Virtual camera recognition method and device based on mobile phone

Abstract

The present invention relates to the field of security identification technology, and in particular to a method and device for virtual camera identification based on a mobile phone. The method comprises the following steps: obtaining the calling authority of a mobile phone camera; performing continuous user portrait recording based on the calling authority of an identification controller of the mobile phone camera to generate a user video frame set; performing video frame set network transmission detection on the user video frame set to generate a dynamic transmission user video frame set and a static user video frame set; performing transmission risk analysis on the dynamic transmission user video frame set to generate dynamic transmission risk data; performing device screen acquisition based on the static user video frame set to generate a device screen display interface; performing user display video reading on the device screen display interface to generate a user video display frame set. The present invention improves the accuracy of virtual camera identification and improves the safety of users through technical means such as continuous recording, difference analysis, and model training.

Description

Technical Field

[0001] The present invention relates to the field of security identification technology, and in particular to a virtual camera identification method and device based on a mobile phone. Background Art

[0002] Image recognition technology based on computer vision is mainly used in the fields of industry and scientific research. These technologies include face recognition, object recognition, etc., but more efficient algorithms and hardware support are needed to implement them on mobile devices. The development of deep learning and neural networks has brought revolutionary changes to image recognition. The emergence of algorithms such as convolutional neural networks (CNN) has improved the accuracy and speed of image recognition, making it possible to implement virtual camera recognition on mobile devices. The success of applications such as Pokémon Go has promoted the development and popularization of augmented reality technology. Virtual camera recognition plays an important role in AR, such as identifying objects in real scenes and displaying related information or virtual objects on the screen. However, the current recognition of virtual cameras is often performed locally through the detection of system hardware, and cannot be dynamically recognized based on long-distance transmission and dynamic images, resulting in low recognition accuracy and poor security. Summary of the invention

[0003] Based on this, it is necessary to provide a mobile phone-based virtual camera recognition method and device to solve at least one of the above technical problems.

[0004] To achieve the above object, a virtual camera recognition method based on a mobile phone is provided, the method comprising the following steps:

[0005] Step S1: obtaining the calling authority of the mobile phone camera; based on the mobile phone camera, using the calling authority of the recognition controller to continuously record the user portrait to generate a user video frame set; performing video frame set network transmission detection on the user video frame set to generate a dynamic transmission user video frame set and a static user video frame set; performing transmission risk analysis on the dynamic transmission user video frame set to generate dynamic transmission risk data;

[0006] Step S2: acquiring the device screen based on the static user video frame set to generate a device screen display interface; reading the user display video on the device screen display interface to generate a user video display frame set; performing feature point distance matching on the standard user video recording frame set and the user video display frame set to obtain matching warning feature points; performing frame set comparison analysis on the standard user video recording frame set and the user video display frame set by matching warning feature points to generate frame set comparison analysis data;

[0007] Step S3: obtaining current network speed data; performing frame set delay difference analysis on the frame set comparison analysis data based on the current network speed data to generate frame set delay difference analysis data; performing motion trajectory difference analysis on the user video display frame set through the frame set delay difference analysis data to generate motion trajectory difference analysis data;

[0008] Step S4: Perform model training based on motion trajectory difference analysis data and frame set comparison analysis data to generate a virtual camera risk prediction model; import static recognition difference data into the virtual camera risk prediction model for risk prediction to generate static transmission risk prediction data; perform entropy weighting on the static transmission risk prediction data and the dynamic transmission risk data to generate a virtual camera recognition weight; compare the virtual camera recognition weight with a preset standard virtual camera recognition weight threshold, and when the virtual camera recognition weight is greater than or equal to the preset standard virtual camera recognition weight threshold, a virtual camera recognition alarm prompt is issued to the mobile phone.

[0009] The present invention can detect existing security risks and protect the user's privacy from being violated by analyzing and comparing the user video frame set. Through network transmission detection of dynamic transmission user video frame sets and static user video frame sets, possible data leakage and attack behaviors can be discovered and prevented in time. Based on model training and data analysis, a virtual camera risk prediction model and static transmission risk prediction data are generated to effectively predict existing security risks. When the virtual camera recognition weight exceeds the preset threshold, the mobile phone can be promptly issued a virtual camera recognition alarm prompt to remind the user to pay attention to the existing security issues. Improve the user experience and let the user feel more at ease when using the mobile phone camera without worrying about privacy leakage and security issues. Therefore, the present invention improves the accuracy of virtual camera recognition and improves the safety of users through technical means such as continuous recording, difference analysis, and model training.

[0010] In this specification, a virtual camera recognition device based on a mobile phone is provided, including a recognition controller for executing the above-mentioned virtual camera recognition method based on a mobile phone. The recognition controller of the virtual camera recognition device based on a mobile phone includes:

[0011] The transmission risk identification module is used to obtain the calling authority of the mobile phone camera; based on the mobile phone camera, the recognition controller calling authority is used to continuously record the user portrait to generate a user video frame set; the user video frame set is subjected to video frame set network transmission detection to generate a dynamic transmission user video frame set and a static user video frame set; the dynamic transmission user video frame set is subjected to transmission risk analysis to generate dynamic transmission risk data;

[0012] The frame set comparison module is used to acquire the device screen based on the static user video frame set to generate the device screen display interface; read the user display video on the device screen display interface to generate the user video display frame set; perform feature point distance matching on the standard user video recording frame set and the user video display frame set to obtain matching warning feature points; perform frame set comparison analysis on the standard user video recording frame set and the user video display frame set by matching the warning feature points to generate frame set comparison analysis data;

[0013] The delay analysis module is used to obtain the current network speed data; perform frame set delay difference analysis on the frame set comparison analysis data based on the current network speed data to generate frame set delay difference analysis data; perform motion trajectory difference analysis on the user video display frame set through the frame set delay difference analysis data to generate motion trajectory difference analysis data;

[0014] The risk prediction module is used to perform model training based on motion trajectory difference analysis data and frame set comparison analysis data to generate a virtual camera risk prediction model; import static recognition difference data into the virtual camera risk prediction model for risk prediction to generate static transmission risk prediction data; perform entropy weighting on the static transmission risk prediction data and the dynamic transmission risk data to generate a virtual camera recognition weight; compare the virtual camera recognition weight with a preset standard virtual camera recognition weight threshold, and when the virtual camera recognition weight is greater than or equal to the preset standard virtual camera recognition weight threshold, a virtual camera recognition alarm prompt is issued to the mobile phone.

[0015] The beneficial effect of the present invention is that through continuous user portrait recording and transmission risk analysis, the monitoring and identification capabilities of potential privacy leaks are improved, and user privacy protection is enhanced. The static user video frame set and device screen acquisition are used for difference analysis to effectively identify whether the device screen content is screenshotted or recorded to prevent security risks. The virtual camera risk prediction model generated by frame set comparison analysis and model training performs entropy weighting on static transmission risks and dynamic transmission risks to provide accurate virtual camera identification weights and alarm prompts. Using frame set delay difference analysis and motion trajectory difference analysis, combined with current network speed data, real-time monitoring and accurate identification of virtual camera recording behavior are achieved. By combining a variety of technical means, users can feel more at ease when using mobile phones to avoid privacy leaks and unsafe camera behaviors. Static identification difference data and frame set comparison analysis effectively prevent illegal behaviors such as screenshots and screen recordings, and protect user privacy and security. It integrates technologies such as continuous recording, difference analysis, and model training to improve the efficiency and accuracy of identifying potential privacy leaks and security risks. The virtual camera recognition weight is compared with the preset standard recognition weight threshold, allowing personalized risk assessment and providing customized safety tips for different user needs. Therefore, the present invention improves the accuracy of virtual camera recognition and improves the safety of users through technical means such as continuous recording, difference analysis, and model training. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of the steps of a virtual camera recognition method based on a mobile phone;

[0017] Figure 2 for Figure 1 Detailed implementation steps of step S1 in FIG.

[0018] Figure 3 for Figure 2 Detailed implementation steps of step S15;

[0019] Figure 4 for Figure 1 Detailed implementation steps of step S2 in FIG.

