A video synchronization method based on image multi-dimensional feature value comparison

Abstract

The present invention relates to the field of image communication, and in particular to a video synchronization method based on image multi-dimensional feature value comparison. The method comprises the following steps: obtaining the original comparison video source, and performing discrete wavelet transform and image enhancement processing on it to generate enhanced video frames with higher quality; performing multi-dimensional feature encoding on the enhanced video frames to generate video feature vectors, and simulating protective transmission to obtain analog comparison videos at the receiving end; screening preliminary synchronization frame pairs by preset matching frame pair thresholds, and performing timestamp correction to generate accurate correction synchronization frame pairs. Based on the correction synchronization frame pairs, the enhanced video frames and the analog video at the receiving end are synchronized in the time domain and the feature coordinate space is compared. When the feature coordinates are consistent, a template error correction synchronization method is adopted; when they are inconsistent, a correction synchronization method is adopted; the present invention can achieve high-precision and high-stability video synchronization.

Description

Technical Field

[0001] The invention relates to the field of image communication, and in particular to a video synchronization method based on image multi-dimensional feature value comparison. Background Art

[0002] In the early stage, from the 1990s to the early 2000s, video synchronization methods mainly used some simple visual features, such as pixel grayscale values ​​and histograms, for frame-level synchronization. Although this method is relatively simple to implement, it is very sensitive to noise, illumination changes, and occlusion interference factors, and the synchronization accuracy is low, which is difficult to meet the needs of practical applications. In the mid-stage, from the mid-2000s to the mid-2010s, researchers began to use methods based on key point detection and feature descriptors to extract more robust visual features, such as SIFT and SURF, for video synchronization. This method has improved the accuracy and robustness of synchronization to some extent, but there are still some problems for dynamic scenes and complex occlusion situations, and the synchronization performance needs to be further improved. In the recent stage, with the rise of deep learning technology, researchers have begun to use CNN and RNN network structures to extract more robust high-dimensional visual features for video synchronization tasks. This deep learning-based method has overcome the limitations of previous methods to a certain extent, but it also faces the problems of requiring a large amount of labeled data for network training, limited generalization ability, and still facing challenges for complex scenes. Summary of the invention

[0003] Based on this, it is necessary to provide a video synchronization method based on image multi-dimensional feature value comparison to solve at least one of the above technical problems.

[0004] To achieve the above object, a video synchronization method based on image multi-dimensional feature value comparison includes the following steps:

[0005] Step S1: obtaining a comparison video source; performing discrete wavelet transform on the comparison video source to generate multi-frequency domain component data; performing image enhancement processing on the multi-frequency domain component data to generate an enhanced video frame;

[0006] Step S2: Perform multi-dimensional feature encoding on the enhanced video frame to generate a video feature vector; perform protective transmission simulation on the video feature vector to obtain a receiving end simulation comparison video;

[0007] Step S3: Screening the synchronization frame pairs of the analog comparison video at the receiving end based on a preset matching frame pair threshold to obtain preliminary synchronization frame pairs; performing timestamp correction on the preliminary synchronization frame pairs to generate corrected synchronization frame pairs;

[0008] Step S4: performing time domain synchronous coordinate positioning of the enhanced video frame and the analog comparison video at the receiving end by correcting the synchronous frame pair, generating enhanced video frame feature coordinates and analog comparison video feature coordinates; performing feature space position comparison of the enhanced video frame feature coordinates and the analog comparison video feature coordinates, and obtaining feature coordinate space matching results;

[0009] Step S5: when the enhanced video frame feature coordinates are equal to the simulated comparison video feature coordinates, the feature coordinate space matching result is defined as the feature space synchronization result, and based on the feature space synchronization result and the enhanced video frame feature coordinate comparison, the video source is subjected to video template synchronization communication to execute the template error correction type video synchronization method;

[0010] Step S6: When the enhanced video frame feature coordinates are not equal to the simulated comparison video feature coordinates, the feature coordinate space matching result is defined as a feature space asynchronous result, and based on the feature space asynchronous result and the enhanced video frame feature coordinate comparison, the video source is subjected to video correction synchronization communication to execute a correction type video synchronization method.

[0011] The present invention obtains multi-frequency domain components of the comparison video through discrete wavelet transform, performs image enhancement processing on these components, and generates richer enhanced video frames. Multi-dimensional feature encoding is performed on the enhanced video frame to generate a video feature vector that can be used for subsequent comparison, thereby improving the comparison accuracy. Timestamp correction is performed on the preliminary synchronization frame pair, and the synchronization deviation is dynamically adjusted to ensure the time alignment of the correction synchronization frame pair. The feature coordinates of the enhanced video frame and the simulated comparison video are compared, and it is determined whether the synchronization state is reached through feature matching in the spatial dimension. According to the feature space comparison result, different synchronization methods are adopted to realize adaptive synchronization template correction or synchronization deviation correction. The multi-dimensional features of the image are fully utilized, and the accurate synchronization between the enhanced video and the simulated comparison video is realized through adaptive time and space calibration. The synchronization method driven by the feature value comparison can better adapt to the complex video source environment, and improve the synchronization effect and stability. Therefore, the video synchronization method based on the multi-dimensional feature value comparison of the image makes full use of image enhancement, feature encoding and synchronization correction technology, and realizes high-precision video synchronization while ensuring video quality and transmission security.

[0012] The invention is beneficial in that the original comparison video source is analyzed in the frequency domain by using discrete wavelet transform, and the multi-frequency domain component data is image enhanced, thereby improving the quality of the video frame and the information reliability. The enhanced video frame is multi-dimensionally encoded to generate a video feature vector, and the protective transmission is simulated to ensure the integrity of the feature information. Compared with directly transmitting the video stream, this feature encoding and transmission method can effectively reduce the transmission bandwidth and improve the transmission stability. The preliminary screening is performed by presetting the matching frame pair threshold to obtain the preliminary synchronization frame pair, and further by timestamp correction, the correction synchronization frame pair is generated, thereby improving the accuracy of the synchronization frame pair. Based on the correction synchronization frame pair, the enhanced video frame and the receiving end simulated comparison video are compared and synchronized in the feature space position, thereby realizing high-precision video synchronization. When the feature space is not synchronized, the template error correction type or correction type video synchronization method can be used to automatically correct the synchronization deviation, thereby improving the robustness of the video synchronization. The image preprocessing, feature coding transmission, synchronization screening, and space matching links are organically integrated to realize the full automation of video synchronization, thereby greatly improving the work efficiency. Therefore, this video synchronization method based on image multi-dimensional feature value comparison makes full use of image enhancement, feature coding and synchronization correction technology to achieve high-precision video synchronization while ensuring video quality and transmission security. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A schematic diagram of the steps of a video synchronization method based on image multi-dimensional feature value comparison;

[0014] Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart;

[0015] Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0016] 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

[0017] The technical method of the present invention is described clearly and completely below 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.

[0018] 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.

[0019] 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.

[0020] To achieve this, please refer to Figures 1 to 3 , a video synchronization method based on image multi-dimensional feature value comparison, comprising the following steps:

[0021] Step S1: obtaining a comparison video source; performing discrete wavelet transform on the comparison video source to generate multi-frequency domain component data; performing image enhancement processing on the multi-frequency domain component data to generate an enhanced video frame;

[0022] Step S2: Perform multi-dimensional feature encoding on the enhanced video frame to generate a video feature vector; perform protective transmission simulation on the video feature vector to obtain a receiving end simulation comparison video;

[0023] Step S3: Screening the synchronization frame pairs of the analog comparison video at the receiving end based on a preset matching frame pair threshold to obtain preliminary synchronization frame pairs; performing timestamp correction on the preliminary synchronization frame pairs to generate corrected synchronization frame pairs;

[0024] Step S4: performing time domain synchronous coordinate positioning of the enhanced video frame and the analog comparison video at the receiving end by correcting the synchronous frame pair, generating enhanced video frame feature coordinates and analog comparison video feature coordinates; performing feature space position comparison of the enhanced video frame feature coordinates and the analog comparison video feature coordinates, and obtaining feature coordinate space matching results;

[0025] Step S5: when the enhanced video frame feature coordinates are equal to the simulated comparison video feature coordinates, the feature coordinate space matching result is defined as the feature space synchronization result, and based on the feature space synchronization result and the enhanced video frame feature coordinate comparison, the video source is subjected to video template synchronization communication to execute the template error correction type video synchronization method;

[0026] Step S6: When the enhanced video frame feature coordinates are not equal to the simulated comparison video feature coordinates, the feature coordinate space matching result is defined as a feature space asynchronous result, and based on the feature space asynchronous result and the enhanced video frame feature coordinate comparison, the video source is subjected to video correction synchronization communication to execute a correction type video synchronization method.

