Stable mirror operation simulation method based on ultra-high-definition virtual machine position

Abstract

The invention provides a stable mirror operation simulation method based on an ultra-high-definition virtual machine position, and relates to the technical field of image acquisition. When the 8K virtual camera shooting technology is used for carrying out mirror moving simulation, a moving track of a virtual lens is input, a filter is used for carrying out mirror moving smooth processing on the moving track, and a stable mirror moving simulation image is output; wherein the filter is internally provided with a damping coefficient used for adjusting the smooth processing granularity of the operation mirror, and corresponding small / medium / large damping coefficients are set based on a high-speed / common / low-speed operation mirror scene, so that the shooting effect similar to that of a professional physical stabilizer can be obtained without carrying out anti-shake processing on a picture under the condition of not carrying out excessive algorithm improvement; the method is suitable for various scenes such as sports events, common scenes and detail display; finally, the invention also provides a two-point scheme as a replacement of a central point scheme so as to obtain a more stable mirror moving effect.

Description

Technical Field

[0001] The present invention relates to the field of image acquisition technology, and in particular to a stable camera movement simulation method based on ultra-high-definition virtual machine positioning. Background Art

[0002] In professional videography, camera stability is crucial to obtaining high-quality images. To achieve this goal, cameras are often mounted on professional stabilization equipment such as cranes and Steadicams.

[0003] These devices use specific mechanical structures with inertia and damping effects to achieve stable push, pull, shake, shift and other framing operations. For example, the Steadicam, through its complex suspension system and balance mechanism, allows photographers to reduce camera shake when shooting on the move, thereby obtaining smooth and stable lens effects; but there are also problems such as high equipment cost and difficulty in deployment.

[0004] With the advancement of technology, the emergence of 8K virtual position shooting technology has provided new tools for cameramen. This technology allows users to capture part of the 8K ultra-high-definition picture and move and zoom in real time through control methods such as mouse, touch screen, and joystick, and simulate partial push, pull, shake, and move virtual framing.

[0005] This technology has been applied in Sony's XR multi-camera switching solution. Through the combination of software and hardware technology, it realizes real-time content preview and seamless switching of multiple cameras, improving shooting efficiency; however, there are still some limitations in practical applications; for example: these devices rely on digital signal sensing of fingers, which may cause stagnation, shaking, jumping and other problems, which limits their performance in simulating the effects of professional physical stabilizers.

[0006] Therefore, there is an urgent need for a stable camera movement simulation method based on ultra-high-definition virtual position to obtain professional camera movement shooting effects similar to those of a professional physical stabilizer. Summary of the invention

[0007] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a stable camera movement simulation method based on ultra-high-definition virtual position to obtain a professional camera movement shooting effect similar to a professional physical stabilizer.

[0008] The purpose of the present invention is achieved through the following technical solutions:

[0009] The present invention provides a stable camera movement simulation method based on ultra-high-definition virtual position. When using 8K virtual position shooting technology to perform camera movement simulation: the movement trajectory of the virtual lens is input, the movement trajectory is smoothed by using a filter, and a stable camera movement simulation image is output; wherein a damping coefficient for adjusting the granularity of the camera movement smoothing processing is provided in the filter, and corresponding small / medium / large damping coefficients are set based on fast / normal / slow camera movement scenes.

[0010] As a further solution, the following specific steps are used to simulate stable camera movement:

[0011] Input step 1: On the device screen of the ultra-high-definition virtual position, the user manually drags to control the movement of the virtual lens and records the movement trajectory of the lens center point over a period of time;

[0012] Input step 2: During the process of manually dragging and controlling the movement of the virtual lens, zoom the virtual lens multiple times and record the changes in the lens image size over a period of time;

[0013] Step 11: Fit the moving trajectory of the center point of the lens through a filter in the local area window; wherein the filter fits the moving trajectory through a polynomial curve and introduces it into the minimization objective function E1 of the least squares method;