[0020] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0021] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0022] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0023] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0024] To achieve this, please refer to Figures 1 to 4 , a virtual camera recognition method based on a mobile phone, comprising the following steps:

[0025] Step S1: obtaining the calling authority of the mobile phone camera; based on the mobile phone camera, using the calling authority of the recognition controller to continuously record the user portrait to generate a user video frame set; performing video frame set network transmission detection on the user video frame set to generate a dynamic transmission user video frame set and a static user video frame set; performing transmission risk analysis on the dynamic transmission user video frame set to generate dynamic transmission risk data;

[0026] Step S2: acquiring the device screen based on the static user video frame set to generate a device screen display interface; reading the user display video on the device screen display interface to generate a user video display frame set; performing feature point distance matching on the standard user video recording frame set and the user video display frame set to obtain matching warning feature points; performing frame set comparison analysis on the standard user video recording frame set and the user video display frame set by matching warning feature points to generate frame set comparison analysis data;

[0027] Step S3: obtaining current network speed data; performing frame set delay difference analysis on the frame set comparison analysis data based on the current network speed data to generate frame set delay difference analysis data; performing motion trajectory difference analysis on the user video display frame set through the frame set delay difference analysis data to generate motion trajectory difference analysis data;

[0028] Step S4: Perform model training based on motion trajectory difference analysis data and frame set comparison analysis data to generate a virtual camera risk prediction model; import static recognition difference data into the virtual camera risk prediction model for risk prediction to generate static transmission risk prediction data; perform entropy weighting on the static transmission risk prediction data and the dynamic transmission risk data to generate a virtual camera recognition weight; compare the virtual camera recognition weight with a preset standard virtual camera recognition weight threshold, and when the virtual camera recognition weight is greater than or equal to the preset standard virtual camera recognition weight threshold, a virtual camera recognition alarm prompt is issued to the mobile phone.

[0029] The present invention can detect existing security risks and protect the user's privacy from being violated by analyzing and comparing the user video frame set. Through network transmission detection of dynamic transmission user video frame sets and static user video frame sets, possible data leakage and attack behaviors can be discovered and prevented in time. Based on model training and data analysis, a virtual camera risk prediction model and static transmission risk prediction data are generated to effectively predict existing security risks. When the virtual camera recognition weight exceeds the preset threshold, the mobile phone can be promptly issued a virtual camera recognition alarm prompt to remind the user to pay attention to the existing security issues. Improve the user experience and let the user feel more at ease when using the mobile phone camera without worrying about privacy leakage and security issues. Therefore, the present invention improves the accuracy of virtual camera recognition and improves the safety of users through technical means such as continuous recording, difference analysis, and model training.

[0030] In the embodiment of the present invention, reference Figure 1 The above is a schematic diagram of the steps of a virtual camera recognition method based on a mobile phone of the present invention. In this example, the virtual camera recognition method based on a mobile phone includes the following steps:

[0031] Step S1: obtaining the calling authority of the mobile phone camera; based on the mobile phone camera, using the calling authority of the recognition controller to continuously record the user portrait to generate a user video frame set; performing video frame set network transmission detection on the user video frame set to generate a dynamic transmission user video frame set and a static user video frame set; performing transmission risk analysis on the dynamic transmission user video frame set to generate dynamic transmission risk data;

[0032] In an embodiment of the present invention, the camera call permission is requested from the user by using the permission request API provided by the mobile phone operating system, such as the requestPermissions() method of Android or the permission request framework of iOS. After the user grants the permission, the application can access the mobile phone camera for recording and analysis. The continuous portrait recording function is implemented by calling the camera API using a mobile application development framework (such as Java of Android or Swift / Objective-C of iOS). The recognition controller is used to control the recording behavior, such as setting parameters such as recording time, frame rate, and resolution. During the recording process, the continuous user portraits are recorded as videos, and each frame is saved as an image to form a user video frame set. The user video frame set is transmitted to a specified server or target device using a network transmission protocol (such as HTTP, TCP / IP, etc.). During the transmission process, a network monitoring tool or a custom transmission detection algorithm is used to monitor the data transmission situation in real time, including indicators such as transmission rate, packet loss rate, and delay. According to the monitoring results, the user video frame set is divided into a dynamic transmission user video frame set and a static user video frame set. The dynamic transmission user video frame set refers to the video frames that change during the transmission process, while the static user video frame set is the video frame that does not change.

[0033] Step S2: acquiring the device screen based on the static user video frame set to generate a device screen display interface; reading the user display video on the device screen display interface to generate a user video display frame set; performing feature point distance matching on the standard user video recording frame set and the user video display frame set to obtain matching warning feature points; performing frame set comparison analysis on the standard user video recording frame set and the user video display frame set by matching warning feature points to generate frame set comparison analysis data;

[0034] In an embodiment of the present invention, a static user video frame set is analyzed by using image processing technology (such as OpenCV library) to identify the position and boundary of the device screen. According to the recognition result, the size, position and direction of the device screen are determined to generate a virtual representation of the device screen display interface. A video stream is obtained from a user display video source (such as a mobile phone screen, a computer screen, etc.). A video processing library (such as FFmpeg) is used to read the video stream and parse it into a series of video frames. The user display portion is extracted from the parsed video frame to form a user video display frame set. A computer vision library (such as OpenCV) is used to extract feature points of a standard user video recording frame set and a user video display frame set. The two feature point sets are matched, usually using feature descriptors (such as SIFT, SURF, etc.) and matching algorithms (such as matching algorithms based on feature distances). According to the matching results, the matched feature points and the distances between them are determined, and then the matching warning feature points are identified. The matched warning feature points are used as references to find the corresponding areas in the standard user video recording frame set and the user video display frame set. The found areas are compared and analyzed, and indicators such as pixel differences, color differences, and shape changes can be calculated. Based on the comparison and analysis results, frame set comparison and analysis data is generated, including warning area, degree of difference, matching degree and other information.

[0035] Step S3: obtaining current network speed data; performing frame set delay difference analysis on the frame set comparison analysis data based on the current network speed data to generate frame set delay difference analysis data; performing motion trajectory difference analysis on the user video display frame set through the frame set delay difference analysis data to generate motion trajectory difference analysis data;

[0036] In an embodiment of the present invention, the network speed data of the current device is obtained by using a network tool or API (such as Speedtest.net). This can be an indicator such as download speed, upload speed, delay, etc. The generated frame set comparison analysis data is associated with the current network speed data, and the impact of network delay on frame transmission is considered. Analysis of the transmission delay between different frames can be achieved by calculating the difference between timestamps. Higher delays will cause delay differences in frame sets. Based on the delay difference, frame set delay difference analysis data is generated, including information such as delay value and delay change. The user video display frame set is time-aligned using the delay difference data to solve the delay problem between frame sets. Computer vision technology (such as optical flow method) is used to perform motion analysis on the time-aligned frame set to detect the motion trajectory in each frame. By comparing the differences between the motion trajectories, abnormal motion or inconsistency in the user video display frame set can be detected. Based on the difference in motion trajectories, motion trajectory difference analysis data is generated, including information such as trajectory offset and speed change.

[0037] Step S4: Perform model training based on motion trajectory difference analysis data and frame set comparison analysis data to generate a virtual camera risk prediction model; import static recognition difference data into the virtual camera risk prediction model for risk prediction to generate static transmission risk prediction data; perform entropy weighting on the static transmission risk prediction data and the dynamic transmission risk data to generate a virtual camera recognition weight; compare the virtual camera recognition weight with a preset standard virtual camera recognition weight threshold, and when the virtual camera recognition weight is greater than or equal to the preset standard virtual camera recognition weight threshold, a virtual camera recognition alarm prompt is issued to the mobile phone.