[0027] The present invention obtains multi-frequency domain components of the comparison video through discrete wavelet transform, performs image enhancement processing on these components, and generates richer enhanced video frames. Multi-dimensional feature encoding is performed on the enhanced video frame to generate a video feature vector that can be used for subsequent comparison, thereby improving the comparison accuracy. Timestamp correction is performed on the preliminary synchronization frame pair, and the synchronization deviation is dynamically adjusted to ensure the time alignment of the correction synchronization frame pair. The feature coordinates of the enhanced video frame and the simulated comparison video are compared, and it is determined whether the synchronization state is reached through feature matching in the spatial dimension. According to the feature space comparison result, different synchronization methods are adopted to realize adaptive synchronization template correction or synchronization deviation correction. The multi-dimensional features of the image are fully utilized, and the accurate synchronization between the enhanced video and the simulated comparison video is realized through adaptive time and space calibration. The synchronization method driven by the feature value comparison can better adapt to the complex video source environment, and improve the synchronization effect and stability. Therefore, the video synchronization method based on the multi-dimensional feature value comparison of the image makes full use of image enhancement, feature encoding and synchronization correction technology, and realizes high-precision video synchronization while ensuring video quality and transmission security.

[0028] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a video synchronization method based on image multi-dimensional feature value comparison of the present invention. In this example, the video synchronization method based on image multi-dimensional feature value comparison includes the following steps:

[0029] Step S1: obtaining a comparison video source; performing discrete wavelet transform on the comparison video source to generate multi-frequency domain component data; performing image enhancement processing on the multi-frequency domain component data to generate an enhanced video frame;

[0030] In an embodiment of the present invention, first, a comparison video source is obtained through a comparison system. Then, a discrete wavelet transform (DWT) algorithm is applied to process the comparison video source to decompose the video signal into multiple frequency domain components. In this process, the video signal is decomposed using wavelet functions such as Haar, Daubechies or Symlet to obtain low-frequency domain and high-frequency domain component data. Then, image enhancement processing is performed on these frequency domain component data, and the specific steps include applying histogram equalization to the low-frequency domain components to improve their contrast and brightness, and edge enhancement processing is performed on the high-frequency domain components, such as using a high-pass filter to increase the details and clarity of the image. Finally, these processing results are synthesized into an enhanced video frame to improve the visual effect and information clarity of the video.

[0031] Step S2: Perform multi-dimensional feature encoding on the enhanced video frame to generate a video feature vector; perform protective transmission simulation on the video feature vector to obtain a receiving end simulation comparison video;

[0032] In an embodiment of the present invention, a multi-dimensional feature coding technology is first applied to the enhanced video frame, for example, the color histogram, texture features and shape features of the image are extracted, and features are extracted using algorithms such as SIFT, SURF or ORB, and these feature data are encoded into a video feature vector. After feature coding, the generated video feature vector is used to simulate the protective transmission process. A transmission simulation tool is used to simulate packet loss, noise interference or data loss during data transmission to obtain a simulated comparison video at the receiving end. This simulation process is performed based on a channel model, such as adding noise or introducing a data loss model to ensure the robustness and reliability of feature transmission.

[0033] Step S3: Screening the synchronization frame pairs of the analog comparison video at the receiving end based on a preset matching frame pair threshold to obtain preliminary synchronization frame pairs; performing timestamp correction on the preliminary synchronization frame pairs to generate corrected synchronization frame pairs;

[0034] In an embodiment of the present invention, synchronization frame pairs are screened based on a preset matching frame pair threshold. The specific operation includes matching feature points in the analog comparison video of the receiving end and calculating the matching value. The matching value is compared with the preset threshold. When the matching value is greater than the threshold, it is considered to be a valid synchronization frame pair, and these preliminary synchronization frame pairs are screened out. Next, the preliminary synchronization frame pairs are corrected by using timestamps, and the timestamp information of the frame pairs is adjusted to correct the time deviation caused by transmission delay or timing disorder. The frame pairs are corrected by timestamps to obtain corrected synchronization frame pairs, thereby ensuring the time synchronization accuracy of the video.

[0035] Step S4: performing time domain synchronous coordinate positioning of the enhanced video frame and the analog comparison video at the receiving end by correcting the synchronous frame pair, generating enhanced video frame feature coordinates and analog comparison video feature coordinates; performing feature space position comparison of the enhanced video frame feature coordinates and the analog comparison video feature coordinates, and obtaining feature coordinate space matching results;

[0036] In an embodiment of the present invention, the time domain synchronous coordinate positioning of the enhanced video frame and the analog comparison video of the receiving end is performed using the correction synchronization frame pair. The specific operation includes applying the time domain coordinate positioning algorithm, such as the geometric transformation method based on image registration, to accurately align the coordinates of the feature points in the enhanced video frame and the analog comparison video. The generated enhanced video frame feature coordinates and the analog comparison video feature coordinates will be used for feature space position comparison. By calculating the spatial distance and matching degree between the feature coordinates, their similarity is evaluated, and the feature coordinate space matching result is obtained. The core of this step is to ensure the spatial consistency of the feature points in different video sources through precise coordinate alignment.

[0037] Step S5: when the enhanced video frame feature coordinates are equal to the simulated comparison video feature coordinates, the feature coordinate space matching result is defined as the feature space synchronization result, and based on the feature space synchronization result and the enhanced video frame feature coordinate comparison, the video source is subjected to video template synchronization communication to execute the template error correction type video synchronization method;

[0038] In the embodiment of the present invention, when the feature coordinate space matching result shows that the feature coordinates of the enhanced video frame are equal to the feature coordinates of the simulated comparison video, it is defined as a feature space synchronization result. Based on this result, the feature coordinates of the enhanced video frame are used to compare the video source for video template synchronization communication. The specific operation includes creating a synchronization template and applying it to the comparison video source to correct errors caused by time or space asynchronization. The synchronization of the video source is achieved through a template matching algorithm, such as matching based on image feature points or global image alignment technology.

[0039] Step S6: When the enhanced video frame feature coordinates are not equal to the simulated comparison video feature coordinates, the feature coordinate space matching result is defined as a feature space asynchronous result, and based on the feature space asynchronous result and the enhanced video frame feature coordinate comparison, the video source is subjected to video correction synchronization communication to execute a correction type video synchronization method.

[0040] In an embodiment of the present invention, when the feature coordinate space matching result shows that the feature coordinates of the enhanced video frame are not equal to the feature coordinates of the simulated comparison video, it is defined as a feature space asynchronous result. At this time, based on this asynchronous result, the feature coordinates of the enhanced video frame are subjected to video correction synchronization communication. The specific operation includes constructing a correction synchronization template according to the asynchronous result, and the template is used to adjust the feature coordinates in the comparison video source. A correction technology, such as geometric transformation correction or image adjustment algorithm, is used to correct the deviation caused by the synchronization error.