[0014] Step 12: Derivative the polynomial coefficients of the minimized objective function E1 and set the result to zero, to obtain the linear equation of the polynomial coefficients of the fitting movement trajectory in step 11;

[0015] Step 13: Add the moving damping coefficient to the minimization objective function E1, and substitute the polynomial coefficient to calculate the moving smoothing value of the center of the local area window;

[0016] Step 21: Fitting the size change of the lens picture through a filter in the local area window; wherein the filter fits the size change of the lens picture through a polynomial curve and introduces it into the minimization objective function E2 of the least square method;

[0017] Step 22: Derivative the polynomial coefficients of the minimized objective function E2 and set the result to zero, to obtain a linear equation of the polynomial coefficients of the fitting movement trajectory in step 21;

[0018] Step 23: Add the scaling damping coefficient to the minimization objective function E2, and substitute the polynomial coefficient to calculate the scaling smoothing value of the center of the local area window;

[0019] Output steps: move the device screen at the center of the local area window by moving the smoothing value, and scale the device screen at the center of the local area window by scaling the smoothing value.

[0020] As a further solution, in step 1, the moving trajectory of the lens center point is represented by a horizontal axis coefficient and a vertical axis coefficient;

[0021] C t ={x t ,y t}; t∈{1,2,...,T};

[0022] Among them, t is the acquisition time variable, T is the total number of acquisition points, C t is the coordinate of the lens center point corresponding to the acquisition point at time t, x t is the horizontal axis coefficient corresponding to the lens center point coordinate at time t, y t is the vertical axis coefficient corresponding to the coordinate of the lens center point at time t; the moving trajectory of the center point is expressed as [C1, C2, ..., C T ].

[0023] As a further solution, in step 2, the lens image size is represented by a width coefficient and a height coefficient;

[0024] S t ={w t ,h t}; t∈{1,2,...,T};

[0025] Among them, t is the acquisition time variable, T is the total number of acquisition points, S t is the lens image size corresponding to the acquisition point at time t, w t is the width coefficient of the lens image size at time t, h t is the height coefficient corresponding to the lens image size at time t; the change of lens image size is expressed as [S1, S2, ..., S T ].

[0026] As a further solution, in step 11, the local area window is set to 2m+1 acquisition points, and the local area window center point set [C i-m , ..., C i-1 , C i , C i+1 ..., C i+m ] and the local area window screen size set [S i-m , ..., S i-1 , S i , S i+1 ..., S i+m ]; where i represents the center acquisition point number of the local area window center, and m represents the local area window half window.

[0027] As a further solution, the n-order polynomial P(x) is fitted to each central acquisition point in the local area window center point set:

[0028]

[0029] Among them, n represents the number of polynomial fitting, j represents the polynomial fitting variable, x represents the horizontal axis coefficient of the central acquisition point to be fitted, and x i Indicates the horizontal axis coefficient at the central collection point i, a j represents the polynomial coefficients used for the jth fit.

[0030] As a further solution, the minimization objective function E1 about the moving trajectory is:

[0031]

[0032] Among them, m represents the local area window half window, j represents the polynomial fitting variable, and y i+j represents the vertical axis coefficient at the center acquisition point i+j, P(x i+j ) represents the n-order polynomial P(x) at the central acquisition point i+j.

[0033] As a further solution, a moving damping coefficient is introduced into the minimization objective function E1:

[0034]

[0035] Among them, λ is the motion damping coefficient, k represents the polynomial damping variable, a k are the polynomial coefficients used for the k-th fit.

[0036] As a further solution, the horizontal axis coefficient and the vertical axis coefficient in the minimization objective function E1 are replaced by the width coefficient and the height coefficient, and the zoom damping coefficient is set to replace the movement damping coefficient, and the minimization objective function E2 about the size of the inner lens picture is obtained:

[0037]

[0038] Among them, h i+j represents the height coefficient at the center acquisition point i+j, P(w i+j ) represents the n-order polynomial P(w) at the central acquisition point i+j, w represents the width coefficient of the central acquisition point to be fitted, w i Represents the width coefficient at the center collection point i, Represents the scaling damping factor.