[0038] In an embodiment of the present invention, by using machine learning or deep learning algorithms, such as convolutional neural networks (CNN) or recurrent neural networks (RNN), model training is performed based on motion trajectory difference analysis data and frame set comparison analysis data. A supervised learning method can be used, with frame set comparison analysis data and motion trajectory difference analysis data as input features and corresponding risk labels as outputs. The model is trained to predict the risk level of the virtual camera, for example, whether there is an abnormal motion trajectory or frame set delay difference. Using the trained virtual camera risk prediction model, the static recognition difference data is imported into the model for risk prediction to generate static transmission risk prediction data. The static transmission risk prediction data may include the risk level of each camera static picture, such as the recognition of abnormal images or abnormal objects. The static transmission risk prediction data and the dynamic transmission risk data are entropy weighted to comprehensively consider static and dynamic risks. The entropy weight method is a comprehensive evaluation method that calculates the weight of each indicator according to the importance and degree of difference of each indicator. The virtual camera recognition weight after entropy weighting is compared with the preset standard virtual camera recognition weight threshold. If the virtual camera recognition weight is greater than or equal to the preset standard virtual camera recognition weight threshold, the virtual camera recognition alarm prompt is triggered.

[0039] Preferably, step S1 comprises the following steps:

[0040] Step S11: Obtaining the mobile phone camera calling authority;

[0041] Step S12: Based on the mobile phone camera, the recognition controller calls the authority to continuously record the user's portrait to generate a user video frame set;

[0042] Step S13: performing frame image preprocessing on the user video frame set to generate a standard user video recording frame set, wherein the frame image preprocessing includes image brightness enhancement, image contrast enhancement and image denoising;

[0043] Step S14: performing video frame set network transmission detection through the standard user video recording frame set, and generating a dynamic transmission user video frame set when it is detected that the standard user video recording frame set is being transmitted over the network; and generating a static user video frame set when it is not detected that the standard user video recording frame set is being transmitted over the network;

[0044] Step S15: performing transmission signal analysis on the dynamic transmission user video frame set to generate dynamic transmission signal analysis data; performing transmission risk analysis on the dynamic transmission signal analysis data to generate dynamic transmission risk data.

[0045] The present invention ensures that the application can access the mobile phone camera for subsequent video recording and processing. Continuous recording is performed through the mobile phone camera to generate a user's video frame set as input data for subsequent processing. Image preprocessing is performed on the user video frame set, including brightness enhancement, contrast enhancement and denoising processing, to improve image quality and reduce noise interference, and a standard video recording frame set is generated. The dynamic nature of the user video frame set is determined by detecting whether the standard user video recording frame set is transmitted over the network. If network transmission is detected, a dynamic transmission user video frame set is generated; if network transmission is not detected, a static user video frame set is generated. Transmission signal analysis is performed on the dynamic transmission user video frame set to identify potential transmission problems and risks. Risk analysis is performed on the transmission signal analysis data to evaluate the reliability and security of the transmission, and corresponding dynamic transmission risk data is generated.

[0046] As an example of the present invention, refer to Figure 2 As shown, in this example, step S1 includes:

[0047] Step S11: Obtaining the mobile phone camera calling authority;

[0048] In an embodiment of the present invention, the permission to use the camera is requested from the user. This is usually done by popping up a dialog box or a system alert, explaining to the user why the camera needs to be accessed and requesting permission. On iOS, the AVCaptureDevice class can be used to request camera permissions, while on Android, a permission request library or a built-in permission framework can be used. Once the user is requested for permission, the response needs to be processed. If the user agrees, the camera continues to be used. If the user refuses, it needs to be handled accordingly, whether to provide additional explanations or guidance, or to provide alternative functions without permission. At certain points in the application, the current camera permission status needs to be checked. This can be achieved by calling the corresponding API to determine whether the user has granted permission. This check is to ensure that the user has authorized permission before trying to access the camera. While the user is using the application, they can change the permission settings at any time. Therefore, a permission check should be performed before each attempt to access the camera, and the situation of permission changes should be handled when necessary.

[0049] Step S12: Based on the mobile phone camera, the recognition controller calls the authority to continuously record the user's portrait to generate a user video frame set;

[0050] In an embodiment of the present invention, in an application, it is first necessary to request the user to grant camera call permission. This is usually achieved by calling an appropriate permission request method, such as using AVCaptureDevice.requestAccess(for:.video) on iOS. Once the camera permission is obtained, the camera device can be initialized to prepare for recording. This usually involves using relevant APIs to configure and connect to the camera of the mobile phone. After initializing the camera, it is necessary to set the recording parameters, such as frame rate, resolution, video encoding format, etc. These parameters will affect the quality and size of the final generated video. By calling the start recording method, start capturing video frames from the camera. In this step, a loop or timer can be set to continuously record a certain length of video. Whenever the camera captures a new video frame, it can be saved to a buffer or processed directly. This process involves image processing techniques, such as portrait detection, filter application, etc. During the recording process, the processed video frames are saved one by one into a collection so that a user video frame set can be generated later. When the recording reaches a preset duration or the user stops recording, the recording process ends and the relevant resources are released, such as closing the camera connection. The generated user video frame set is saved to a suitable storage location for subsequent processing or display.

[0051] Step S13: performing frame image preprocessing on the user video frame set to generate a standard user video recording frame set, wherein the frame image preprocessing includes image brightness enhancement, image contrast enhancement and image denoising;

[0052] In an embodiment of the present invention, the contrast and brightness distribution of an image are enhanced by using a histogram equalization algorithm. Histogram equalization is performed according to the local area of ​​the image to avoid over-enhancing noise. The brightness of the image is adjusted using a gamma function to make it brighter or dimmer. The pixel values ​​of the image are linearly stretched to enhance the contrast of the image. The pixel values ​​of the image are mapped to a target histogram to enhance the contrast of the image and match the brightness distribution of the target image. The image is blurred using a Gaussian filter to reduce high-frequency noise in the image. Pixel points are replaced with the median in the neighborhood to eliminate salt and pepper noise in the image. De-noising the image using a wavelet transform can effectively retain the detail information of the image.

[0053] Step S14: performing video frame set network transmission detection through the standard user video recording frame set, and generating a dynamic transmission user video frame set when it is detected that the standard user video recording frame set is being transmitted over the network; and generating a static user video frame set when it is not detected that the standard user video recording frame set is being transmitted over the network;

[0054] In an embodiment of the present invention, video frame set data is obtained from a standard user video recording frame set, and these data can be a pre-recorded video file or a real-time captured video frame. Use appropriate network monitoring methods and tools to detect whether the standard user video recording frame set is undergoing network transmission, which involves monitoring network data traffic, network connection status or other related indicators. According to the result of the network transmission detection, determine whether the transmission type of the standard user video recording frame set is dynamic or static. If it is detected that network transmission is in progress, a dynamic transmission user video frame set is generated. If no network transmission is detected, a static user video frame set is generated. If it is detected that network transmission is in progress, it is necessary to dynamically obtain the video frames being transmitted, and the video frame data can be obtained in real time through network requests, streaming media protocols, etc. If no network transmission is detected, the video frame data in the standard user video recording frame set is directly used to generate a static user video frame set.

[0055] Step S15: performing transmission signal analysis on the dynamic transmission user video frame set to generate dynamic transmission signal analysis data; performing transmission risk analysis on the dynamic transmission signal analysis data to generate dynamic transmission risk data.

[0056] In an embodiment of the present invention, by acquiring a dynamically transmitted user video frame set, real-time video frame data can be acquired through network transmission or video frames can be dynamically generated. Signal analysis is performed on the dynamically transmitted video frame set, which involves the following aspects: analyzing the transmission rate of the video frame to determine the frame rate of the video. Analyzing the resolution of the video frame to determine the clarity and size of the video. Analyzing the transmission speed of the video data to determine the efficiency and stability of data transmission. The results of the transmission signal analysis are organized into a data form, including information such as frame rate, resolution, and data transmission speed, for subsequent risk analysis. According to the dynamic transmission signal analysis data, a transmission risk analysis is performed, which includes the following contents: analyzing the delay existing during network transmission to determine the real-time and stability of the transmission. Analyzing the frame loss of the video frame during the transmission process to determine the smoothness of the video transmission. Analyzing whether the data is complete and not tampered with during the transmission process to ensure the security and reliability of the data. The results of the transmission risk analysis are organized into a data form, including information such as network delay, frame loss rate, and data integrity, for subsequent processing and decision-making.