[0041] Preferably, step S1 comprises the following steps:

[0042] Step S11: Collecting the data stream in the comparison system to obtain a comparison video source;

[0043] Step S12: performing inter-frame image cutting on the comparison video source to obtain a comparison video frame;

[0044] Step S13: performing discrete wavelet transform on the video frames to generate multi-frequency domain component data, wherein the multi-frequency domain component data includes low-frequency domain components and high-frequency domain components;

[0045] Step S14: performing histogram grayscale conversion on the low-frequency domain component to obtain a low-frequency domain enhanced video frame;

[0046] Step S15: performing adaptive directional filtering on the high-frequency domain components to generate a high-frequency domain enhanced video frame;

[0047] Step S16: performing weighted fusion on the low-frequency domain enhanced video frame and the high-frequency domain enhanced video frame to generate an enhanced video frame.

[0048] The benefit of the present invention is that it can effectively extract key images in the comparison video, lay the foundation for subsequent frequency domain analysis and enhancement processing, reduce unnecessary data processing, and improve the computational efficiency of the overall algorithm. The video frame can be divided into low-frequency domain and high-frequency domain components containing different frequency band information, which provides a basis for targeted enhancement processing and is conducive to making full use of the multi-dimensional feature information in the video. The brightness and contrast of the low-frequency domain video frame can be significantly improved, the detail information can be enhanced, and more abundant and accurate feature data can be provided for the subsequent multi-dimensional feature value comparison. The important edge features in the high-frequency domain can be highlighted, the clarity of the video frame can be enhanced, and the feature information of the video frame can be further enriched, which is conducive to the subsequent precise synchronization. The advantages of low-frequency domain and high-frequency domain enhancement are fully integrated to generate enhanced video frames with higher quality, laying a good foundation for multi-dimensional feature value comparison. It embodies the organic combination of targeted image processing technology and video synchronization algorithm, thereby effectively improving the reliability and accuracy of the synchronization process.

[0049] In an embodiment of the present invention, the real-time data stream in the comparison system is transmitted to a data processing center via a wired or wireless network. In the data processing center, an efficient video acquisition software such as FFmpeg is used to capture and record the data stream. The video acquisition software is set to a continuous capture mode to ensure the integrity and real-time performance of the data stream. In the process of inter-frame image cutting of the video source, a video processing library such as OpenCV is used to read the video file. A fixed number of frames are extracted from the video stream per second, for example, 30 frames are extracted per second. The extracted video frames are stored as an image sequence, and each frame is used as a separate image file. Discrete wavelet transform is performed on the video frames using image processing tools such as MATLAB or Python's PyWavelets library. The specific steps include selecting an appropriate wavelet basis function, such as Haar or Daubechies wavelet, and performing a two-dimensional discrete wavelet transform on each video frame. Discrete wavelet transform decomposes the video frame into multi-frequency domain component data, including low-frequency domain components and high-frequency domain components. When processing the low-frequency domain component, the grayscale histogram of the low-frequency domain component is first calculated using an image histogram calculation tool such as the calcHist function of OpenCV. Then, the grayscale histogram of the low-frequency domain component is converted by applying a histogram equalization technique, such as the equalizeHist function of OpenCV. In this way, the contrast of the image is enhanced to obtain a low-frequency domain enhanced video frame. When the high-frequency domain component is subjected to adaptive directional filtering, an image directional filtering technique such as a Gabor filter or a directional guided filter is used. The specific operation includes selecting appropriate filter parameters, such as the frequency and directional parameters of the Gabor filter. The high-frequency domain component is subjected to directional filtering to extract edge and texture details in the image. The processed high-frequency domain component is the high-frequency domain enhanced video frame. When the low-frequency domain enhanced video frame and the high-frequency domain enhanced video frame are weightedly fused, a weighted average algorithm is used. The specific steps include determining appropriate weighting coefficients, such as setting the weights of the low-frequency domain and the high-frequency domain to 0.6 and 0.4, respectively. A weighted average calculation is performed on each pixel to generate an enhanced video frame.

[0050] Preferably, step S15 comprises the following steps:

[0051] Step S151: performing pyramid decomposition layer by layer on the high-frequency domain components to obtain multi-scale high-frequency domain components;

[0052] Step S152: Calculate the directional gradient of the multi-scale high-frequency domain components to generate component directional detail information;

[0053] Step S153: performing multi-directional filtering processing on the high-frequency domain components based on the component directional detail information to obtain multi-directional filtering response information;

[0054] Step S154: weight mapping is performed on the multi-directional filtering response information to obtain weight data for each direction;

[0055] Step S155: performing weight distribution directional filtering processing on the high-frequency domain components based on the weight data of each direction to generate a high-frequency domain enhanced video frame.

[0056] The benefit of the present invention lies in that it can effectively extract detail information of different scales in the high-frequency domain components, and provide necessary input data for subsequent directional gradient calculation and multi-directional filtering processing. It can accurately obtain the detail features of each direction in the high-frequency domain components, and perform targeted filtering and enhancement on the detail features in different directions, which can effectively highlight the important edge information in the video frame and improve the clarity of the picture. Reasonable weight data is calculated based on the filtering response information in each direction, which provides a basis for the final weighted fusion processing and ensures the scientific nature of the processing results. The weight information in different directions is fully utilized to perform adaptive filtering on the high-frequency domain components to generate clearer high-frequency domain enhanced video frames that highlight edge details. The processing of high-frequency domain components is further enriched and optimized, and the important feature information in the video frame is better highlighted, providing more reliable basic data for subsequent multi-dimensional feature value comparison, thereby improving the accuracy and robustness of video synchronization.

[0057] In an embodiment of the present invention, an image processing tool such as MATLAB or Python's OpenCV library is used to perform a pyramid decomposition operation on the high-frequency domain components layer by layer. The specific method includes using a Gaussian pyramid decomposition technique to downsample the high-frequency domain components layer by layer to obtain image components of different scales. The image size contained in each layer of the pyramid is reduced in sequence, and high-frequency domain information of different scales is gradually extracted. This process retains the high-frequency details in the image and generates multi-scale high-frequency domain components. When calculating the directional gradient of the multi-scale high-frequency domain components, a Sobel operator or a Scharr operator gradient calculation method is used. The specific steps include applying a directional gradient operator to the high-frequency domain components of each scale to calculate the gradient components of the image in the horizontal and vertical directions. Combining the gradients in these two directions, a directional detail information image of each scale is generated. Based on the component directional detail information, when performing multi-directional filtering processing on the high-frequency domain components, a Gabor filter or a directional guide filter is used. The specific steps include selecting filter parameters of different directions for the high-frequency domain components of each scale, and applying these filters to perform convolution operations on the high-frequency domain components. The filtering result in each direction reflects the detail information in that direction. When weight mapping the multi-directional filtering response information, a weight calculation method based on the gradient amplitude is used. The specific operation includes calculating the gradient amplitude of the filter response in each direction, normalizing these amplitudes, and generating weight mapping data. By setting a threshold or weighting function, the weight data is further adjusted to reflect the importance of different directions. Based on the weight data of each direction, when performing weight distribution directional filtering processing on the high-frequency domain component, the multi-directional filter response information and the weight data of each direction are combined. The specific operation includes weighting the filter response information of each direction and the corresponding weight data to generate a weighted directional filtering result. The weighted results of all directions are superimposed to obtain the final high-frequency domain enhanced video frame.

[0058] Preferably, step S2 comprises the following steps:

[0059] Step S21: extracting multi-dimensional features from the enhanced video frame to obtain a video feature data set;

[0060] Step S22: performing vector encoding on the video feature data set to generate a video feature vector;

[0061] Step S23: Building a simulation transmission channel for the enhanced video frame to generate a simulation transmission channel;

[0062] Step S24: Based on the simulated transmission channel and the video feature vector, feature protection transmission simulation is performed on the enhanced video frame to obtain a receiving end simulated comparison video.

