as described in any one of claims 1 to 9.

Images