[0057] Preferably, step S15 comprises the following steps:

[0058] Step S151: performing transmission signal detection on the dynamically transmitted user video frame set, and when it is detected that the dynamically transmitted user video frame set is transmitted as an input signal, performing camera device code recognition on the mobile phone according to the dynamically transmitted user video frame set to generate input signal detection data;

[0059] Step S152: when it is detected that the dynamic transmission user video frame set is transmitted as the output signal, the transmission device machine code is identified according to the dynamic transmission user video frame set to obtain the transmission device machine code;

[0060] Step S153: Perform camera digital signature authentication by transmitting the device identification machine code to generate a camera digital signature; perform signature matching between the abnormal camera digital signature and a preset virtual camera digital signature library to generate output signal detection data;

[0061] Step S154: integrating the input signal detection data and the output signal detection data to generate dynamic transmission signal analysis data; performing transmission risk analysis on the dynamic transmission signal analysis data to generate dynamic transmission risk data.

[0062] The present invention detects the dynamic transmission of user video frame sets as input signal transmission, and the system can monitor in real time whether the user video frame sets are transmitted, so as to respond in time. Identifying the mobile phone camera device code and the transmission device machine code helps to determine the source and destination of the signal, thereby improving the security and reliability of the signal. By performing camera digital signature authentication on the transmission device machine code, the system can verify the identity and integrity of the device to prevent tampering or counterfeiting. Integrating input signal detection data and output signal detection data helps to comprehensively analyze the characteristics and behavior of the transmission signal, so as to better understand the potential risks in the transmission process. By performing transmission risk analysis on the dynamic transmission signal analysis data, the system can identify and evaluate potential transmission risks, which helps to take corresponding security measures in time to protect data and user privacy.

[0063] As an example of the present invention, refer to Figure 3 As shown, in this example, step S15 includes:

[0064] Step S151: performing transmission signal detection on the dynamically transmitted user video frame set, and when it is detected that the dynamically transmitted user video frame set is transmitted as an input signal, performing camera device code recognition on the mobile phone according to the dynamically transmitted user video frame set to generate input signal detection data;

[0065] In an embodiment of the present invention, a transmission signal detection is performed on a dynamically transmitted user video frame set to determine whether it is transmitted as an input signal. This can be achieved by real-time monitoring and analysis of the video frame set. After the signal transmitted by the video frame set is detected, the next step is entered. Image processing and computer vision technology are used to identify the camera device code for the detected dynamically transmitted user video frame set. This includes the following steps: preprocessing the video frame set, such as denoising, contrast enhancement, etc., to improve the recognition accuracy of the device code. Locating the position of the camera device code in the video frame set involves object detection and positioning algorithms. The located device code area is identified using optical character recognition (OCR) technology or other pattern recognition methods. The recognition results are verified and checked to ensure accuracy and reliability. The identified camera device code and related information are organized into data form, including information such as device code content and recognition time, as input signal detection data.

[0066] Step S152: when it is detected that the dynamic transmission user video frame set is transmitted as the output signal, the transmission device machine code is identified according to the dynamic transmission user video frame set to obtain the transmission device machine code;

[0067] In an embodiment of the present invention, by utilizing computer vision technology, such as a convolutional neural network (CNN), feature extraction is performed on a video frame set to capture the visual features of the transmission device machine code. Object detection and positioning algorithms (such as YOLO, SSD, etc.) are used to locate the position of the transmission device machine code in the video frame set, which requires training a special model to identify the target area. The located machine code area is identified. Optical character recognition (OCR) technology can be used in combination with a deep learning model (such as a recurrent neural network) to identify text information. The identified machine code is verified and checked to ensure accuracy and reliability. Valid information, such as a device identification code, is extracted from the identified machine code.

[0068] Step S153: Perform camera digital signature authentication by transmitting the device identification machine code to generate a camera digital signature; perform signature matching between the abnormal camera digital signature and a preset virtual camera digital signature library to generate output signal detection data;

[0069] In an embodiment of the present invention, a digital signature is generated by using a unique identification code of a camera transmission device in combination with a hash algorithm (such as SHA-256). This ensures that each camera has a unique signature and that each signature is generated based on the unique identification of a specific device. When a video signal transmitted by a camera is received, the transmission device machine code of the camera is used for authentication. By verifying the signature of the transmission device machine code, it is ensured that the source of the video signal is legal and credible. Public key encryption technology can be used to verify the digital signature to ensure the authenticity and integrity of the signature. A preset virtual camera digital signature library is maintained, which contains digital signatures of known legal cameras. These digital signatures are generated in advance by the manufacturer and stored in a database. The received camera digital signature is compared with the signature in the preset virtual camera digital signature library. If the received signature matches any signature in the library, it is marked as legal. Otherwise, it is marked as abnormal. According to the signature matching result, the unmatched signature is regarded as an abnormal situation. An abnormality detection algorithm (such as isolation forest, support vector machine, etc.) can be used to identify abnormal signatures, thereby marking the credibility of the camera signal. According to the signature matching result, output signal detection data is generated. This data includes the location, timestamp, anomaly type and other information of the abnormal camera for subsequent processing and analysis.

[0070] Step S154: integrating the input signal detection data and the output signal detection data to generate dynamic transmission signal analysis data; performing transmission risk analysis on the dynamic transmission signal analysis data to generate dynamic transmission risk data.

[0071] In an embodiment of the present invention, the input signal detection data and the output signal detection data are merged into one data set by using a data integration tool (such as the Pandas library in Python), which involves matching and aligning data fields to ensure the consistency of the data structure. Dynamic transmission signal analysis is performed on the integrated data set. Machine learning techniques, such as cluster analysis, anomaly detection, or timing analysis, are used to detect potential signal patterns or abnormal situations. For example, clustering algorithms (such as K-means clustering) can be used to group data points into different transmission modes, and abnormal clusters can be identified as potential abnormal signals. Based on the dynamic transmission signal analysis data, a deep analysis of transmission risks is performed. Using threat modeling technology, the attack path is modeled to identify potential attack targets and threat vectors. At the same time, vulnerability analysis technology is used to evaluate the vulnerabilities of the system or application to determine potential security risks. Using risk assessment techniques (such as risk matrix, attack tree analysis), the identified transmission risks are quantitatively or qualitatively evaluated, including the potential impact, possibility, and severity of the risk. Based on the results of the transmission risk analysis, dynamic transmission risk data is generated.

[0072] Preferably, step S2 comprises the following steps:

[0073] Step S21: acquiring the device screen based on the static user video frame set to generate a device screen display interface; segmenting the device screen display interface into user display frames to generate user display frames; reading the user display video from the user display frames to generate a user video display frame set;

[0074] Step S22: extracting frame image feature points from the standard user video recording frame set to obtain user recording frame image feature data; performing frame image feature point matching on the user video display frame set according to the user recording frame image feature data to generate a frame image feature point matching pair;

[0075] Step S23: performing matching distance difference analysis on the frame image feature point matching pairs to generate frame image feature point matching difference distance data; comparing the frame image feature point matching difference distance data with a preset standard feature point matching difference distance threshold, and when the frame image feature point matching difference distance data is less than the standard feature point matching difference distance threshold, generating a matching warning feature point;

[0076] Step S24: performing frame set comparison analysis on the standard user video recording frame set and the user video display frame set by matching the warning feature points to generate frame set comparison analysis data.

[0077] The present invention acquires the device screen based on a static user video frame set, so that the system can acquire the display interface of the device screen and divide it into a user display frame. In this way, it can be ensured that the application or interface being used by the user is captured. Then, the system reads the video from the user display frame and generates a user video display frame set to provide data for subsequent analysis. The frame image feature points are extracted from the standard user video recording frame set to obtain the feature data of the user recorded frame image. Then, the system matches these feature data with the user video display frame set to generate a frame image feature point matching pair. This helps to determine the position and features of the frame image recorded by the user in the screen display. The system performs a matching distance difference analysis on the frame image feature point matching pair to generate frame image feature point matching difference distance data. Then, these data are compared with a preset standard feature point matching difference distance threshold. When the matching difference distance data is less than the threshold, the system will generate a matching warning feature point to indicate the existence of recording behavior. By matching the warning feature points, the standard user video recording frame set and the user video display frame set are compared and analyzed to generate frame set comparison analysis data, which helps to determine whether there is a recording behavior inconsistent with the standard recording frame set and provide further analysis and warning.