[0063] The benefit of the present invention is that it can comprehensively extract a variety of visual features in the video frame, such as color, texture, and edge, providing rich basic data for subsequent feature comparison and synchronization. The multidimensional feature data is encoded into a compact feature vector form, which improves the efficiency of data storage and transmission and provides convenience for subsequent feature comparison. A simulated transmission channel environment is constructed to simulate the actual network transmission conditions, which helps to evaluate the loss and interference that the feature data will suffer during the transmission process. Under the simulated transmission channel conditions, the robustness and accuracy of the feature data are tested, providing a reliable data basis for subsequent actual feature comparison and synchronization. The video feature data can be comprehensively prepared and verified to ensure that the subsequent feature comparison and synchronization process is more stable and reliable. This is of great significance for improving the accuracy and practicality of video synchronization.

[0064] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0065] Step S21: extracting multi-dimensional features from the enhanced video frame to obtain a video feature data set;

[0066] In an embodiment of the present invention, image processing and machine learning tools such as OpenCV and scikit-learn are used to perform multi-dimensional feature extraction on enhanced video frames. The specific operation includes first applying an edge detection algorithm such as the Canny algorithm to each frame to extract edge features. Then, the color distribution of the image is analyzed using a color histogram to generate color features. In addition, key points and descriptors are extracted by using the SIFT (Scale-Invariant Feature Transform) or SURF (Speeded-Up Robust Features) algorithm to obtain local features. These features are combined to form a multi-dimensional feature vector, which is stored in a video feature data set to provide detailed feature information for subsequent processing.

[0067] Step S22: performing vector encoding on the video feature data set to generate a video feature vector;

[0068] In an embodiment of the present invention, when vector encoding is performed on a video feature data set, a PCA (Principal Component Analysis) or t-SNE (t-distributed Stochastic Neighbor Embedding) dimensionality reduction algorithm is used. The specific steps include standardizing the feature data so that its mean is 0 and its variance is 1. Then, the PCA algorithm is applied to reduce the high-dimensional feature vector to a lower dimension while retaining the main information of the feature. The generated low-dimensional feature vector is the video feature vector.

[0069] Step S23: Building a simulation transmission channel for the enhanced video frame to generate a simulation transmission channel;

[0070] In the embodiment of the present invention, when building a simulated transmission channel, a network simulation tool such as ns-3 or OMNeT++ is used. The specific operation includes defining the network topology, such as setting the transmission path and bandwidth limit. Then, the data transmission in the real network environment is simulated, and the transmission delay, packet loss rate and noise interference parameters are set. Through these settings, a simulated channel that can truly simulate video transmission is generated.

[0071] Step S24: Based on the simulated transmission channel and the video feature vector, feature protection transmission simulation is performed on the enhanced video frame to obtain a receiving end simulated comparison video.

[0072] In an embodiment of the present invention, error correction coding technology such as LDPC code or Turbo code is used to enhance the error correction coding of the video feature vector. The specific operation includes first selecting an appropriate error correction coding scheme and setting coding parameters according to the length of the video feature vector and the transmission environment. Then, the error correction coding algorithm is applied to convert the video feature vector into a protective feature vector, and redundant information is added to improve the anti-interference ability. Through error correction coding, the generated protective feature vector can effectively detect and correct errors during transmission to ensure the integrity of the data. The enhanced video frame is optimized for feature protection using an optimization algorithm based on the protective feature vector. The specific operation includes associating the protective feature vector with the enhanced video frame, and embedding the redundant information of the feature vector into the video frame through weighted fusion or embedding technology. This process embeds redundant feature information in the video frame, so that the original features of the video frame can be restored through the redundant information even if it is interfered during transmission, and a feature protection video frame is generated. A network simulation tool such as ns-3 or OMNeT++ is used to simulate the transmission of the feature protection video frame. The specific steps include setting the transmission path and parameters such as bandwidth, delay, and packet loss rate in the simulation environment to simulate real network transmission conditions. The characteristic protective video frames are transmitted through the simulated transmission channel, and the packet loss, delay and bit error phenomena that occur during the transmission process are recorded. At the receiving end, the protective feature vector is used to perform error repair on the simulated protective video frames. The specific operation includes first decoding the received simulated protective video frames and extracting the embedded protective feature vectors. Then, an error correction decoding algorithm, such as LDPC or Turbo decoding, is applied to detect and repair the bit errors and lost data generated during the transmission process using the redundant information in the protective feature vector. Through the error repair process, the generated receiving end simulated comparison video can be restored to a state close to the original state, ensuring the quality and integrity of the video.

[0073] Preferably, step S24 comprises the following steps:

[0074] Step S241: performing error correction coding enhancement on the video feature vector to generate a protective feature vector;

[0075] Step S242: performing feature protection optimization on the enhanced video frame according to the protective feature vector to obtain a feature protective video frame;

[0076] Step S243: performing transmission simulation on the characteristic protective video frame based on the simulated transmission channel to generate a simulated protective video frame;

[0077] Step S244: Use the protective feature vector to perform receiving end error repair on the simulated protective video frame to obtain a receiving end simulated comparison video.

[0078] The benefit of the present invention lies in that the anti-interference ability of the feature vector is enhanced through error correction coding technology, ensuring that the feature data is protected from noise and interference during the transmission process. The enhanced feature vector is used to perform targeted protective processing on the video frame to enhance the robustness of key features in the video frame, ensuring that it can better resist losses during the transmission process. The transmission characteristics of the feature protective video frame are tested in a simulated transmission environment to provide reliable test data support for the error repair processing at the receiving end. The protective feature vector is used to effectively repair the errors of the simulated video frame at the receiving end, and the video information that suffers interference loss during the transmission process is restored to the greatest extent. The anti-interference ability of the video feature data is further optimized and enhanced to ensure that it can still maintain a high degree of integrity and accuracy in the actual network transmission environment. This is crucial to improving the stability and reliability of the final video synchronization.

[0079] In an embodiment of the present invention, error correction coding technology such as LDPC code or Turbo code is used to enhance the error correction coding of the video feature vector. The specific operation includes first selecting an appropriate error correction coding scheme and setting coding parameters according to the length of the video feature vector and the transmission environment. Then, the error correction coding algorithm is applied to convert the video feature vector into a protective feature vector, and redundant information is added to improve the anti-interference ability. Through error correction coding, the generated protective feature vector can effectively detect and correct errors during transmission to ensure the integrity of the data. The enhanced video frame is optimized for feature protection using an optimization algorithm based on the protective feature vector. The specific operation includes associating the protective feature vector with the enhanced video frame, and embedding the redundant information of the feature vector into the video frame through weighted fusion or embedding technology. This process embeds redundant feature information in the video frame, so that the original features of the video frame can be restored through the redundant information even if it is interfered during transmission, and a feature protection video frame is generated. A network simulation tool such as ns-3 or OMNeT++ is used to simulate the transmission of the feature protection video frame. The specific steps include setting the transmission path and parameters such as bandwidth, delay, and packet loss rate in the simulation environment to simulate real network transmission conditions. The characteristic protective video frames are transmitted through the simulated transmission channel, and the packet loss, delay and bit error phenomena that occur during the transmission process are recorded. At the receiving end, the protective characteristic vector is used to perform error repair on the simulated protective video frames. The specific operation includes first decoding the received simulated protective video frames and extracting the embedded protective characteristic vectors. Then, an error correction decoding algorithm, such as LDPC or Turbo decoding, is applied to detect and repair the bit errors and lost data generated during the transmission process using the redundant information in the protective characteristic vector. Through the error repair process, the generated receiving end simulated comparison video can be restored to a state close to the original state, ensuring the quality and integrity of the video.

