Universal video image stabilization method based on multi-strategy fusion

Abstract

The invention discloses a universal video image stabilization method based on multi-strategy fusion. The problem of video jitter in a complex dynamic scene is solved. Various algorithms are combined, so that the adaptability and robustness of the system are improved, and the method is widely applied to different types of camera equipment and video scenes. In a motion estimation link, an nR-AKAZE feature matching algorithm is adopted, feature points in video frames can be rapidly detected, and robust descriptors are generated. And then, Fourier transform is carried out through a phase correlation method, so that fine translation amount estimation is realized, and high-precision motion detection is ensured. An improved method based on parameter filtering is introduced in the motion compensation stage, and a Gaussian filter is used for smoothing a global motion vector, so that the influence of unintentional motion is reduced, and error accumulation is avoided. In addition, when the image-stabilized video frame is generated, the continuity and smooth transition of the visual effect are ensured through the image correction and zero value filling technology. And an efficient and accurate real-time video image stabilization processing solution is integrally provided.

Description

Technical Field

[0001] The present invention belongs to the technical field of video stabilization, and in particular relates to a universal video stabilization method based on multiple strategy fusion. Background Art

[0002] The continuous advancement of science and technology and the improvement of living standards have led to the widespread application of electronic video equipment in various fields. Video sequences are not only a tool for information transmission, but also contain rich in-depth content. They are widely used in many fields such as civil, industrial and military. However, when shooting in complex environments, camera shaking can cause unstable movement or deformation in the video, which not only increases visual fatigue, but also makes subsequent video analysis and interpretation more difficult. In order to solve these problems, video stabilization technology came into being. It can effectively remove unnecessary random jitter and lay a good foundation for further video processing. Video stabilization not only improves the viewing experience, but also provides a guarantee for the accuracy of various application scenarios.

[0003] There are three main types of video stabilization technology: mechanical stabilization, optical stabilization, and electronic stabilization. Mechanical stabilization technology uses physical movement to suppress jitter, but because it relies on a complex mechanical structure, it often increases the weight of the system, is difficult to miniaturize, and is easily damaged in certain extreme environments. Optical stabilization technology achieves stabilization through lens movement, but this method also has limitations, mainly because it has high requirements for the lens, is expensive, and does not work well in low light conditions.

[0004] In comparison, electronic image stabilization technology directly achieves video stabilization through image processing algorithms, showing significant advantages such as simple operation, small size and high flexibility, which makes it the main trend of current image stabilization research and development. However, the current electronic image stabilization technology needs to balance accuracy and efficiency in practical applications. High-precision stabilization can improve video quality, but it often takes a long time to process, which may affect real-time performance; while high-efficiency algorithms can process videos quickly, but may sacrifice stabilization effects. Therefore, it is necessary to find the best balance between the two to ensure that the technology can provide a high-quality video image visual experience and can quickly adapt to various devices and environments for easy deployment and application. In this way, the practical application of electronic image stabilization technology will be more extensive and flexible. Summary of the invention

[0005] In order to solve the technical problems existing in the background technology, the present invention provides a universal video stabilization method based on the fusion of multiple strategies, which integrates the advantages of multiple intelligent image processing algorithms, including motion estimation based on feature matching and phase correlation, motion compensation based on parameter filtering and trajectory smoothing, and visual optimization based on image transformation and zero value filling. This technology aims to comprehensively consider accuracy, generalization and real-time performance to effectively reduce the visual discomfort caused by jitter, improve the user's viewing experience and the accuracy of subsequent analysis, while ensuring that the processing speed meets real-time requirements and can flexibly adapt to different devices and application scenarios.

[0006] In order to solve the technical problem, the technical solution of the present invention is:

[0007] A general video stabilization method based on multiple strategy fusion, the method comprising:

[0008] Motion estimation: Accurately estimate the translation of video frames through feature matching and phase correlation methods;

[0009] Motion compensation: Based on the estimated translation amount, the motion vector is smoothed by applying Gaussian filtering to obtain the smoothed transformation parameters and the processed current frame information;

[0010] Visual optimization: Based on the obtained transformation parameters and current frame information, the stabilized video sequence is obtained by using image transformation and zero-filling methods.

