Full-automatic shooting control method based on stability control algorithm

Abstract

The invention relates to the technical field of automatic shooting, and discloses a full-automatic shooting control method based on a stability control algorithm, and the method comprises the steps: constructing a three-dimensional rectangular coordinate system through a main camera and an auxiliary camera, synchronously collecting images, extracting the feature points of the images, and calculating the three-dimensional coordinates of a target through binocular parallax. A trend prediction algorithm is constructed by using three-frame image data, and the three-dimensional position of the target in the next frame is accurately predicted. According to the prediction result, the steering angles needed by the main camera and the auxiliary camera are calculated respectively, and then the corresponding rotation angular speeds are calculated through a stable control function. Introducing an angular velocity compensation mechanism, dynamically adjusting the angular velocity in combination with the target acceleration and the motor response inertial parameters, monitoring the shooting directions of the two cameras in real time, judging whether the view included angles of the two cameras are overlapped or not, extracting an image edge point set, judging whether intersection exists or not, and once it is confirmed that pictures are overlapped, judging whether the intersection exists or not; therefore, the yaw angles of the two cameras can be adjusted, so that the shooting areas are separated again.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic shooting, and in particular to a full-automatic shooting control method based on a stable control algorithm. Background Art

[0002] In modern multi-scene filming applications, such as stage performances, sports events, drone tracking, and intelligent surveillance, the accuracy and stability of automated filming technology are critical factors affecting film quality and target visibility. Traditional filming control methods often struggle to achieve stable tracking and accurate focus, especially under complex conditions such as rapid target motion, frequent occlusions, or overlapping fields of view. Currently, most mainstream automated filming systems rely on image recognition combined with single-camera gimbal control, performing simple target detection and velocity estimation for directional correction. These approaches suffer from three core limitations: First, they cannot accurately determine three-dimensional spatial position, relying only on a two-dimensional image to estimate the target's position, resulting in significant errors. Second, most systems employ proportional control models or rely on empirical weighting coefficients for angular velocity adjustment. These response strategies are not universal and require frequent manual tuning across different devices, which can easily lead to system instability or overshoot. Third, existing solutions ignore the potential for "field of view interference" in multi-camera collaborative control, which can easily lead to resource redundancy, filming interference, and even target loss or loss of focus in real-world scenarios. In addition, during dynamic shooting, due to the inertia of the device itself, the motor response always lags behind the actual changes of the target. Traditional methods rarely model and compensate for the angular velocity deviation caused by this "inertial disturbance", which makes the shooting results perform poorly in terms of image stability and angle continuity.

[0003] To address these issues, a fully automatic shooting control method with rigorous logic, complete structure, high practicality, and innovation is urgently needed. This method should be able to solve three-dimensional coordinates based on the camera's dual viewpoint structure, predict target motion by combining multi-frame trend prediction, and achieve high-precision and stable control through a weight-free control function and angular velocity compensation mechanism. Furthermore, this method must be able to automatically identify and avoid overlapping fields of view of multiple cameras, ensuring stable tracking while achieving coordinated image distribution, avoiding field of view conflicts, and improving overall film quality and system intelligence. Summary of the Invention

[0004] The present invention provides a fully automatic shooting control method based on a stable control algorithm, which is used to promote the solution of the problems mentioned in the above background technology.

[0005] The present invention provides the following technical solution: a fully automatic shooting control method based on a stable control algorithm, comprising: Construct a three-dimensional rectangular coordinate system, collect images and calculate the three-dimensional coordinates of the target; Predict the three-dimensional coordinates of the target in the next frame through the three-point trend prediction algorithm; Calculate the steering angles that the main camera and auxiliary camera need to adjust based on the 3D coordinates of the target in the next frame; Calculate the angular velocity of the main camera and auxiliary camera through the stable control function; Calculate angular velocity compensation and optimize the angular velocity of the main camera and auxiliary camera; Monitor the rotation of the main camera and auxiliary camera in real time, and execute adjustment strategies when overlapping images occur.

[0006] Optionally, constructing a three-dimensional rectangular coordinate system, performing image acquisition, and calculating the three-dimensional coordinates of the target includes: Get the main viewpoint camera and auxiliary viewpoint cameras ; pass and Real-time image acquisition; Get and The images in the tth frame are respectively recorded as ; Where t is the frame number; Establish a three-dimensional rectangular coordinate system, specifically: Take the optical center of the main camera as the coordinate origin ; Starting from the coordinate origin, draw a horizontal ray to the right on the main camera image as the horizontal axis; Starting from the coordinate origin, draw a vertical downward ray on the main camera image as the vertical axis; Starting from the coordinate origin, along the direction of the main camera pointing to the scene, draw a ray perpendicular to the image plane extending outward as the depth direction; Construct a feature point extraction algorithm as follows: in: is the main photographic response graph; is the auxiliary response diagram; and They are the pixel points in the main response map and the auxiliary response map respectively Gray value of and They are The first-order gradient of the inner pixel in the horizontal and vertical directions; and They are The first-order gradient of the inner pixel in the horizontal and vertical directions; The first-order gradient is used to measure the change in grayscale value of a pixel in the horizontal and vertical directions. The faster the change, the larger the value. Extract separately The maximum response value is taken as the target feature point coordinate ; Construct a standard binocular disparity formula and calculate the three-dimensional coordinates of the target based on the t-th frame image, as follows: in: F is the main view camera and auxiliary viewpoint cameras focal length; D is the main view camera and auxiliary viewpoint cameras the spacing between; They are and The center pixel coordinates of is the three-dimensional coordinate of the target when the t-th frame image is collected.