[0080] Preferably, step S3 comprises the following steps:

[0081] Step S31: performing feature descriptor comparison on the simulated comparison video at the receiving end and the comparison video source based on the video feature vector to generate matching feature point pairs;

[0082] Step S32: Screening the matching feature point pairs based on a preset matching frame pair threshold to obtain preliminary synchronization frame pairs;

[0083] Step S33: Calculate the timestamp difference of the preliminary synchronization frame pair to obtain a timestamp difference value;

[0084] Step S34: performing timestamp correction on the preliminary synchronization frame pair using the timestamp difference value to generate a corrected synchronization frame pair.

[0085] The benefit of the present invention lies in that by using the video feature vector to compare the receiving end analog video and the original comparison video at the feature descriptor level, the corresponding feature points between the two videos can be accurately found, providing a reliable basis for subsequent synchronization. According to the preset matching frame pair threshold, reliable synchronization frame pairs are screened out from a large number of feature point pairs, effectively removing inaccurate or unstable matching points and improving the accuracy of synchronization. The timestamp difference value between the preliminary synchronization frame pairs is calculated, providing the necessary basis data for subsequent timestamp correction. The timestamp difference value is used to perform accurate timestamp correction on the preliminary synchronization frame pairs to ensure that the two videos are completely aligned in the time dimension, thereby improving the final synchronization accuracy. Focusing on the key links of feature descriptor comparison, synchronization frame pair screening, timestamp difference calculation and correction, the accuracy and reliability of video synchronization are effectively improved. This is of great significance for realizing high-quality comparison video synchronization applications.

[0086] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes:

[0087] Step S31: performing feature descriptor comparison on the simulated comparison video at the receiving end and the comparison video source based on the video feature vector to generate matching feature point pairs;

[0088] In an embodiment of the present invention, an image matching algorithm such as SIFT (Scale-Invariant Feature Transform) or SURF (Speeded-Up Robust Features) is used to compare the feature descriptors of the simulated comparison video at the receiving end and the comparison video source. The specific operation includes first extracting feature points and feature descriptors in the receiving end and the comparison video source, and then applying a feature matching algorithm such as FLANN (Fast Library for Approximate Nearest Neighbors) for matching. The matching process calculates the Euclidean distance between feature descriptors, selects the point pair with the smallest distance as the matching feature point pair, and generates a set of matching feature point pairs.

[0089] Step S32: Screening the matching feature point pairs based on a preset matching frame pair threshold to obtain preliminary synchronization frame pairs;

[0090] In an embodiment of the present invention, a feature matching algorithm is used to calculate the matching value of each pair of matching feature points. The specific operation includes applying a distance metric between feature descriptors, such as Euclidean distance or Hamming distance, to calculate the matching value of the matching feature point pair. The matching value reflects the similarity between the feature point pairs, and the smaller the value, the better the match. For each pair of feature points, its matching value is calculated, and the matching values ​​of all feature point pairs are recorded. According to a preset matching frame pair threshold, the calculated feature point pair matching values ​​are compared. When the matching value is less than the preset threshold, these feature point pairs are defined as unsynchronized frame pairs. The specific operation includes setting a threshold, such as the maximum allowable value of the matching value. When the matching value is lower than the threshold, it is considered that the matching effect is not good. These unsynchronized frame pairs are removed from the matching feature point pair set. The removal process deletes the matching point pairs by using an image processing tool, such as the filtering function in MATLAB. The feature point pairs whose matching values ​​are greater than the preset threshold are processed. The specific operation includes setting a threshold, such as the minimum required value of the matching value. When the matching value is higher than the threshold, it is considered that these feature point pairs are well matched. These feature point pairs are defined as synchronized frame pairs, and the pairs are integrated. The integration operation can ensure effective synchronization in time and space by performing time alignment and data fusion on each pair of synchronized frames. The generated preliminary synchronized frame pair set contains frame pairs with high matching quality, laying the foundation for further synchronization processing.

[0091] Step S33: Calculate the timestamp difference of the preliminary synchronization frame pair to obtain a timestamp difference value;

[0092] In an embodiment of the present invention, the timestamp of the preliminary synchronization frame pair is extracted and calculated. The specific operation includes reading the timestamp information of each frame, and calculating the timestamp difference between the analog comparison video frame and the comparison video source frame at the receiving end. By comparing the timestamps of the frame pairs, the timestamp difference value of each pair of preliminary synchronization frames is calculated. The calculation formula of the timestamp difference value is: timestamp difference value = receiving end frame timestamp - comparison source frame timestamp. The purpose of this step is to provide accurate time difference information for subsequent time correction.

[0093] Step S34: performing timestamp correction on the preliminary synchronization frame pair using the timestamp difference value to generate a corrected synchronization frame pair.

[0094] In an embodiment of the present invention, the calculated timestamp difference value is used to perform timestamp correction on the preliminary synchronization frame pair. The specific operation includes adjusting the timestamp of each pair of preliminary synchronization frames so that the timestamp of the receiving end frame is synchronized with the timestamp of the comparison source frame. The correction method can use weighted average or least square method to smooth all timestamp difference values ​​to obtain the corrected timestamp difference value. Then, the corrected timestamp is applied to the simulated comparison video frame of the receiving end to generate a corrected synchronization frame pair. Through timestamp correction, it is ensured that the final synchronization frame pair is completely consistent in time, providing an accurate time reference for subsequent video synchronization processing.

[0095] Preferably, performing synchronous frame pair screening on matching feature point pairs based on a preset matching frame pair threshold comprises:

[0096] Calculate the matching value of the matching feature point pair to obtain the matching value of the feature point pair;

[0097] When the feature point pair matching value is less than the preset matching frame pair threshold, the matching feature point pair is defined as an unsynchronized frame pair; and the unsynchronized frame pair is eliminated;

[0098] When the feature point pair matching value is greater than a preset matching frame pair threshold, the matching feature point pair is defined as a synchronization frame pair, and the synchronization frame pair is integrated to obtain a preliminary synchronization frame pair.

[0099] The benefit of the present invention lies in that detailed matching value calculations are performed on the compared feature point pairs, which can quantify the degree of matching between the feature point pairs and provide a basis for the subsequent screening of synchronous frame pairs. When the feature point pair matching value is lower than the preset threshold, it is defined as an asynchronous frame pair and eliminated, which effectively removes unreliable feature point matching and improves the accuracy of the synchronization result. When the feature point pair matching value is higher than the preset threshold, it is defined as a synchronous frame pair, and these synchronous frame pairs are integrated to obtain a preliminary set of synchronous frame pairs, which provides key intermediate results for subsequent timestamp difference calculation and correction. The matching values ​​of feature point pairs are fully utilized to effectively screen out synchronous frame pairs, laying a solid foundation for the key steps of video synchronization. This method based on feature point pair matching can more accurately locate synchronous and asynchronous frame pairs, thereby improving the quality and reliability of the final synchronization result.

[0100] In an embodiment of the present invention, a feature matching algorithm is used to calculate the matching value of each pair of matching feature points. The specific operation includes applying a distance metric between feature descriptors, such as Euclidean distance or Hamming distance, to calculate the matching value of the matching feature point pair. The matching value reflects the similarity between the feature point pairs, and the smaller the value, the better the match. For each pair of feature points, its matching value is calculated, and the matching values ​​of all feature point pairs are recorded. According to a preset matching frame pair threshold, the calculated feature point pair matching values ​​are compared. When the matching value is less than the preset threshold, these feature point pairs are defined as unsynchronized frame pairs. The specific operation includes setting a threshold, such as the maximum allowable value of the matching value. When the matching value is lower than the threshold, it is considered that the matching effect is not good. These unsynchronized frame pairs are removed from the matching feature point pair set. The removal process deletes the matching point pairs by using an image processing tool, such as the filtering function in MATLAB. The feature point pairs whose matching values ​​are greater than the preset threshold are processed. The specific operation includes setting a threshold, such as the minimum required value of the matching value. When the matching value is higher than the threshold, it is considered that these feature point pairs are well matched. These feature point pairs are defined as synchronized frame pairs, and the pairs are integrated. The integration operation can ensure effective synchronization in time and space by performing time alignment and data fusion on each pair of synchronized frames. The generated preliminary synchronized frame pair set contains frame pairs with high matching quality, laying the foundation for further synchronization processing.