[0039] As a further solution, a corner point L is set at the lower left corner of the device screen. t, set the corner point R in the upper right corner of the device screen t , record the corner points L t and corner point R t The moving trajectory is set to minimize the objective function E1(L t ) and minimize the objective function E1(R t ), by minimizing the objective function E1(L t ) and minimize the objective function E1(R t ) are jointly regulated; among them,

[0040]

[0041] in, Indicates the corner point L t The coordinate at time t corresponds to the horizontal axis coefficient, Indicates the corner point L t The coordinate t corresponds to the vertical axis coefficient, λ l Indicates that λ l Indicates the corner point L t Set the movement damping coefficient; Represents the corner point R t The coordinate at time t corresponds to the horizontal axis coefficient, Represents the corner point R t The coordinate t corresponds to the vertical axis coefficient, λ r Represents the corner point R t Sets the motion damping factor.

[0042] As a further solution, the filter is set by an SG filter, a Kalman filter or a bilateral filter.

[0043] Compared with the related art, the stable camera movement simulation method based on ultra-high-definition virtual position provided by the present invention has the following advantages:

[0044] When the present invention uses 8K virtual machine position shooting technology to perform camera movement simulation: by inputting the movement trajectory of the virtual lens, using a filter to perform camera movement smoothing processing on the movement trajectory, and outputting a stable camera movement simulation image; wherein, the filter is provided with a damping coefficient for adjusting the granularity of the camera movement smoothing processing, and based on the fast / normal / slow camera movement scenes, corresponding small / medium / large damping coefficients are set, so that without making too much algorithm improvement and without anti-shake processing of the picture, a shooting effect similar to that of a professional physical stabilizer can be obtained, which is suitable for various scenes such as sports events, ordinary scenes and detail displays; finally, the present invention also provides a two-point solution as an alternative to the center point solution to obtain a more stable camera movement effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1A schematic diagram of the steps of a stable camera movement simulation method based on ultra-high-definition virtual position provided by the present invention;

[0046] Figure 2 A schematic diagram of specific steps for performing stable camera movement simulation provided by the present invention;

[0047] Figure 3 A schematic diagram of the center point trajectory movement and width and height scaling provided by the present invention;

[0048] Figure 4 A schematic diagram of two-point trajectory movement and width-height scaling provided by the present invention;

[0049] Figure 5 This is a comparison chart of the smoothing results of different filters provided by the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0051] See also Figure 1 The present invention provides a stable camera movement simulation method based on ultra-high-definition virtual position. When using 8K virtual position shooting technology to perform camera movement simulation: by inputting the movement trajectory of the virtual lens, using a filter to perform camera movement smoothing processing on the movement trajectory, and outputting a stable camera movement simulation image; wherein a damping coefficient for adjusting the granularity of the camera movement smoothing processing is set in the filter, and based on fast / normal / slow camera movement scenes, corresponding small / medium / large damping coefficients are set.

[0052] It should be noted that: this embodiment adds a smoothing algorithm to the selection box (corresponding to the "virtual lens"); that is, when the user moves the selection box or scales the selection box, the timing data of the user operation itself is smoothed by the smoothing algorithm on the basis of recording the user operation data. In this way, there is no need to perform anti-shake processing on the picture, thereby obtaining a professional camera movement effect similar to that of a professional physical stabilizer.