[0011] Furthermore, the motion estimation specifically includes:

[0012] Use feature matching to preliminarily estimate the translation:

[0013] Use the nR-AKAZE algorithm to detect feature points on video frames and quickly generate descriptors; use the BF algorithm to match descriptors through Hamming distance and find the best match by calculating the distance between feature points; use the RANSAC algorithm to remove incorrect matching points to improve matching accuracy and stability; calculate the displacement of the remaining valid matching points and average the displacement to obtain a preliminary translation component;

[0014] Phase correlation fine estimation of translation:

[0015] Preliminary motion estimation: The preliminary motion vector obtained by feature matching and other methods will be used to perform preliminary image transformation on the current frame to make it visually closer to the reference frame; Fourier transform: Fourier transform is performed on the current frame and the reference frame after preliminary transformation. This step converts the image from the spatial domain to the frequency domain, making the analysis of the translation transformation between images more efficient; Frequency domain processing: Multiply the Fourier transform result of the current frame with the conjugate complex number of the Fourier transform of the reference frame; This operation is to use the frequency domain characteristics to emphasize the phase information between images through complex operations and effectively filter out information related to translation; Inverse Fourier transform: Inverse Fourier transform is performed on the above product result to obtain a related image; In the related image, the value of each position indicates the degree of similarity between the current frame and the reference frame at the corresponding position; Locate the maximum value: In the related image, find the position of the maximum value; The position of the maximum value corresponds to the optimal translation between the current frame and the reference frame, that is, the fine motion vector; Motion vector superposition: The obtained fine motion vector is superimposed with the preliminary motion vector to obtain the final motion vector, which represents the comprehensive translation of the current frame relative to the reference frame.

[0016] Furthermore, the motion compensation specifically includes:

[0017] Parameter filtering and trajectory smoothing: Smooth the unintentional motion in the final motion vector; introduce a Gaussian filter to perform real-time smoothing based on the global transformation parameters of the current frame and adjacent stable output frames and unstabilized input frames; use a Gaussian filter to perform neighborhood weighting and smoothing on the motion vectors of the current frame and consecutive frames to obtain the smoothed transformation parameters and the processed current frame information, thereby achieving stable video output.

[0018] Furthermore, the visual optimization specifically includes:

[0019] Image transformation and black edge filling: Use the smoothed transformation parameters to correct the image of the current frame to reduce the impact of unintentional motion and retain intentional motion; for the black edges caused by image transformation, a zero-value filling method is used to fill the black edges of the corrected current frame with the stable output content of the previous frame to reduce the sense of visual incoherence; in this process, the corresponding part of the previous frame is seamlessly spliced ​​with the edge of the current frame to ensure the integrity of the image and the naturalness of the transition; finally, all the stabilized continuous frames are written into the video file in sequence to ensure the smoothness of the output video and improve the visual quality.

[0020] Compared with the prior art, the advantages of the present invention are:

[0021] First of all, the present invention provides a universal video stabilization method based on multiple strategy fusion, which aims to solve the problem of video jitter in complex dynamic scenes. By integrating multiple algorithms and technical means, the method is not only highly adaptable and robust, but also can be widely used in different types of camera equipment and video scenes. Whether in the processing of real-time video streams or in later video interpretation, the present invention shows good results, ensures stable video output, and has wide application value.

[0022] Secondly, in the motion estimation stage, the nR-AKAZE feature matching algorithm is used. This algorithm can quickly detect feature points in video frames and generate robust descriptors to remain invariant under different conditions. After that, the phase correlation method is used to perform fine translation estimation through Fourier transform, thereby improving the accuracy and robustness of motion estimation and ensuring that small image displacements can be identified.

[0023] Finally, in the motion compensation stage, the present invention introduces an improved method based on parameter filtering. The estimated global motion vector is smoothed by a Gaussian filter. This method not only focuses on the motion of the current frame, but also combines the information of the previous and next frames to reduce the impact of unintentional motion and avoid error accumulation. When the stabilized video frame is generated, the smoothed transformation parameters are used for image correction, and the zero values ​​in the image are filled by using the corresponding part of the previous frame after transformation, ensuring the smooth transition and coherence of the visual effect.

[0024] Based on this series of designs, the present invention constructs an efficient and accurate video stabilization processing solution, which is particularly suitable for real-time applications in dynamic scenes. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 , an overall flow chart of a universal video stabilization method based on multiple strategy fusion according to the present invention;

[0026] Figure 2 , the main flow chart of a universal video stabilization method based on multiple strategy fusion of the present invention. DETAILED DESCRIPTION

[0027] The specific implementation mode of the present invention is described below in conjunction with embodiments:

[0028] It should be noted that the structures, proportions, sizes, etc. shown in this specification are only used to match the contents disclosed in the specification so that people familiar with this technology can understand and read them, and are not used to limit the conditions under which the present invention can be implemented. Any structural modification, change in proportional relationship or adjustment of size should still fall within the scope of the technical content disclosed in the present invention without affecting the effects and purposes that can be achieved by the present invention.