[0101] Preferably, step S4 comprises the following steps:

[0102] Step S41: performing video synchronization simulation on the enhanced video frame and the receiving end analog comparison video by correcting the synchronization frame pair to obtain a synchronized enhanced video frame and a synchronized analog comparison video;

[0103] Step S42: positioning feature coordinates of the synchronous enhanced video frame based on the video feature vector to generate feature coordinates of the enhanced video frame;

[0104] Step S43: positioning feature coordinates of the synchronous simulation comparison video based on the video feature vector to generate feature coordinates of the simulation comparison video;

[0105] Step S44: performing feature space position comparison on the enhanced video frame feature coordinates and the simulated comparison video feature coordinates to obtain feature coordinate space matching results.

[0106] The benefit of the present invention lies in that the enhanced video frame and the analog comparison video at the receiving end are synchronously simulated by using the correction synchronization frame pair, ensuring that the two video sources are completely aligned in the time dimension, providing a reliable basis for the subsequent feature coordinate comparison. The feature coordinates of the synchronous enhanced video frame and the synchronous analog comparison video are located respectively, and the feature point coordinate information of the two video sources is extracted for the subsequent spatial position comparison. The feature coordinates of the enhanced video frame and the analog comparison video are compared in detail in space to obtain the matching results of the feature points of the two video sources in space, providing a basis for coordinate alignment for the final video fusion. Through video synchronization simulation, feature coordinate positioning and spatial position comparison, the high consistency of the enhanced video frame and the analog comparison video in both time and space dimensions is ensured. This provides a reliable basic condition for subsequent video fusion and helps to achieve more accurate comparison video enhancement applications.

[0107] In an embodiment of the present invention, a video synchronization simulation is performed using a correction synchronization frame pair to achieve synchronization between an enhanced video frame and a receiving end simulation comparison video. The specific operation includes first aligning the timestamps of the correction synchronization frame pair to ensure that the two sets of video data are completely consistent in time. Then, a video processing tool, such as a video processing function in MATLAB or OpenCV, is applied to synchronize the enhanced video frame and the receiving end simulation comparison video. The synchronization process involves aligning the two sets of video frames on the time axis, and generating synchronized enhanced video frames and synchronized simulation comparison videos through interpolation or resampling technology. This step ensures that the two sets of videos are synchronized in time, providing a consistent data basis for subsequent feature coordinate positioning and comparison. When locating feature coordinates of the synchronized enhanced video frame, a feature extraction and matching algorithm, such as SIFT or SURF, is used to calculate the coordinates of the feature points in combination with feature descriptors. The specific operation includes first extracting feature points in the synchronized enhanced video frame, and then using feature descriptor matching technology to determine the coordinates of each feature point in the image. The generated feature coordinates include the (x, y) coordinate position of each feature point, which is used to describe the feature distribution in the enhanced video frame. First, feature extraction algorithms such as ORB (Oriented FAST and Rotated BRIEF) are used to extract feature points in the synchronized simulation comparison video. Then, the coordinates of each feature point are determined by calculating feature descriptors. The generated feature coordinates reflect the positions of each feature point in the simulation comparison video, providing data support for feature space position comparison. The feature coordinates of the enhanced video frame and the feature coordinates of the simulation comparison video are spatially compared and processed using matching algorithms such as RANSAC (Random Sample Consensus). The specific operation includes first representing the two sets of feature coordinates as coordinate point sets, and then using the RANSAC algorithm to match the coordinate points. The feature coordinate matching effect between the two videos is evaluated by calculating the transformation matrix between the matching point pairs, such as the homography matrix or the affine transformation matrix. The generated feature coordinate space matching results show the corresponding relationship between the enhanced video frame and the simulation comparison video in the feature space. The purpose of this step is to evaluate the accuracy of video synchronization through the matching results and provide data support for subsequent analysis and processing.

[0108] Preferably, step S5 comprises the following steps:

[0109] Step S51: when the enhanced video frame feature coordinates are equal to the simulated comparison video feature coordinates, the feature coordinate space matching result is defined as the feature space synchronization result, and the enhanced video frame feature coordinates are clustered based on the feature space synchronization result to obtain feature coordinate clustering data;

[0110] Step S52: performing cluster center positioning on the feature coordinate clustering data to generate the feature coordinate cluster center;

[0111] Step S53: performing inter-frame transformation estimation on the feature coordinate cluster center to generate an inter-frame spatial transformation matrix;

[0112] Step S54: constructing an error correction synchronization template for the synchronous simulation comparison video based on the inter-frame spatial transformation matrix to obtain an error correction video synchronization template;

[0113] Step S55: Perform video synchronization communication on the video source by comparing the error-correcting video synchronization template, so as to execute the template error-correcting video synchronization method.

[0114] The benefit of the present invention lies in that when the feature coordinates of the enhanced video frame and the simulated comparison video are completely consistent, it is defined as a feature space synchronization result, which provides a reliable benchmark basis for subsequent feature coordinate clustering. Based on the feature space synchronization result, the feature coordinates of the enhanced video frame are clustered, and the key feature points in the video can be effectively extracted, laying the foundation for the subsequent transformation matrix estimation. The spatial position information of the important feature points in the video frame can be obtained by locating the cluster center of the feature coordinate clustering data. Based on the feature coordinate clustering center, the spatial transformation matrix between frames is estimated, which provides key geometric transformation parameters for the construction of the video synchronization template. Using the inter-frame spatial transformation matrix, an error correction synchronization template for the comparison video is constructed, which can effectively correct the spatiotemporal distortion of the comparison video and improve the accuracy of video synchronization. Make full use of the spatial position information of the feature coordinates, and construct a reliable video synchronization template through clustering and transformation estimation methods. This provides an important basis for the subsequent template error correction type video synchronization, which can greatly improve the accuracy and robustness of video synchronization.

[0115] In an embodiment of the present invention, the feature coordinates of the enhanced video frame are compared with the feature coordinates of the simulated comparison video. If the coordinate points match successfully, the feature space synchronization result is considered to be valid. The specific operation includes first comparing all the matched feature coordinate points. When the coordinate values ​​are equal, it is defined as the feature space synchronization result. Then, based on these synchronization results, a clustering algorithm such as the K-means algorithm or DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is used to cluster the feature coordinates in the enhanced video frame. The clustering process includes setting clustering parameters, such as the number of clusters or density thresholds, grouping the feature coordinates into different clusters, and generating feature coordinate clustering data. When locating the cluster center of the feature coordinate clustering data, a clustering algorithm is used to calculate the center point of each cluster. The specific operation includes using the results of the clustering algorithm to calculate the mean coordinates of each cluster as the cluster center position. The mean is calculated for all feature coordinates in each cluster to obtain the cluster center coordinates. The calculation formula for the cluster center is: Cluster center coordinates = ,in, is the X-axis coordinate of the coordinate point in the cluster, is the Y-axis coordinate of the coordinate point in the cluster, is the number of feature points, is a variable. The generated feature coordinate cluster center is used to describe the position of each cluster in the feature space and provide data support for inter-frame transformation estimation. When inter-frame transformation estimation is performed on the feature coordinate cluster center, the spatial transformation matrix from one frame to another is calculated. The specific operation includes estimating the spatial transformation matrix of the feature coordinate cluster center by using the least squares method or the RANSAC algorithm through the coordinate data of the feature point matching. The transformation matrix can be an affine transformation matrix or a perspective transformation matrix, depending on the geometric transformation of the feature point. During the calculation process, the error and transformation type between the point pairs need to be considered. The generated inter-frame spatial transformation matrix is ​​used to describe the geometric transformation relationship between the two video frames and provide a basis for the subsequent synchronization template construction. The inter-frame spatial transformation matrix is ​​used to construct an error correction synchronization template for the synchronization simulation comparison video. The specific operation includes applying the generated transformation matrix to the feature coordinates of the synchronization simulation comparison video and correcting the geometric distortion in the video frame. Using video processing tools such as OpenCV, image geometric transformation operations are performed and the video frame is error corrected according to the transformation matrix. Through transformation and correction, the generated error correction video synchronization template can repair the geometric errors in the video and ensure the synchronization accuracy of the video in subsequent processing. Use the error-correcting video synchronization template to compare the video source for synchronization communication to ensure accurate synchronization of video data. The specific operation includes applying the error-correcting video synchronization template to the comparison video source and performing video synchronization through image registration technology. Use video processing software or programming tools, such as the image registration function in MATLAB, to adjust and correct the video source according to the error-correcting synchronization template comparison. Compare the synchronized comparison video source with other synchronized video data to verify the synchronization effect. Execute the template error-correcting video synchronization method to ensure the consistency of the video source at different time points and spatial positions.