[0078] As an example of the present invention, refer to Figure 4 As shown, in this example, step S2 includes:

[0079] Step S21: acquiring the device screen based on the static user video frame set to generate a device screen display interface; segmenting the device screen display interface into user display frames to generate user display frames; reading the user display video from the user display frames to generate a user video display frame set;

[0080] In an embodiment of the present invention, a static image of the device screen is obtained by using a screen recording software or a screenshot API provided by the operating system. This can capture the display content of the current device screen without interfering with the user. The display interface of the device screen is generated by the acquired static image of the device screen, which involves image processing techniques such as image segmentation, feature extraction, etc., for subsequent analysis and processing. After the device screen display interface is generated, the interface is analyzed and processed to determine the application or interface that the user is using, which involves image segmentation and pattern recognition techniques to distinguish different display frames, such as application windows, browser tabs, etc. For each user display frame, the video content of the corresponding area is extracted from the static image of the device screen, specifically by extracting an image sequence and forming a video frame set at a certain frame rate. The video frame set composed of the extracted user display frame video content is saved as a user video display frame set, and the video frame can be simply saved as an image sequence, or further video encoding and decoding and processing can be performed.

[0081] Step S22: extracting frame image feature points from the standard user video recording frame set to obtain user recording frame image feature data; performing frame image feature point matching on the user video display frame set according to the user recording frame image feature data to generate a frame image feature point matching pair;

[0082] In the embodiment of the present invention, a feature point extraction algorithm in computer vision is used, such as SIFT (Scale Invariant Feature Transform), SURF (Speeded Robust Features) or ORB (Oriented FAST and Rotated BRIEF). For each standard user video recording frame, the key points in the image are detected using the selected feature point extraction algorithm. For each key point, its descriptor is calculated to describe the feature information of its surrounding area. Feature matching is performed using feature point descriptors. Common methods include matching algorithms based on feature distances, such as nearest neighbor matching or matching algorithms based on machine learning, such as RANSAC (Random Sampling Consensus) algorithm. For each frame in the user video display frame set, the same feature point extraction algorithm is used to extract feature points of the frame image and calculate its descriptor. The feature point descriptors in the standard user video recording frame set are matched with the feature point descriptors in the user video display frame set. The feature descriptors are matched using a feature point matching algorithm, and a frame image feature point matching pair is generated based on the matching results, that is, each standard frame is associated with its matching frame in the user video display frame set.

[0083] Step S23: performing matching distance difference analysis on the frame image feature point matching pairs to generate frame image feature point matching difference distance data; comparing the frame image feature point matching difference distance data with a preset standard feature point matching difference distance threshold, and when the frame image feature point matching difference distance data is less than the standard feature point matching difference distance threshold, generating a matching warning feature point;

[0084] In an embodiment of the present invention, the differences between matching points are compared by using a distance measurement method (such as Euclidean distance, Hamming distance, etc.). For each pair of frame image feature point matching pairs, the distance difference between them is calculated. The distance difference data for each pair of matching points is saved to form a matching difference distance data set. The generated frame image feature point matching difference distance data is compared with a preset standard feature point matching difference distance threshold. If the matching difference distance is less than the preset threshold, it means that the match is relatively accurate. For feature points whose matching difference distance is less than the standard threshold, they are marked as matching warning feature points, indicating that the matching of these feature points is relatively reliable.

[0085] Step S24: performing frame set comparison analysis on the standard user video recording frame set and the user video display frame set by matching the warning feature points to generate frame set comparison analysis data.

[0086] In an embodiment of the present invention, feature points are extracted from each frame in the standard user video recording frame set and the user video display frame set, and their descriptors are calculated. A feature point matching algorithm, such as a descriptor-based matching algorithm, is used to match the feature points in the standard user video recording frame set with the feature points in the user video display frame set. According to the matching warning feature points generated in step S23, matching warning feature points are filtered out from the feature point matching results. Each frame in the standard user video recording frame set is compared with the corresponding frame in the user video display frame set. For each pair of matching frames, a similarity measurement method is used to calculate the similarity between them, and corresponding frame set comparison analysis data is generated. The similarity data of each pair of matching frames and the matching warning feature point information are summarized to generate frame set comparison analysis data.

[0087] Preferably, step S22 includes the following steps:

[0088] Step S221: performing background separation on a standard user video recording frame set to generate background separation data; performing background color feature marking on the background separation data to generate background color feature data;

[0089] Step S222: performing portrait separation on the standard user video recording frame set using the background separation data to generate portrait separation data;

[0090] Step S223: marking facial feature points of the standard user video recording frame set according to the portrait separation data to generate facial feature data of the portrait;

[0091] Step S224: integrating the background color feature data and the facial features feature data to generate the user recording frame image feature data; performing frame image feature point matching on the user video display frame set according to the user recording frame image feature data to generate a frame image initial feature point matching pair;

[0092] Step S225: performing external point elimination on the initial feature point matching pairs of the frame image to generate frame image eliminated feature point matching pairs; performing geometric verification on the frame image eliminated feature point matching pairs to generate feature point geometric verification data;

[0093] Step S226: using the feature point geometry verification data, perform feature point registration on the frame image after removing the feature point matching pairs, to generate the frame image feature point matching pairs.

[0094] The present invention can separate the foreground and background in the video through background separation, which is convenient for subsequent processing. Background color feature marking can extract and mark the color features in the background, which is helpful for further analysis and processing of the background. Portrait separation enables the characters in the video to be separated from the background, which is of great significance for identifying and analyzing the actions and behaviors of the characters. By marking the feature points of the facial features, the key features of the face, such as eyes, nose, mouth, etc., can be extracted to provide basic data for subsequent face recognition, expression analysis, etc. By integrating the background color feature data and the facial feature data of the portrait, the background and character information can be comprehensively considered to improve the accuracy and efficiency of subsequent processing. Through feature point matching, the corresponding relationship between video frames can be found to achieve video frame registration. Outlier elimination and geometric verification help to eliminate abnormal points in matching and improve the accuracy and stability of matching. The final feature point registration makes the matching between video frames more accurate, providing a reliable basis for subsequent video analysis and processing.

[0095] In an embodiment of the present invention, the background and foreground in the video frame are separated by adopting a traditional method based on pixel value threshold, a statistical model (such as a Gaussian mixture model) or a deep learning method, such as a background subtraction algorithm or a semantic segmentation model based on deep learning. The color features in the background are marked by color space conversion and clustering algorithms in image processing, such as using a K-means clustering algorithm to identify the main background color. An image segmentation algorithm, such as a semantic segmentation model based on region growing, edge detection, graph cuts or deep learning, is used to separate the portrait from the background in the video frame. Face detection and key point detection algorithms, such as a Haar cascade detector, a HOG feature-based detector or a deep learning-based face key point detection model, are used to detect faces and mark their facial feature points. The background color feature data and the facial feature feature data of the portrait can be integrated by simply merging them into a feature vector or using feature fusion techniques, such as principal component analysis (PCA) or multimodal fusion methods. Use feature descriptors and matching algorithms, such as SIFT, SURF, ORB and other descriptors combined with nearest neighbor search-based matching algorithms to match feature points of user recorded frame image feature data. Use RANSAC (Random Sampling Consensus) algorithm or model fitting-based methods to identify and remove outliers in the match to improve the accuracy of the match. Use basic matrix-based or homography-based geometric verification methods to verify the geometric relationship between matching point pairs to further screen matching point pairs. Use optimization-based methods, such as optimization algorithms that minimize reprojection errors, such as Levenberg-Marquardt algorithm or gradient descent algorithm, to align matching point pairs to obtain the final frame image feature point matching pairs.