[0029] At the same time, the terms such as "upper", "lower", "left", "right", "middle" and "one" cited in this specification are only for the convenience of description and are not used to limit the scope of implementation of the present invention. Changes or adjustments to their relative relationships should be regarded as the scope of implementation of the present invention without substantially changing the technical content.

[0030] Embodiment 1:

[0031] The following is a detailed explanation of the input and output of the three steps (motion estimation, motion compensation, and visual optimization) of the video stabilization solution of the present invention.

[0032] 1. Motion Estimation

[0033] enter:

[0034] Video frame sequence: multiple consecutive image frames from which motion information is extracted.

[0035] Feature detection algorithm: Parameters used by the nR-AKAZE algorithm (such as threshold, number of key points, etc.).

[0036] Image processing tools: Mathematical tools for Fourier transform and phase correlation calculations.

[0037] Output:

[0038] Preliminary global motion vector: A rough translation calculated in the first stage that represents the motion of the current frame relative to the reference frame.

[0039] Accurate global motion vector: The motion vector is finely adjusted by the phase correlation method to provide more accurate translation information.

[0040] 2. Motion Compensation

[0041] enter:

[0042] Superposition of precise global motion vector and preliminary global motion vector: The output from the first step contains the information of intended and unintentional camera motion.

[0043] Information about the current and adjacent frames: the global motion vectors of the stabilized output frame and the unstabilized input frame for smoothing.

[0044] Gaussian filter parameters: mathematical parameters used for neighborhood weighting.

[0045] Output:

[0046] Smoothed transformation parameters: After Gaussian filtering, the global motion vector affected by unintentional motion is removed, providing stable motion information.

[0047] 3. Visual optimization

[0048] enter:

[0049] Smoothed transformation parameters: The smoothed result from the second step, used for image rectification.

[0050] Current frame and adjacent frames: the stabilized output frame and the unstabilized input frame images.

[0051] Output:

[0052] Stable output video frames: After transformation and black edge filling, the final video frames are output with more stable and coherent visual effects.

[0053] Final video file: Combine the processed frames into a video file in sequence, ensuring that the parameters are consistent with the original video.

[0054] Summary: The output of motion estimation comes from the motion features of the video frame, which is used to obtain preliminary and accurate global motion vectors. The motion compensation stage uses the superimposed motion vectors for smoothing, aiming to eliminate unintentional motion and stabilize the current frame. Visual optimization ultimately outputs a stable and coherent image sequence and generates a playable stable video file. This process can effectively improve the quality and stability of video stabilization through step-by-step processing, and ultimately obtain a visually smoother and more comfortable viewing experience.

[0055] Embodiment 2:

[0056] This embodiment 2 is a further explanation of embodiment 1, which mainly includes three steps: motion estimation, motion compensation, and visual optimization. The overall framework and specific process of the technical solution are as follows: Figure 1 , 2 shown.

[0057] 1. Motion Estimation Based on Feature Matching and Phase Correlation

[0058] In video image stabilization, we usually only need to focus on the translation of the image and ignore rotation, scaling and other distortions, because in most cases, the jitter between video frames is mainly manifested as translation transformation. First, an accelerated KAZE (non-Rotation Accelerated KAZE, nR-AKAZE) feature detection algorithm that ignores rotation operations is used to quickly extract key features in video frames and preliminarily estimate the translation amount. This process minimizes redundant calculation operations, has high efficiency, and is suitable for real-time processing. Next, a more precise translation estimation is performed using the phase correlation method. Phase correlation uses the characteristics of Fourier transform and can more accurately identify small translations between images by analyzing phase information in the frequency domain. This strategy of first using feature matching to obtain a rough estimate and then fine-tuning it through phase correlation fully combines the advantages of both, effectively improving the accuracy and robustness of image stabilization processing without affecting efficiency, thereby providing more stable and clear image output in dynamic scenes.

[0059] 1.1. Feature matching based on nR-AKAZE

[0060] The feature matching process based on nR-AKAZE mainly includes the steps of feature point detection, descriptor generation, feature matching, error elimination and displacement estimation.

[0061] First, the nR-AKAZE (non-Rotation AKAZE) algorithm is used to detect feature points in the input image and quickly generate descriptors for each feature point while ignoring the rotation operation. These descriptors can maintain high robustness under different image conditions by capturing local image information, thus providing a solid foundation for subsequent matching.