[0116] Preferably, step S6 comprises the following steps:

[0117] Step S61: when the enhanced video frame feature coordinates are not equal to the simulated comparison video feature coordinates, the feature coordinate space matching result is defined as a feature space asynchronous result, and a coordinate trajectory of the enhanced video frame feature coordinates is constructed based on the feature space asynchronous result to obtain a video frame feature trajectory;

[0118] Step S62: performing drift point analysis on the video frame feature trajectory to generate feature trajectory drift points;

[0119] Step S63: performing timestamp positioning on the characteristic trajectory drift point to generate a characteristic drift timestamp;

[0120] Step S64: synchronously correcting the simulated comparison video feature coordinates based on the feature drift timestamp to obtain the corrected comparison video feature coordinates;

[0121] Step S65: constructing a correction synchronization template for the synchronous simulation comparison video based on the correction comparison video feature coordinates to obtain a correction video synchronization template;

[0122] Step S66: Use the corrected video synchronization template to compare the video source for video synchronization communication, so as to execute the corrected video synchronization method.

[0123] The benefit of the present invention lies in that when the feature coordinates of the enhanced video frame are not equal to the feature coordinates of the simulated comparison video, it is defined as a feature space asynchronization result, indicating that the two video sources have deviations in time and space and need further correction. Based on the feature space asynchronization result, the feature coordinates of the enhanced video frame are constructed by trajectory, which can reflect the motion trajectory of the feature points in the video frame and provide a data basis for the subsequent drift point analysis. The constructed feature trajectory is subjected to drift point analysis to identify the abnormal position changes of the feature points between the video frames, providing valuable clues for timestamp positioning. Based on the feature trajectory drift points, the timestamp information of the abnormal changes of the feature points in the video frame is located, which lays the foundation for the subsequent correction of the feature coordinates of the comparison video. Using the feature drift timestamp, the feature coordinates of the simulated comparison video are corrected, and the deviation of the feature points caused by the time and space distortion in the comparison video can be corrected. Based on the corrected comparison video feature coordinates, a correction synchronization template for the comparison video is constructed, and the application of the correction template can effectively correct the time and space distortion of the comparison video and improve the accuracy of video synchronization. In the case of feature space asynchrony, a reliable corrective video synchronization template is constructed through trajectory analysis, timestamp positioning and coordinate correction methods. This provides an important guarantee for subsequent corrective video synchronization and greatly improves the accuracy and robustness of video synchronization.

[0124] In the embodiment of the present invention, by comparing the enhanced video frame feature coordinates and the simulated comparison video feature coordinates, if the coordinate points are found to be mismatched, it is defined as a feature space asynchronous result. The specific operation includes first extracting all matching feature point coordinates and recording these mismatched coordinate points. Then, the coordinate trajectory is constructed using these asynchronous feature point coordinates. The construction of the coordinate trajectory uses a trajectory analysis algorithm, such as a Kalman filter or a particle filter, to model the motion trajectory of the feature points. When performing drift point analysis on the video frame feature trajectory, the goal is to identify and locate abnormal points in the trajectory. The specific operation includes using a trajectory analysis algorithm, such as a sliding window method or an outlier detection algorithm, to analyze data points in the video frame feature trajectory. These algorithms identify drift points that deviate greatly from the normal trajectory by calculating the deviation or change trend of the feature points. The generated feature trajectory drift points reflect the abnormal phenomena caused by synchronization problems in the video frame. When performing timestamp positioning on the feature trajectory drift points, the timestamp information of each drift point is first recorded. The specific operation includes using the timestamp recording mechanism of the video frame to determine the specific time when each drift point occurs. During the processing, time series analysis techniques, such as timestamp alignment or interpolation methods, are used to accurately calculate the timestamps of drift points. These timestamps reflect the specific moment when the feature points drift in the video frame, and the generated feature drift timestamps are used for subsequent synchronization correction operations. Based on the feature drift timestamps, when the feature coordinates of the simulated comparison video are synchronized and corrected, the timestamp information is used to adjust the feature coordinates in the simulated comparison video. The specific operation includes using the feature drift timestamp to adjust the feature coordinate position in the video frame and correct the deviation caused by time asynchronization. The correction process can correct the feature coordinates by using the time interpolation method or the geometric transformation technology based on the timestamp. The corrected feature coordinates are called the corrected comparison video feature coordinates, which ensure that the feature points in the video are consistent with the actual video data and provide accurate basic data for the construction of the correction synchronization template. When the correction synchronization template is constructed using the corrected comparison video feature coordinates, the synchronization template is first generated based on the corrected feature coordinates. The specific operation includes applying image processing technology to map the corrected feature coordinates to the video frame and perform geometric correction. A correction video synchronization template can be constructed using a template matching algorithm or image registration technology, such as feature point-based registration or global transformation. When executing the correction video synchronization method, the correction video synchronization template is applied to the comparison video source for synchronous communication. The specific operation includes registering the correction template with the comparison video source, and applying the transformation parameters in the template to adjust the comparison video source through image processing tools such as OpenCV or MATLAB. During the synchronous communication process, the correction data in the template is used to adjust the feature coordinates in the comparison video source to correct the errors caused by time or space asynchrony.

[0125] The invention is beneficial in that the original comparison video source is analyzed in the frequency domain by using discrete wavelet transform, and the multi-frequency domain component data is image enhanced, thereby improving the quality of the video frame and the information reliability. The enhanced video frame is multi-dimensionally encoded to generate a video feature vector, and the protective transmission is simulated to ensure the integrity of the feature information. Compared with directly transmitting the video stream, this feature encoding and transmission method can effectively reduce the transmission bandwidth and improve the transmission stability. The preliminary screening is performed by presetting the matching frame pair threshold to obtain the preliminary synchronization frame pair, and further by timestamp correction, the correction synchronization frame pair is generated, thereby improving the accuracy of the synchronization frame pair. Based on the correction synchronization frame pair, the enhanced video frame and the receiving end simulated comparison video are compared and synchronized in the feature space position, thereby realizing high-precision video synchronization. When the feature space is not synchronized, the template error correction type or correction type video synchronization method can be used to automatically correct the synchronization deviation, thereby improving the robustness of the video synchronization. The image preprocessing, feature coding transmission, synchronization screening, and space matching links are organically integrated to realize the full automation of video synchronization, thereby greatly improving the work efficiency. Therefore, this video synchronization method based on image multi-dimensional feature value comparison makes full use of image enhancement, feature coding and synchronization correction technology to achieve high-precision video synchronization while ensuring video quality and transmission security.

[0126] 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.