[0053] See also Figure 2 As a further solution, the following specific steps are used to simulate the stable camera movement:

[0054] Input step 1: On the device screen of the ultra-high-definition virtual position, the user manually drags to control the movement of the virtual lens and records the movement trajectory of the lens center point over a period of time;

[0055] Input step 2: During the process of manually dragging and controlling the movement of the virtual lens, zoom the virtual lens multiple times and record the changes in the lens image size over a period of time;

[0056] Step 11: Fit the moving trajectory of the center point of the lens through a filter in the local area window; wherein the filter fits the moving trajectory through a polynomial curve and introduces it into the minimization objective function E1 of the least squares method;

[0057] Step 12: Derivative the polynomial coefficients of the minimized objective function E1 and set the result to zero, to obtain the linear equation of the polynomial coefficients of the fitting movement trajectory in step 11;

[0058] Step 13: Add the moving damping coefficient to the minimization objective function E1, and substitute the polynomial coefficient to calculate the moving smoothing value of the center of the local area window;

[0059] Step 21: Fitting the size change of the lens picture through a filter in the local area window; wherein the filter fits the size change of the lens picture through a polynomial curve and introduces it into the minimization objective function E2 of the least square method;

[0060] Step 22: Derivative the polynomial coefficients of the minimized objective function E2 and set the result to zero, to obtain a linear equation of the polynomial coefficients of the fitting movement trajectory in step 21;

[0061] Step 23: Add the scaling damping coefficient to the minimization objective function E2, and substitute the polynomial coefficient to calculate the scaling smoothing value of the center of the local area window;

[0062] Output steps: move the device screen at the center of the local area window by moving the smoothing value, and scale the device screen at the center of the local area window by scaling the smoothing value.

[0063] It should be noted that the smoothing algorithm mainly smoothes the moving trajectory and scaling of the center point of the virtual lens to simulate the damping of the camera arm and lens focusing in reality, thereby maintaining the stability of the imaging and reducing the jitter of the video picture; among them, the collected data is smoothed. In the stage of selecting the filter, the Savitzky-Golay filter (hereinafter referred to as the SG filter) uses a polynomial curve fitted in a local area window and fits the data by the least squares method in the main time domain. In this way, the filtering can reduce noise while maintaining the characteristics of the original data. The disadvantage is that the window size and polynomial order need to be defined in advance.

[0064] This embodiment cleverly utilizes the stage of "needing to define the window size and polynomial order in advance" and introduces the damping coefficient at this stage The degree of smoothness is controlled by adjusting the local area window size (the window size can be set according to needs) and the polynomial order (a larger coefficient will produce a smoother curve but will also lose more details), thereby achieving stable camera movement simulation of ultra-high-definition virtual positions without making too many improvements.

[0065] As a further solution, in step 1, the moving trajectory of the lens center point is represented by a horizontal axis coefficient and a vertical axis coefficient;

[0066] C t ={x t ,y t}; t∈{1,2,...,T};

[0067] Among them, t is the acquisition time variable, T is the total number of acquisition points, C t is the coordinate of the lens center point corresponding to the acquisition point at time t, x t is the horizontal axis coefficient corresponding to the lens center point coordinate at time t, y t is the vertical axis coefficient corresponding to the coordinate of the lens center point at time t; the moving trajectory of the center point is expressed as [C1, C2, ..., C T ].

[0068] It should be noted that the collection points of the moving trajectory and the lens image size correspond to each other, which is conducive to matching the subsequent moving smoothing value and scaling smoothing value to specific points and providing data support for subsequent fitting operations.

[0069] As a further solution, in step 2, the lens image size is represented by a width coefficient and a height coefficient;

[0070] S t ={w t ,h t}; t∈{1,2,...,T};

[0071] Among them, t is the acquisition time variable, T is the total number of acquisition points, S t is the lens image size corresponding to the acquisition point at time t, w t is the width coefficient of the lens image size at time t, h t is the height coefficient corresponding to the lens image size at time t; the change of lens image size is expressed as [S1, S2, ..., S T ].