[0062] Next, the BF (Brute Force Matcher) algorithm is used to match the descriptors extracted from the previous and next frames of the video. BF finds the best matching pair by calculating the distance between descriptors. In order to improve the accuracy of the matching, the RANSAC (Random Sample Consensus) algorithm is then applied to eliminate false matches. This process can effectively identify and remove matching points that do not conform to the overall transformation relationship through random sampling and model fitting, thereby enhancing the reliability of the matching results.

[0063] Finally, for the remaining valid matching points, the displacement between them is calculated. The specific method is to extract the displacement of each pair of matching points in the previous and next frames of the video, and average all the displacements to obtain the preliminary translation component.

[0064] 1.2. Phase correlation method

[0065] The translation component obtained by feature matching is applied to the video frame for preliminary transformation, so that the states of the previous and next frames of the video are visually closer. Next, the phase correlation method is used to perform secondary translation calculation on the video frame after preliminary transformation.

[0066] The phase correlation method uses the properties of Fourier transform to determine the displacement component between two images by calculating the frequency domain product of the two images. Specifically, the two images are first Fourier transformed, the product of the Fourier domain of one image and the conjugate complex number of the Fourier domain of the other image is calculated, and then the result is inverse Fourier transformed to obtain a new image, in which the maximum value position corresponds to the translation between the two images. This method can more accurately identify small displacements through frequency domain analysis and is highly robust to noise.

[0067] The obtained secondary translation is superimposed on the preliminary translation to obtain a more accurate overall translation component, that is, the global motion vector of the current frame relative to the reference frame.

[0068] 2. Motion compensation based on parameter filtering and trajectory smoothing

[0069] The global motion vector obtained by motion estimation contains intentional motion components (such as normal scanning motion of the camera) and unintentional motion components (such as random jitter of the camera). In order to eliminate the unintentional motion components in the video, it is usually necessary to perform smoothing compensation on the estimated global motion vector. However, traditional methods often smooth the accumulated global motion parameters, which easily introduces cumulative errors and eventually causes drift in the video sequence.

[0070] In order to solve this problem, the present invention adopts an improved real-time motion smoothing method. This method not only smoothes the unstable input frame sequence, but also introduces the stabilized output frame information, thereby more effectively eliminating high-frequency jitter. Specifically, the global transformation parameters calculated in real time between the current frame and the specified adjacent stabilized output frame and the unstabilized input frame are smoothed by a Gaussian filter to form a dynamic motion compensation mechanism.

[0071] In this process, the Gaussian filter is used as a motion filter to perform neighborhood weighting and iteration on the motion vectors of the current frame and several consecutive frames before and after, and finally obtain the smoothed transformation parameters. This motion compensation method based on parameter filtering and trajectory smoothing avoids the error accumulation problem caused by simple cumulative estimation in traditional methods. By achieving a smoother and more stable video output effect without increasing the computational overhead, the quality of motion compensation is significantly improved.

[0072] 3. Visual optimization based on image transformation and zero-value padding

[0073] The current frame is transformed using the smoothed transformation parameters, and the image is corrected by repositioning each pixel of the image. In this way, the unintentional motion component can be reduced or eliminated while retaining the intentional motion component of the video frame, thus achieving the effect of video stabilization. However, image transformation will cause zero values ​​to appear at the edges of the image, that is, black edges will be generated, affecting the visual effect.

[0074] To solve this problem, the previous frame after transformation (stabilization output) can be used to fill the black edges of the current frame transformation, making the image more complete and natural. Finally, all consecutive stabilized frames are written to the video file in sequence, and the parameters of the output video are ensured to be consistent with the original video, thus generating a playable stable output video.

[0075] Embodiment 3:

[0076] The present invention performs effective image stabilization processing on RGB and infrared video sequences in multiple offline and real-time scenarios. The motion estimation method that combines nR-AKAZE feature matching and phase correlation has significant improvements in stability and real-time performance compared to traditional feature matching algorithms such as SIFT (Scale-Invariant Feature Transform). Specifically, the present invention achieves a higher frame rate (Frames Per Second, FPS) during the processing process. After testing, for a video sequence with a size of 640*480 pixels in a 16vCPU Intel(R)Xeon(R)Platinum 8481C environment, the processing speed reaches more than 30FPS, and the video processing is smooth, which is suitable for real-time applications in dynamic environments.