[0127] 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 video synchronization method based on image multi-dimensional feature value comparison, characterized in that: The following steps are involved: Step S1: Obtain comparison video source; Performing discrete wavelet transform on the video source to generate multi-frequency domain component data; Perform image enhancement processing on the multi-frequency domain component data to generate an enhanced video frame, wherein step S1 includes the following steps: Step S11: collecting data streams in the comparison system to obtain a comparison video source; Step S12: performing inter-frame image cutting on the comparison video source to obtain a comparison video frame; Step S13: performing discrete wavelet transform on the video frames to generate multi-frequency domain component data, wherein the multi-frequency domain component data includes low-frequency domain components and high-frequency domain components; Step S14: performing histogram grayscale conversion on the low-frequency domain component to obtain a low-frequency domain enhanced video frame; Step S15: performing adaptive directional filtering on the high-frequency domain components to generate a high-frequency domain enhanced video frame, wherein step S15 includes the following steps: Step S151: performing pyramid decomposition layer by layer on the high-frequency domain components to obtain multi-scale high-frequency domain components; Step S152: Calculate the directional gradient of the multi-scale high-frequency domain components to generate component directional detail information; Step S153: performing multi-directional filtering processing on the high-frequency domain component based on the component directional detail information to obtain multi-directional filtering response information; Step S154: weight mapping is performed on the multi-directional filtering response information to obtain weight data for each direction; Step S155: performing weight distribution directional filtering processing on the high-frequency domain components based on the weight data of each direction to generate a high-frequency domain enhanced video frame; Step S16: performing weighted fusion on the low-frequency domain enhanced video frame and the high-frequency domain enhanced video frame to generate an enhanced video frame; Step S2: Perform multi-dimensional feature encoding on the enhanced video frame to generate a video feature vector; perform protective transmission simulation on the video feature vector to obtain a receiving end simulation comparison video; Step S3: Screening the synchronization frame pairs of the analog comparison video at the receiving end based on a preset matching frame pair threshold to obtain preliminary synchronization frame pairs; performing timestamp correction on the preliminary synchronization frame pairs to generate corrected synchronization frame pairs; Step S4: performing time domain synchronous coordinate positioning of the enhanced video frame and the analog comparison video at the receiving end by correcting the synchronous frame pair, generating enhanced video frame feature coordinates and analog comparison video feature coordinates; performing feature space position comparison of the enhanced video frame feature coordinates and the analog comparison video feature coordinates, and obtaining feature coordinate space matching results; Step S5: when the enhanced video frame feature coordinates are equal to the simulated comparison video feature coordinates, the feature coordinate space matching result is defined as the feature space synchronization result, and based on the feature space synchronization result and the enhanced video frame feature coordinate comparison, the video source is subjected to video template synchronization communication to execute the template error correction type video synchronization method; Step S6: When the enhanced video frame feature coordinates are not equal to the simulated comparison video feature coordinates, the feature coordinate space matching result is defined as a feature space asynchronous result, and based on the feature space asynchronous result and the enhanced video frame feature coordinate comparison, the video source is subjected to video correction synchronization communication to execute a correction type video synchronization method.

2. The video synchronization method based on image multi-dimensional feature value comparison according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: extracting multi-dimensional features from the enhanced video frame to obtain a video feature data set; Step S22: performing vector encoding on the video feature data set to generate a video feature vector; Step S23: Building a simulation transmission channel for the enhanced video frame to generate a simulation transmission channel; Step S24: Based on the simulated transmission channel and the video feature vector, feature protection transmission simulation is performed on the enhanced video frame to obtain a receiving end simulated comparison video.

3. The video synchronization method based on image multi-dimensional feature value comparison according to claim 2, characterized in that: Step S24 includes the following steps: Step S241: performing error correction coding enhancement on the video feature vector to generate a protective feature vector; Step S242: performing feature protection optimization on the enhanced video frame according to the protective feature vector to obtain a feature protective video frame; Step S243: performing transmission simulation on the characteristic protective video frame based on the simulated transmission channel to generate a simulated protective video frame; Step S244: Use the protective feature vector to perform receiving end error repair on the simulated protective video frame to obtain a receiving end simulated comparison video.

4. The video synchronization method based on image multi-dimensional feature value comparison according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: performing feature descriptor comparison on the simulated comparison video at the receiving end and the comparison video source based on the video feature vector to generate matching feature point pairs; Step S32: Screening the matching feature point pairs based on a preset matching frame pair threshold to obtain preliminary synchronization frame pairs; Step S33: Calculate the timestamp difference of the preliminary synchronization frame pair to obtain a timestamp difference value; Step S34: performing timestamp correction on the preliminary synchronization frame pair using the timestamp difference value to generate a corrected synchronization frame pair.

5. The video synchronization method based on image multi-dimensional feature value comparison according to claim 4 is characterized in that: The synchronous frame pair screening of the matching feature point pairs based on the preset matching frame pair threshold includes: Calculate the matching value of the matching feature point pair to obtain the matching value of the feature point pair; When the feature point pair matching value is less than the preset matching frame pair threshold, the matching feature point pair is defined as an unsynchronized frame pair; and the unsynchronized frame pair is eliminated; When the feature point pair matching value is greater than a preset matching frame pair threshold, the matching feature point pair is defined as a synchronization frame pair, and the synchronization frame pair is integrated to obtain a preliminary synchronization frame pair.

6. The video synchronization method based on image multi-dimensional feature value comparison according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: performing video synchronization simulation on the enhanced video frame and the receiving end analog comparison video by correcting the synchronization frame pair to obtain a synchronized enhanced video frame and a synchronized analog comparison video; Step S42: positioning feature coordinates of the synchronous enhanced video frame based on the video feature vector to generate feature coordinates of the enhanced video frame; Step S43: positioning feature coordinates of the synchronous simulation comparison video based on the video feature vector to generate feature coordinates of the simulation comparison video; Step S44: performing feature space position comparison on the enhanced video frame feature coordinates and the simulated comparison video feature coordinates to obtain feature coordinate space matching results.

7. The video synchronization method based on image multi-dimensional feature value comparison according to claim 1, characterized in that: Step S5 includes the following steps: Step S51: when the enhanced video frame feature coordinates are equal to the simulated comparison video feature coordinates, the feature coordinate space matching result is defined as the feature space synchronization result, and the enhanced video frame feature coordinates are clustered based on the feature space synchronization result to obtain feature coordinate clustering data; Step S52: performing cluster center positioning on the feature coordinate clustering data to generate the feature coordinate cluster center; Step S53: performing inter-frame transformation estimation on the feature coordinate cluster center to generate an inter-frame spatial transformation matrix; Step S54: constructing an error correction synchronization template for the synchronous simulation comparison video based on the inter-frame spatial transformation matrix to obtain an error correction video synchronization template; Step S55: Perform video synchronization communication on the video source by comparing the error-correcting video synchronization template, so as to execute the template error-correcting video synchronization method.

8. The video synchronization method based on image multi-dimensional feature value comparison according to claim 1, characterized in that: Step S6 includes the following steps: Step S61: when the enhanced video frame feature coordinates are not equal to the simulated comparison video feature coordinates, the feature coordinate space matching result is defined as a feature space asynchronous result, and a coordinate trajectory of the enhanced video frame feature coordinates is constructed based on the feature space asynchronous result to obtain a video frame feature trajectory; Step S62: performing drift point analysis on the video frame feature trajectory to generate feature trajectory drift points; Step S63: performing timestamp positioning on the characteristic trajectory drift point to generate a characteristic drift timestamp; Step S64: synchronously correcting the simulated comparison video feature coordinates based on the feature drift timestamp to obtain the corrected comparison video feature coordinates; Step S65: constructing a correction synchronization template for the synchronous simulation comparison video based on the correction comparison video feature coordinates to obtain a correction video synchronization template; Step S66: Use the corrected video synchronization template to compare the video source for video synchronization communication, so as to execute the corrected video synchronization method.

Images