[0072] As a further solution, in step 11, the local area window is set to 2m+1 acquisition points, and the local area window center point set [C i-m , ..., C i-1 , C i , Ci+1 ..., C i+m ] and the local area window screen size set [S i-m , ..., S i-1 , S i , S i+1 ..., S i+m ]; where i represents the center acquisition point number of the local area window center, and m represents the local area window half window.

[0073] It should be noted that this step is mainly to determine the size of the local area window and ensure that the center point of the local area window can be found (2m+1 is an odd number), and then construct a set of local area window center points at different points and match them with the picture size one by one, and provide data support for subsequent fitting operations.

[0074] As a further solution, the n-order polynomial P(x) is fitted to each central acquisition point in the local area window center point set:

[0075]

[0076] Among them, n represents the number of polynomial fitting, j represents the polynomial fitting variable, x represents the horizontal axis coefficient of the central acquisition point to be fitted, and x i Indicates the horizontal axis coefficient at the central collection point i, a j represents the polynomial coefficients used for the jth fit.

[0077] It should be noted that a larger polynomial coefficient will result in a smoother curve, but it will also lose more details of the user's operation. Here we can determine a suitable polynomial coefficient through multiple debugging, so as to ensure the smoothness of the curve and retain the detailed operations of the user.

[0078] In addition, we can also adjust the damping coefficient to adapt to different application scenarios. In relatively intense scenes such as sports events, we use a smaller λ damping coefficient to ensure the smoothness of the virtual lens movement and sacrifice a little stability; in ordinary scenes, we use a moderate λ damping coefficient to balance stability and smoothness; in scenes that display details, we use a larger λ damping coefficient, which will relatively give up some authenticity, thereby greatly improving the stability of the lens.

[0079] As a further solution, the minimization objective function E1 about the moving trajectory is:

[0080]

[0081] Among them, m represents the local area window half window, j represents the polynomial fitting variable, and y i+jrepresents the vertical axis coefficient at the center acquisition point i+j, P(x iij ) represents the n-order polynomial P(x) at the central acquisition point i+j.

[0082] As a further solution, a moving damping coefficient is introduced into the minimization objective function E1:

[0083]

[0084] Among them, λ is the motion damping coefficient, k represents the polynomial damping variable, a k are the polynomial coefficients used for the kth fit.

[0085] As a further solution, the horizontal axis coefficient and the vertical axis coefficient in the minimization objective function E1 are replaced by the width coefficient and the height coefficient, and the zoom damping coefficient is set to replace the movement damping coefficient, and the minimization objective function E2 about the size of the inner lens picture is obtained:

[0086]

[0087] Among them, h i+j represents the height coefficient at the center acquisition point i+j, P(w i+j ) represents the n-order polynomial P(w) at the central acquisition point i+j, w represents the width coefficient of the central acquisition point to be fitted, w i Represents the width coefficient at the center collection point i, Represents the scaling damping factor.

[0088] As a further solution, a corner point L is set at the lower left corner of the device screen. t , set the corner point R in the upper right corner of the device screen t , record the corner points L t and corner point R t The moving trajectory is set to minimize the objective function E1(L t ) and minimize the objective function E1(R t ), by minimizing the objective function E1(L t ) and minimize the objective function E1(R t ) are jointly regulated; among them,

[0089]

[0090] in, Indicates the corner point L t The coordinate at time t corresponds to the horizontal axis coefficient, Indicates the corner point L t The coordinate t corresponds to the vertical axis coefficient, λ l Indicates that λ lIndicates the corner point L t Set the movement damping coefficient; Represents the corner point R t The coordinate at time t corresponds to the horizontal axis coefficient, Represents the corner point R t The coordinate t corresponds to the vertical axis coefficient, λ r Represents the corner point R t Sets the motion damping factor.