[0077] After image stabilization, the inter-frame transform fidelity (ITF) of the video is significantly improved compared to the original video. This means that the image quality remains more stable and the details are clearer between different frames. This improvement has been verified in multiple static / motion test scenes, showing the significant advantages of the present invention in video quality, especially in scenes with irregular fast movement and severe shaking, which can effectively reduce jitter and blur.

[0078] Compared with other image stabilization methods, the present invention achieves a better balance between accuracy and efficiency. For example, traditional image stabilization methods may sacrifice processing speed while improving stability, while the present invention ensures efficient processing without reducing image quality through a multi-strategy fusion solution design. Through comparative analysis, it is proved that the image stabilization technology of the present invention can provide high-quality stabilization effects while maintaining real-time processing efficiency.

[0079] This method is particularly suitable for video stabilization applications with a wide range of scene requirements on mobile devices and low computing resource scenarios, allowing users to obtain smoother and clearer video data on different devices. The wide application potential of this technology will bring significant improvements to areas such as video surveillance, drone aerial photography, and mobile device photography.

[0080] The specific implementation of the present invention includes three aspects: motion estimation, motion compensation, and visual optimization.

[0081] 1. Motion Estimation Based on Feature Matching and Phase Correlation

[0082] In the technical solution of the present invention, the purpose of motion estimation is to identify and correct the inter-frame translation caused by camera shake. This method combines feature matching and phase correlation technology to improve accuracy and robustness.

[0083] Feature matching based on nR-AKAZE

[0084] The original video frame is converted into a single-channel grayscale image, and a nR-AKAZE algorithm is used to detect feature points P = {p1, p2, ..., p n}, each feature point p i With location and scale information. Generate a descriptor D for each feature point i , capturing local image features. The nR-AKAZE algorithm simplifies the operation used in the original AKAZE to process significant image rotation, that is, cancels the calculation of the main direction of the feature points in the AKAZE algorithm, sets the reference direction angle of the feature points to 0°, and cancels the rotation operation of the feature descriptor.

[0085] The distance between the descriptors of the feature points of the two frames I1 and I2 is calculated using the BF matching algorithm. Finding the best matching pair where d can be calculated using the Hamming distance:

[0086]

[0087] Where m is the length of the descriptor.

[0088] Apply the RANSAC algorithm to remove false matches. Set a model (e.g., translation or affine transformation model) and estimate the model through random sampling to identify matching points that do not conform to the model.

[0089] For the remaining valid matching points Calculate the displacement between them:

[0090]

[0091] Where N is the number of valid matching points, and are the x and y coordinates of the feature points in frame I1, and are the x and y coordinates of frame I2 and its corresponding valid feature points respectively.

[0092] Phase correlation method

[0093] Apply the preliminary translation (Δx1, Δy1) obtained by the nR-AKAZE algorithm to the video frame I1 to obtain the transformed frame I1′:

[0094] I1′(x,y)=I1(x+Δx1,y+Δy1)

[0095] Perform Fourier transform on the transformed frame I1′ and the reference frame I2:

[0096]

[0097] Compute the product of the complex conjugates of F1′ and F2:

[0098]

[0099] Perform an inverse Fourier transform on the product result to get the related image:

[0100]

[0101] Find the maximum position (x0, y0) in the correlation image R(x, y), which corresponds to the fine translation (Δx2, Δy2) between the two images. Superimpose the fine translation with the preliminary translation to obtain the final motion vector:

[0102] Δx=Δx1+Δx2

[0103] Δy=Δy1+Δy2

[0104] 2. Motion compensation based on parameter filtering and trajectory smoothing

[0105] The purpose of motion compensation is to eliminate the unintentional motion components caused by camera motion to improve video quality. This method uses a Gaussian filter to smooth the global motion vector of the current frame neighborhood, combining the unstable input frame and the stable output frame information instead of only considering the original input frame to achieve dynamic motion compensation.

[0106] Assume T i j represents the coordinate transformation from frame i to frame j, and the adjacent frames are represented as:

[0107] NF c ={n:ck≤n≤c+k}

[0108] Then, the smooth motion transformation can be expressed as:

[0109]

[0110] Among them, T USF (j) is the transformation parameter from the current frame to a specific stable output frame j, and T UOF (j) are the transformation parameters between the current frame and the unstable input frame j, and:

[0111]

[0112] This is a standard deviation of Gaussian kernel, * indicates convolution operation. This process is done by N blur iterations to achieve the best smoothing effect. Finally, the smoothed motion frame can be transformed from the original frame by:

[0113] I c ′=S c I c

[0114] 3. Visual optimization based on image transformation and zero-value padding

[0115] During the video stabilization process, when the current frame is corrected using the smoothed transformation parameters, a "black edge" phenomenon of zero pixel values ​​will appear due to the spatial position transformation of the image. In order to improve the visual effect while keeping the video size unchanged, these zero-value areas can be filled with the corresponding parts of the transformed (stabilized) previous frame to achieve a more natural transition. The pixel value of the tth frame at the (x, y) position after zero value filling is expressed as:

[0116]

[0117] Among them, I t ′(x,y) represents the pixel value at the (x,y) position in the tth frame after image transformation.