[0096] Preferably, step S24 includes the following steps:

[0097] Step S241: performing frame set blurriness comparison on a standard user video recording frame set and a user video display frame set to generate blurriness comparison analysis data;

[0098] Step S242: overlap the same frame images of the standard user video recording frame set and the user video display frame set by using the fuzziness comparison analysis data to generate overlapped frame images;

[0099] Step S243: performing non-overlapping region contour analysis on the overlapped frame images to generate non-overlapping region contour data;

[0100] Step S244: confirming the pixel coordinates of the non-overlapping area contour data to obtain the non-overlapping area contour pixel coordinate data; performing pixel coordinate offset comparison on the non-overlapping area contour pixel coordinate data to generate frame set comparison analysis data.

[0101] The present invention can quantitatively evaluate the clarity and quality of video images by performing a fuzziness comparison analysis on a standard user video recording frame set and a user video display frame set. This helps to detect whether the quality of the video display meets expectations and identify potential image blur problems. Based on the fuzziness comparison analysis data, the standard user video recording frame set and the user video display frame set are overlapped with the same frame image. This step can determine whether a frame synchronization problem occurs during the video display process, that is, whether the recorded frame and the displayed frame completely overlap. The overlapped frame image is subjected to a non-overlapping area contour analysis to determine the non-overlapping areas present during the video display process. These areas represent the displacement or deformation of the video frame during the display process. By confirming the non-overlapping area contour pixel coordinate data and comparing the pixel coordinates of the recorded frame and the displayed frame, the degree of displacement or deformation can be quantitatively measured. This helps to accurately locate problems in the video display process and provide quantitative evaluation data.

[0102] In an embodiment of the present invention, image data of a standard user video recording frame set and a user video display frame set are prepared. Blur analysis is performed on each frame image, and blur indexes such as average blur and variance can be calculated using methods such as image gradient and frequency domain analysis. Blur data of the recording frame set and the display frame set are compared and analyzed to generate blur comparison analysis data. Feature points such as SIFT and SURF are extracted from the recording frame set and the display frame set. The extracted feature points are matched to establish a correspondence between the frames. Based on the feature matching results, image transformation is performed on the recording frame set and the display frame set to achieve the overlap of the same frame images. The overlapped frame images are segmented to extract non-overlapping areas. Edge detection is performed on the non-overlapping areas to find the contours. The contour features of the non-overlapping areas, such as area and shape, are analyzed to generate contour data of the non-overlapping areas. The pixel coordinate data of the contour of the non-overlapping areas is determined. The offset of the pixel coordinates of the contour of the non-overlapping areas in the recording frame set and the display frame set is calculated. The pixel coordinate offset data is analyzed to generate frame set comparison analysis data for evaluating the difference and displacement between the frame sets.

[0103] Preferably, step S3 comprises the following steps:

[0104] Step S31: obtaining current network speed data; performing transmission delay loss calculation on the frame set comparison analysis data based on the current network speed data to obtain transmission delay data;

[0105] Step S32: performing frame set delay difference analysis on the frame set comparison analysis data according to the transmission delay data to generate frame set delay difference analysis data;

[0106] Step S33: reconstructing the display frame set of the standard user video recording frame set by analyzing the frame set delay difference data to generate a user video display reconstructed frame set;

[0107] Step S34: performing motion trajectory difference analysis on the user video display reconstruction frame set and the user video display frame set to generate motion trajectory difference analysis data.

[0108] The present invention obtains the current network speed data and calculates the transmission delay, which can help the system understand the network status and further optimize the mode and parameters of video transmission to reduce transmission delay. The frame set delay difference analysis is performed on the frame set comparison analysis data based on the transmission delay data, which helps to understand the impact of transmission delay on the frame set comparison data, and is used to further optimize the transmission strategy or compensate for delays in the display process. The display frame set is reconstructed for the standard user video recording frame set using the frame set delay difference analysis data. This is a compensation mechanism that compensates for the impact of network transmission delay by reconstructing the display frame set, thereby improving the user experience. The motion trajectory difference analysis is performed on the user video display reconstructed frame set and the user video display frame set, which is used to evaluate the difference between the reconstructed frame set and the original frame set, as well as the motion trajectory changes introduced during the reconstruction process, so as to further optimize the reconstruction algorithm or transmission strategy.

[0109] In an embodiment of the present invention, the network speed data of the current device is obtained by using a network speed test tool or API. The difference between two frame sets is compared using image processing and analysis algorithms, which usually involve image feature extraction, matching algorithms, etc. The transmission delay loss is calculated for the frame set comparison analysis data, involving parameters such as the delay of the network communication protocol and the size of the data packet. The transmission delay data is obtained by combining the frame set comparison analysis results and the transmission delay calculation results. The delay difference between the frame sets is statistically analyzed, including the average delay, delay distribution, etc. The analysis results are presented using data visualization tools, such as line charts and bar charts. According to the frame set delay difference analysis data, the standard user video recording frame set is reconstructed to compensate for the display delay caused by the transmission delay. The motion trajectory analysis is performed on the user video display reconstruction frame set and the user video display frame set, including target tracking, motion estimation, etc. The difference between the motion trajectories is calculated using a trajectory matching algorithm to generate motion trajectory difference analysis data.

[0110] Preferably, step S33 includes the following steps:

[0111] Step S331: performing inter-frame difference analysis on the frame set delay difference analysis data to obtain a delay value for each frame;

[0112] Step S332: sorting the delay values ​​of each frame in time order and constructing a delay difference data matrix;

[0113] Step S333: performing video frame interpolation based on the delay difference data matrix to generate missing frame data;

[0114] Step S334: performing missing frame segment analysis on the delay difference data matrix through the missing frame data to generate missing frame segments;

[0115] Step S335: performing frame complementation processing on the standard user video recording frame set according to the missing frame segments to generate a user video display reconstructed frame set.

[0116] The present invention can fill in the missing frames in the video through frame filling processing and interpolation algorithm, thereby improving the display quality and fluency of the video. By analyzing and processing the delayed data, the delay in the transmission process can be reduced, and the efficiency and speed of video transmission can be improved. Optimizing the video display effect can enhance the user's viewing experience and enhance the user's attraction and satisfaction with the video content.

[0117] In an embodiment of the present invention, the delay difference analysis data of the frame set is subjected to inter-frame difference, that is, the delay difference between adjacent frames is calculated. This can be achieved by calculating the time interval between adjacent frames, for example, subtracting the timestamp of the previous frame from the timestamp of the current frame to obtain the delay value between them. The delay values ​​of each frame are sorted in time order, and a delay difference data matrix is ​​constructed. This means that the calculated delay difference values ​​are sorted in time order and organized into a data matrix, in which each row represents a time point and each column represents the delay difference value of a frame. Video frame interpolation is performed based on the delay difference data matrix to generate missing frame data. This step involves using an interpolation algorithm to fill in the missing frames in the delay difference data matrix to ensure the integrity and continuity of the data. Common interpolation algorithms include linear interpolation, spline interpolation, etc. The missing frame segment analysis of the delay difference data matrix is ​​performed through the missing frame data to generate missing frame segments. In this step, it is necessary to identify the segments of continuous missing frames in the delay difference data matrix and mark them as missing frame segments for subsequent processing. The standard user video recording frame set is supplemented according to the missing frame segments to generate a user video display reconstruction frame set. Finally, according to the missing frame segment information previously identified, the standard user video recording frame set is supplemented accordingly to generate a user video display reconstruction frame set, thereby improving the smoothness and integrity of the video display.

[0118] Preferably, step S4 comprises the following steps:

[0119] Step S41: integrating the motion trajectory difference analysis data and the frame set comparison analysis data to obtain static recognition difference data; dividing the static recognition difference data into data sets to generate a model training set and a model test set;

[0120] Step S42: using a convolutional neural network algorithm to perform model training on the model training set to generate a virtual camera risk training model; using a model test set to perform model testing on the virtual camera risk training model to generate a virtual camera risk prediction model;

[0121] Step S43: importing the static recognition difference data into the virtual camera risk prediction model to perform risk prediction and generate static transmission risk prediction data;

[0122] Step S44: Entropy weight the static transmission risk prediction data and the dynamic transmission risk data to generate a virtual camera recognition weight; compare the virtual camera recognition weight with the preset standard virtual camera recognition weight threshold. When the virtual camera recognition weight is greater than or equal to the preset standard virtual camera recognition weight threshold, a virtual camera recognition alarm is issued to the mobile phone.