[0091] It should be noted that: in this embodiment, the movement and zooming of the virtual lens on the screen are operated by smoothing the two diagonal points of the lens frame (two-point solution, the specific operation is shown as follows Figure 4 As shown), the picture is kept stable; compared with the center point solution (the specific operation is shown as Figure 3 As shown in the figure, it has better camera movement stability. The disadvantage is that it will increase the amount of calculation, resulting in increased camera movement delay. We can choose to use the center point solution or the two-point solution according to actual needs.

[0092] As a further solution, the filter is set by an SG filter, a Kalman filter or a bilateral filter.

[0093] It should be noted that no matter which filter is used for fitting, our main purpose is to introduce the damping coefficient into the smoothing algorithm to imitate the impact of the jib and lens focus on the picture in the real scene. Through the smoothing algorithm, the lens picture presented by the virtual position is more stable and in line with reality. At the same time, different degrees of smoothness can be achieved by controlling the damping coefficient to meet the needs of different scenes.

[0094] In a specific embodiment, we conducted a simple smoothing test and compared the SG filter, the Kalman filter and the bilateral filter to obtain the influence of different filters on the smoothing result. The comparison results are shown in FIG. Figure 5 As shown in the figure, the SG filter performs best in the test, so we give priority to using the SG filter. However, this does not affect our ability to replace the SG filter with a Kalman filter, a bilateral filter, etc., according to actual conditions, and adjust the process noise covariance matrix Q and the observation covariance matrix R of the Kalman filter through the damping coefficient to cope with more special scenarios.

[0095] The above embodiments only express preferred implementation modes, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present invention. It should be pointed out that, for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention.

Claims

1. A stable camera motion simulation method based on ultra-high-definition virtual position, characterized in that: When using 8K virtual machine position shooting technology for camera movement simulation: by inputting the movement trajectory of the virtual lens, use the filter to smooth the movement trajectory, and output a stable camera movement simulation image; wherein, the filter is provided with a damping coefficient for adjusting the granularity of the camera movement smoothing processing, and based on the fast / normal / slow camera movement scenes, the corresponding small / medium / large damping coefficients are set.

2. The method for simulating stable camera movement based on ultra-high-definition virtual position according to claim 1, characterized in that: The stable camera movement simulation is carried out through the following specific steps: Input step 1: On the device screen of the ultra-high-definition virtual position, the user manually drags to control the movement of the virtual lens and records the movement trajectory of the lens center point over a period of time; Input step 2: During the process of manually dragging and controlling the movement of the virtual lens, zoom the virtual lens multiple times and record the changes in the lens image size over a period of time; Step 11: Fit the moving trajectory of the center point of the lens through a filter in the local area window; wherein the filter fits the moving trajectory through a polynomial curve and introduces it into the minimization objective function E1 of the least squares method; Step 12: Derivative the polynomial coefficients of the minimized objective function E1 and set the result to zero, to obtain the linear equation of the polynomial coefficients of the fitting movement trajectory in step 11; Step 13: Add the moving damping coefficient to the minimization objective function E1, and substitute the polynomial coefficient to calculate the moving smoothing value of the center of the local area window; Step 21: Fitting the size change of the lens picture through a filter in the local area window; wherein the filter fits the size change of the lens picture through a polynomial curve and introduces it into the minimization objective function E2 of the least square method; Step 22: Derivative the polynomial coefficients of the minimized objective function E2 and set the result to zero, to obtain a linear equation of the polynomial coefficients of the fitting movement trajectory in step 21; Step 23: Add the scaling damping coefficient to the minimization objective function E2, and substitute the polynomial coefficient to calculate the scaling smoothing value of the center of the local area window; Output steps: move the device screen at the center of the local area window by moving the smoothing value, and scale the device screen at the center of the local area window by scaling the smoothing value.