[0118] Finally, the parameters and encoding format of the output video are set to be consistent with the input video, and all consecutive stabilized frames are written to the video file in sequence to generate the stabilized output video.

[0119] The preferred embodiments of the present invention are described in detail above, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge scope of ordinary technicians in this field without departing from the purpose of the present invention.

[0120] Many other changes and modifications may be made without departing from the concept and scope of the present invention.It should be understood that the present invention is not limited to the specific embodiments, and the scope of the present invention is defined by the appended claims.

Claims

1. A general video stabilization method based on multiple strategy fusion, characterized in that: The method comprises: Motion estimation: Accurately estimate the translation of video frames through feature matching and phase correlation methods; Motion compensation: Based on the estimated translation amount, the motion vector is smoothed by applying Gaussian filtering to obtain the smoothed transformation parameters and the processed current frame information; Visual optimization: Based on the obtained transformation parameters and current frame information, the stabilized video sequence is obtained by using image transformation and zero-filling methods.

2. A general video stabilization method based on multiple strategy fusion according to claim 1, characterized in that: The motion estimation specifically includes: Use feature matching to preliminarily estimate the translation: Use the nR-AKAZE algorithm to detect feature points on video frames and quickly generate descriptors; use the BF algorithm to match descriptors through Hamming distance and find the best match by calculating the distance between feature points; use the RANSAC algorithm to remove incorrect matching points to improve matching accuracy and stability; calculate the displacement of the remaining valid matching points and average the displacement to obtain a preliminary translation component; Phase correlation fine estimation of translation: Preliminary motion estimation: The preliminary motion vector obtained by feature matching and other methods will be used to perform preliminary image transformation on the current frame to make it visually closer to the reference frame; Fourier transform: Fourier transform is performed on the current frame and the reference frame after preliminary transformation. This step converts the image from the spatial domain to the frequency domain, making the analysis of the translation transformation between images more efficient; Frequency domain processing: Multiply the Fourier transform result of the current frame with the conjugate complex number of the Fourier transform of the reference frame; This operation is to use the frequency domain characteristics to emphasize the phase information between images through complex operations and effectively filter out information related to translation; Inverse Fourier transform: Inverse Fourier transform is performed on the above product result to obtain a related image; In the related image, the value of each position indicates the degree of similarity between the current frame and the reference frame at the corresponding position; Locate the maximum value: In the related image, find the position of the maximum value; The position of the maximum value corresponds to the optimal translation between the current frame and the reference frame, that is, the fine motion vector; Motion vector superposition: The obtained fine motion vector is superimposed with the preliminary motion vector to obtain the final motion vector, which represents the comprehensive translation of the current frame relative to the reference frame.

3. The general video stabilization method based on multiple strategy fusion according to claim 1, characterized in that: The motion compensation specifically includes: Parameter filtering and trajectory smoothing: Smooth the unintentional motion in the final motion vector; introduce a Gaussian filter to perform real-time smoothing based on the global transformation parameters of the current frame and adjacent stable output frames and unstabilized input frames; use a Gaussian filter to perform neighborhood weighting and smoothing on the motion vectors of the current frame and consecutive frames to obtain the smoothed transformation parameters and the processed current frame information, thereby achieving stable video output.

4. The general video stabilization method based on multiple strategy fusion according to claim 1, characterized in that: The visual optimization specifically includes: Image transformation and black edge filling: Use the smoothed transformation parameters to correct the image of the current frame to reduce the impact of unintentional motion and retain intentional motion; for the black edges caused by image transformation, a zero-value filling method is used to fill the black edges of the corrected current frame with the stable output content of the previous frame to reduce the sense of visual incoherence; in this process, the corresponding part of the previous frame is seamlessly spliced ​​with the edge of the current frame to ensure the integrity of the image and the naturalness of the transition; finally, all the stabilized continuous frames are written into the video file in sequence to ensure the smoothness of the output video and improve the visual quality.

Images