[0123] The present invention integrates static recognition difference data based on motion trajectory difference and frame set comparison analysis data: this means that the results of motion trajectory and frame set comparison analysis are comprehensively considered to obtain more comprehensive static recognition difference data. The data is divided to generate training sets and test sets: this is to prepare for subsequent model training and evaluation to ensure that the model has sufficient generalization ability and prediction accuracy. The convolutional neural network (CNN) algorithm is used for model training: CNN has a wide range of applications in the field of image recognition and processing, can effectively learn and extract features, and has a good effect on model training. Evaluate model performance through test sets: By evaluating the model through test sets, indicators such as model accuracy and recall rate can be obtained to ensure the reliability and effectiveness of the model. Use the trained model to predict risks: import static recognition difference data into the model to obtain static transmission risk prediction data. This step is a key step in applying the model to actual data and making predictions. Perform entropy weighted processing: combine static and dynamic transmission risk data, and perform weighted processing through entropy weight method to obtain the recognition weight of the virtual camera. Compare the recognition weight and the preset threshold: Compare the calculated recognition weight with the preset threshold. If the weight exceeds the threshold, the virtual camera recognition alarm prompt is triggered. This step is to compare the model prediction results with the preset rules to achieve real-time monitoring and warning of risks.

[0124] In an embodiment of the present invention, the motion trajectory is analyzed by using computer vision technology, such as an optical flow algorithm or a target tracking algorithm, to identify the motion trajectory. The frame set is compared and analyzed, and image processing technology, such as a feature matching algorithm (such as SIFT, SURF, etc.) or a deep learning method (such as a Siamese network) is used to detect the difference between images. The motion trajectory analysis data and the frame set comparison analysis data are integrated to obtain static recognition difference data. The static recognition difference data is divided into a model training set and a model test set using a cross-validation or random partitioning method. A convolutional neural network (CNN) model is constructed, including a convolutional layer, a pooling layer, and a fully connected layer. The model is trained using the training set data, and the model parameters are updated using a back propagation algorithm and an optimizer (such as SGD, Adam, etc.). The model performance is evaluated using the test set data, and indicators such as accuracy, recall rate, and F1 score are calculated to evaluate the generalization ability and prediction accuracy of the model. The static recognition difference data is input into the trained virtual camera risk prediction model, and the model is used for risk prediction to obtain static transmission risk prediction data. The static transmission risk prediction data and dynamic transmission risk data are entropy-weighted to comprehensively consider the importance of static and dynamic risk factors. The calculated virtual camera recognition weight is compared with the preset standard threshold to determine whether to trigger the virtual camera recognition alarm. If the conditions are met, the mobile phone will be prompted with a virtual camera recognition alarm to inform the user of the virtual camera risk.

[0125] In this specification, a virtual camera recognition device based on a mobile phone is provided, including a recognition controller for executing the above-mentioned virtual camera recognition method based on a mobile phone. The recognition controller of the virtual camera recognition device based on a mobile phone includes:

[0126] The transmission risk identification module is used to obtain the calling authority of the mobile phone camera; based on the mobile phone camera, the recognition controller calling authority is used to continuously record the user portrait to generate a user video frame set; the user video frame set is subjected to video frame set network transmission detection to generate a dynamic transmission user video frame set and a static user video frame set; the dynamic transmission user video frame set is subjected to transmission risk analysis to generate dynamic transmission risk data;

[0127] The frame set comparison module is used to acquire the device screen based on the static user video frame set to generate the device screen display interface; read the user display video on the device screen display interface to generate the user video display frame set; perform feature point distance matching on the standard user video recording frame set and the user video display frame set to obtain matching warning feature points; perform frame set comparison analysis on the standard user video recording frame set and the user video display frame set by matching the warning feature points to generate frame set comparison analysis data;

[0128] The delay analysis module is used to obtain the current network speed data; perform frame set delay difference analysis on the frame set comparison analysis data based on the current network speed data to generate frame set delay difference analysis data; perform motion trajectory difference analysis on the user video display frame set through the frame set delay difference analysis data to generate motion trajectory difference analysis data;

[0129] The risk prediction module is used to perform model training based on motion trajectory difference analysis data and frame set comparison analysis data to generate a virtual camera risk prediction model; import static recognition difference data into the virtual camera risk prediction model for risk prediction to generate static transmission risk prediction data; perform entropy weighting on the static transmission risk prediction data and the dynamic transmission risk data to generate a virtual camera recognition weight; compare the virtual camera recognition weight with a preset standard virtual camera recognition weight threshold, and when the virtual camera recognition weight is greater than or equal to the preset standard virtual camera recognition weight threshold, a virtual camera recognition alarm prompt is issued to the mobile phone.

[0130] The beneficial effect of the present invention is that through continuous user portrait recording and transmission risk analysis, the monitoring and identification capabilities of potential privacy leaks are improved, and user privacy protection is enhanced. The static user video frame set and device screen acquisition are used for difference analysis to effectively identify whether the device screen content is screenshotted or recorded to prevent security risks. The virtual camera risk prediction model generated by frame set comparison analysis and model training performs entropy weighting on static transmission risks and dynamic transmission risks to provide accurate virtual camera identification weights and alarm prompts. Using frame set delay difference analysis and motion trajectory difference analysis, combined with current network speed data, real-time monitoring and accurate identification of virtual camera recording behavior are achieved. By combining a variety of technical means, users can feel more at ease when using mobile phones to avoid privacy leaks and unsafe camera behaviors. Static identification difference data and frame set comparison analysis effectively prevent illegal behaviors such as screenshots and screen recordings, and protect user privacy and security. It integrates technologies such as continuous recording, difference analysis, and model training to improve the efficiency and accuracy of identifying potential privacy leaks and security risks. The virtual camera recognition weight is compared with the preset standard recognition weight threshold, allowing personalized risk assessment and providing customized safety tips for different user needs. Therefore, the present invention improves the accuracy of virtual camera recognition and improves the safety of users through technical means such as continuous recording, difference analysis, and model training.

[0131] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0132] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A virtual camera recognition method based on a mobile phone, characterized in that: The following steps are involved: Step S1: Obtain the mobile phone camera calling permission; Based on the mobile phone camera, continuous user portrait recording is performed using the recognition controller call authority to generate a user video frame set; video frame set network transmission detection is performed on the user video frame set to generate a dynamic transmission user video frame set and a static user video frame set; transmission risk analysis is performed on the dynamic transmission user video frame set to generate dynamic transmission risk data; step S1 includes the following steps: Step S11: Obtaining the mobile phone camera calling authority; Step S12: Based on the mobile phone camera, the recognition controller calls the authority to continuously record the user's portrait to generate a user video frame set; Step S13: performing frame image preprocessing on the user video frame set to generate a standard user video recording frame set, wherein the frame image preprocessing includes image brightness enhancement, image contrast enhancement and image denoising; Step S14: performing video frame set network transmission detection through the standard user video recording frame set, and generating a dynamic transmission user video frame set when it is detected that the standard user video recording frame set is being transmitted over the network; and generating a static user video frame set when it is not detected that the standard user video recording frame set is being transmitted over the network; Step S15: performing transmission signal analysis on the dynamic transmission user video frame set to generate dynamic transmission signal analysis data; performing transmission risk analysis on the dynamic transmission signal analysis data to generate dynamic transmission risk data; Step S2: acquiring the device screen based on the static user video frame set to generate a device screen display interface; reading the user display video on the device screen display interface to generate a user video display frame set; performing feature point distance matching on the standard user video recording frame set and the user video display frame set to obtain matching warning feature points; performing frame set comparison analysis on the standard user video recording frame set and the user video display frame set by matching warning feature points to generate frame set comparison analysis data; Step S3: obtaining current network speed data; performing frame set delay difference analysis on the frame set comparison analysis data based on the current network speed data to generate frame set delay difference analysis data; performing motion trajectory difference analysis on the user video display frame set through the frame set delay difference analysis data to generate motion trajectory difference analysis data; Step S4: Perform model training based on motion trajectory difference analysis data and frame set comparison analysis data to generate a virtual camera risk prediction model; import static recognition difference data into the virtual camera risk prediction model for risk prediction to generate static transmission risk prediction data; perform entropy weighting on the static transmission risk prediction data and the dynamic transmission risk data to generate a virtual camera recognition weight; compare the virtual camera recognition weight with a preset standard virtual camera recognition weight threshold, and when the virtual camera recognition weight is greater than or equal to the preset standard virtual camera recognition weight threshold, issue a virtual camera recognition alarm prompt to the mobile phone; Step S4 includes the following steps: Step S41: integrating the motion trajectory difference analysis data and the frame set comparison analysis data to obtain static recognition difference data; dividing the static recognition difference data into data sets to generate a model training set and a model test set; Step S42: using a convolutional neural network algorithm to perform model training on the model training set to generate a virtual camera risk training model; using a model test set to perform model testing on the virtual camera risk training model to generate a virtual camera risk prediction model; Step S43: importing the static recognition difference data into the virtual camera risk prediction model to perform risk prediction and generate static transmission risk prediction data; Step S44: Entropy weight the static transmission risk prediction data and the dynamic transmission risk data to generate a virtual camera recognition weight; compare the virtual camera recognition weight with the preset standard virtual camera recognition weight threshold. When the virtual camera recognition weight is greater than or equal to the preset standard virtual camera recognition weight threshold, a virtual camera recognition alarm is issued to the mobile phone.