3. The method for simulating stable camera movement based on ultra-high-definition virtual position according to claim 2, characterized in that: In step 1, the moving trajectory of the lens center point is represented by the horizontal axis coefficient and the vertical axis coefficient; C t ={x t ,y t };t∈{1,2,...,T}; Among them, t is the acquisition time variable, T is the total number of acquisition points, C t is the coordinate of the lens center point corresponding to the acquisition point at time t, x t is the horizontal axis coefficient corresponding to the lens center point coordinate at time t, y t is the vertical axis coefficient corresponding to the coordinate of the lens center point at time t; the moving trajectory of the center point is expressed as [C1, C2, ..., C T ].

4. The method for simulating stable camera movement based on ultra-high-definition virtual position according to claim 3, characterized in that: In step 2, the lens image size is represented by a width coefficient and a height coefficient; S t ={w t ,h t };t∈{1,2,...,T}; Among them, t is the acquisition time variable, T is the total number of acquisition points, S t is the lens image size corresponding to the acquisition point at time t, w t is the width coefficient of the lens image size at time t, h t is the height coefficient corresponding to the lens image size at time t; the change of lens image size is expressed as [S1, S2, ..., S T ].

5. The method for simulating stable camera movement based on ultra-high-definition virtual position according to claim 4, characterized in that: In step 11, the local area window is set to 2m+1 collection points, and the local area window center point set [C i-m , ..., C i-1 , C i , C i+1 ..., C i+m ] and the local area window screen size set [S i-m , ..., S i-1 , S i , S i+1 ..., S i+m ]; where i represents the center acquisition point number of the local area window center, and m represents the local area window half window.

6. The method for simulating stable camera movement based on ultra-high-definition virtual position according to claim 5, characterized in that: Fit the n-order polynomial P(x) to each central acquisition point in the local area window center point concentration: Among them, n represents the number of polynomial fitting, j represents the polynomial fitting variable, x represents the horizontal axis coefficient of the central acquisition point to be fitted, and x i Indicates the horizontal axis coefficient at the central collection point i, a j represents the polynomial coefficients used for the jth fit.

7. The method for simulating stable camera movement based on ultra-high-definition virtual position according to claim 6, characterized in that: The minimization objective function E1 about the moving trajectory is: Among them, m represents the local area window half window, j represents the polynomial fitting variable, and y i+j represents the vertical axis coefficient at the center acquisition point i+j, P(x iij ) represents the n-order polynomial P(x) at the central acquisition point i+j.

8. The method for simulating stable camera movement based on ultra-high-definition virtual position according to claim 7, characterized in that: The moving damping coefficient is introduced into the minimization objective function E1: Among them, λ is the motion damping coefficient, k represents the polynomial damping variable, a k are the polynomial coefficients used for the kth fit.

9. The method for simulating stable camera movement based on ultra-high-definition virtual position according to claim 7, characterized in that: Regarding the minimization objective function E2 of the inner lens image size: Among them, h i+j represents the height coefficient at the center acquisition point i+j, P(w i+j ) represents the n-order polynomial P(w) at the central acquisition point i+j, w represents the width coefficient of the central acquisition point to be fitted, w i Represents the width coefficient at the center collection point i, Represents the scaling damping factor.

10. The method for simulating stable camera movement based on ultra-high-definition virtual position according to claim 8, characterized in that: Also set the corner point L in the lower left corner of the device screen t , set the corner point R in the upper right corner of the device screen t , record the corner points L t and corner point R t The moving trajectory is set to minimize the objective function E1(L t ) and minimize the objective function E1(R t ), by minimizing the objective function E1(L t ) and minimize the objective function E1(R t ) are jointly regulated; among them, in, Indicates the corner point L t The coordinate at time t corresponds to the horizontal axis coefficient, Indicates the corner point L t The coordinate t corresponds to the vertical axis coefficient, λ l Indicates that λ l Indicates the corner point L t Set the movement damping coefficient; Represents the corner point R t The coordinate at time t corresponds to the horizontal axis coefficient, Represents the corner point R t The coordinate t corresponds to the vertical axis coefficient, λ r Represents the corner point R t Sets the motion damping factor.

Images