2. The method for identifying a virtual camera based on a mobile phone according to claim 1, characterized in that: Step S15 includes the following steps: Step S151: performing transmission signal detection on the dynamically transmitted user video frame set, and when it is detected that the dynamically transmitted user video frame set is transmitted as an input signal, performing camera device code recognition on the mobile phone according to the dynamically transmitted user video frame set to generate input signal detection data; Step S152: when it is detected that the dynamic transmission user video frame set is transmitted as the output signal, the transmission device machine code is identified according to the dynamic transmission user video frame set to obtain the transmission device machine code; Step S153: Perform camera digital signature authentication by transmitting the device machine code to generate a camera digital signature; perform signature matching between the abnormal camera digital signature and a preset virtual camera digital signature library to generate output signal detection data; Step S154: integrating the input signal detection data and the output signal detection data to generate dynamic transmission signal analysis data; performing transmission risk analysis on the dynamic transmission signal analysis data to generate dynamic transmission risk data.

3. The method for identifying a virtual camera based on a mobile phone according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: acquiring the device screen based on the static user video frame set to generate a device screen display interface; segmenting the device screen display interface into user display frames to generate user display frames; reading the user display video from the user display frames to generate a user video display frame set; Step S22: extracting frame image feature points from the standard user video recording frame set to obtain user recording frame image feature data; performing frame image feature point matching on the user video display frame set according to the user recording frame image feature data to generate a frame image feature point matching pair; Step S23: performing matching distance difference analysis on the frame image feature point matching pairs to generate frame image feature point matching difference distance data; comparing the frame image feature point matching difference distance data with a preset standard feature point matching difference distance threshold, and when the frame image feature point matching difference distance data is less than the standard feature point matching difference distance threshold, generating a matching warning feature point; Step S24: performing frame set comparison analysis on the standard user video recording frame set and the user video display frame set by matching the warning feature points to generate frame set comparison analysis data.

4. The method for identifying a virtual camera based on a mobile phone according to claim 3, characterized in that: Step S22 includes the following steps: Step S221: performing background separation on a standard user video recording frame set to generate background separation data; performing background color feature marking on the background separation data to generate background color feature data; Step S222: performing portrait separation on the standard user video recording frame set using the background separation data to generate portrait separation data; Step S223: marking facial feature points of the standard user video recording frame set according to the portrait separation data to generate facial feature data of the portrait; Step S224: integrating the background color feature data and the facial features feature data to generate the user recorded frame image feature data; performing frame image feature point matching on the user video display frame set according to the user recorded frame image feature data to generate a frame image initial feature point matching pair; Step S225: performing external point elimination on the initial feature point matching pairs of the frame image to generate frame image eliminated feature point matching pairs; performing geometric verification on the frame image eliminated feature point matching pairs to generate feature point geometric verification data; Step S226: using the feature point geometry verification data, perform feature point registration on the frame image after removing the feature point matching pairs, to generate the frame image feature point matching pairs.

5. The method for identifying a virtual camera based on a mobile phone according to claim 3, characterized in that: Step S24 includes the following steps: Step S241: performing frame set blurriness comparison on a standard user video recording frame set and a user video display frame set to generate blurriness comparison analysis data; Step S242: overlap the same frame images of the standard user video recording frame set and the user video display frame set by using the fuzziness comparison analysis data to generate overlapped frame images; Step S243: performing non-overlapping region contour analysis on the overlapped frame images to generate non-overlapping region contour data; Step S244: confirming the pixel coordinates of the non-overlapping area contour data to obtain the non-overlapping area contour pixel coordinate data; performing pixel coordinate offset comparison on the non-overlapping area contour pixel coordinate data to generate frame set comparison analysis data.

6. The method for identifying a virtual camera based on a mobile phone according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: obtaining current network speed data; performing transmission delay loss calculation on the frame set comparison analysis data based on the current network speed data to obtain transmission delay data; Step S32: performing frame set delay difference analysis on the frame set comparison analysis data according to the transmission delay data to generate frame set delay difference analysis data; Step S33: reconstructing the display frame set of the standard user video recording frame set by analyzing the frame set delay difference data to generate a user video display reconstructed frame set; Step S34: performing motion trajectory difference analysis on the user video display reconstruction frame set and the user video display frame set to generate motion trajectory difference analysis data.

7. The method for identifying a virtual camera based on a mobile phone according to claim 6, characterized in that: Step S33 includes the following steps: Step S331: performing inter-frame difference analysis on the frame set delay difference analysis data to obtain a delay value for each frame; Step S332: sorting the delay values ​​of each frame in time order and constructing a delay difference data matrix; Step S333: performing video frame interpolation based on the delay difference data matrix to generate missing frame data; Step S334: performing missing frame segment analysis on the delay difference data matrix through the missing frame data to generate missing frame segments; Step S335: performing frame complementation processing on the standard user video recording frame set according to the missing frame segments to generate a user video display reconstructed frame set.

8. A virtual camera recognition device based on a mobile phone, characterized in that: The invention comprises a recognition controller for executing the mobile phone-based virtual camera recognition method as claimed in claim 1, wherein the recognition controller of the mobile phone-based virtual camera recognition device comprises: The transmission risk identification module is used to obtain the calling authority of the mobile phone camera; based on the mobile phone camera, the recognition controller calling authority is used to continuously record the user portrait to generate a user video frame set; the user video frame set is subjected to video frame set network transmission detection to generate a dynamic transmission user video frame set and a static user video frame set; the dynamic transmission user video frame set is subjected to transmission risk analysis to generate dynamic transmission risk data; The frame set comparison module is used to acquire the device screen based on the static user video frame set to generate the device screen display interface; read the user display video on the device screen display interface to generate the user video display frame set; perform feature point distance matching on the standard user video recording frame set and the user video display frame set to obtain matching warning feature points; perform frame set comparison analysis on the standard user video recording frame set and the user video display frame set by matching the warning feature points to generate frame set comparison analysis data; The delay analysis module is used to obtain the current network speed data; perform frame set delay difference analysis on the frame set comparison analysis data based on the current network speed data to generate frame set delay difference analysis data; perform motion trajectory difference analysis on the user video display frame set through the frame set delay difference analysis data to generate motion trajectory difference analysis data; The risk prediction module is used to perform model training based on motion trajectory difference analysis data and frame set comparison analysis data to generate a virtual camera risk prediction model; import static recognition difference data into the virtual camera risk prediction model for risk prediction to generate static transmission risk prediction data; perform entropy weighting on the static transmission risk prediction data and the dynamic transmission risk data to generate a virtual camera recognition weight; compare the virtual camera recognition weight with a preset standard virtual camera recognition weight threshold, and when the virtual camera recognition weight is greater than or equal to the preset standard virtual camera recognition weight threshold, a virtual camera recognition alarm prompt is issued to the mobile phone.

Images