[0007] Optionally, predicting the three-dimensional coordinates of the target in the next frame by using a three-point trend prediction algorithm includes: Construct a three-point trend prediction algorithm as follows: The lateral acceleration of the target at frame t : in: and are the speed of the target in the horizontal direction at the t-th frame image and the t-1-th frame image respectively; is the interval between image acquisitions; The longitudinal acceleration of the target at frame t : in: and are the speed of the target in the vertical direction at the t-th frame image and the t-1-th frame image respectively; The acceleration of the target in the depth direction at frame t : in: and are the speed of the target in the depth direction at the t-th frame image and the t-1-th frame image respectively; Predict the position of the target on the horizontal axis when the t+1 frame image is collected: Predict the target's position on the vertical axis when the t+1th frame image is collected: Predict the position of the target in the depth direction when the t+1th frame image is collected: in, Represents the three-dimensional coordinates of the target at the t+1th frame image.

[0008] Optionally, calculating the steering angles that the main camera and the auxiliary camera need to adjust according to the three-dimensional coordinates of the target in the next frame includes: Adjust the main view camera at the same time and auxiliary viewpoint cameras Steering angle; Define the current moment The direction vector is , the predicted target direction vector is ; Construct the camera steering angle calculation formula as follows: in: is the three-dimensional coordinate of the camera; for and The angle between them is the steering angle; They are and The vector modulus of .

[0009] Optionally, the calculating the angular velocity of the main camera and the auxiliary camera by using a stabilization control function includes: Construct a stable control function and calculate the target angular velocity , as follows: in: Indicates the response cycle; is the number of feedback increment cycles, which is 3 frames in this scheme; is the rate of change of angle.

[0010] Optionally, the calculating of angular velocity compensation to optimize the angular velocity of rotation of the main camera and the auxiliary camera includes: Set the acceleration vector ; in, are the inertial disturbance acceleration of the target in the horizontal, vertical and depth directions respectively. The specific calculation process is as follows: Construct the compensation correction function as follows: in: is angular velocity compensation; The inertial parameters of the camera response; is the acceleration vector The specific expression is ; calculate The final angular velocity , specifically: .

[0011] Optionally, the real-time monitoring of the rotation of the main camera and the auxiliary camera and executing an adjustment strategy when image overlap occurs include: Get the current time Direction vector ; Define the current moment The direction vector is ; Calculate the field of view angle between the main camera and the auxiliary camera , as follows: Get the horizontal field of view angle of the main camera and auxiliary camera ; The horizontal field of view angle refers to the angle of the field of view that the camera can cover in the horizontal direction; like , it is determined that there is overlap between the field of view of the main camera and the auxiliary camera, and the adjustment strategy is executed; The adjustment strategy is as follows: Extract separately and The edge point set in , specifically: in, and Respectively and The maximum value of the corresponding responsivity of all pixels in ; Determine whether the images from the main camera and the auxiliary camera overlap: in, Represents the edge point set Coordinates within the intersection; Get the yaw angle of the current main camera and the yaw angle of the auxiliary camera ; The yaw angle of the main camera and the yaw angle of the auxiliary camera Make adjustments as follows: .

[0012] The present invention has the following beneficial effects: This fully automatic shooting control method, based on a stable control algorithm, achieves high-precision, real-time calculation of the target's position in three-dimensional space through three-dimensional coordinate construction and image acquisition. A dual-camera primary and secondary viewpoint system is constructed, with the optical center of the primary camera as the coordinate origin. A spatial rectangular coordinate system is constructed in the horizontal, vertical, and depth directions, ensuring that the spatial calculation of the target in the image has a clear physical meaning. Edge extraction is performed using a proprietary response point difference operator. This operator constructs a response value function based on the horizontal and vertical differences in pixel grayscale within the image, and extracts highly responsive points from each of the primary and secondary images. By setting the response threshold to 20% of the maximum value, it effectively preserves structurally significant edge regions in the image, eliminates background and invalid feature interference, and provides stable keypoint input for three-dimensional coordinate calculation. This method, based on accessible parameters and a realistic physical model, effectively improves the stability and solveability of image features. It can accurately identify the target's three-dimensional position, especially in scenes with complex backgrounds or drastic lighting changes, resolving the practical problem of unstable target position accuracy caused by background interference in existing techniques.

[0013] 2. This fully automatic capture control method, based on a stable control algorithm, uses the target's coordinate changes across three consecutive image frames to construct a velocity and acceleration model, enabling advance prediction of the target's position in the next frame. This method abandons the traditional approach of relying solely on differential calculations between the first two frames and instead uses a joint modeling approach based on data from all three frames. By extracting instantaneous acceleration values ​​in the lateral, longitudinal, and depth directions and combining them with the current velocity vector, the predicted target coordinates at the next moment are calculated. This algorithm offers two significant advantages: First, it accounts for nonlinear motion characteristics such as non-uniform speed or sudden changes in direction, making it widely applicable to tracking targets with dramatic or erratic movements. Second, all required data is derived from spatial coordinate changes within the image frames, allowing parameters to be directly derived without the need for fuzzy factors, ensuring the determinism and stability of the computational model. By precalculating the target's potential position, the system provides "pre-aiming" capability, enabling adjustment commands to be executed before the target moves rapidly or is obscured. This significantly reduces jitter or target deviation caused by tracking delay, effectively improving the smoothness and robustness of tracking capture.

[0014] 3. This is a fully automatic shooting control method based on a stable control algorithm. By introducing the three-dimensional direction vector angle as a control variable, the adjustment angle of the camera is uniformly incorporated into the space vector framework for processing. Specifically, the vector formed by the current camera's line of sight direction and the predicted target position is used as input to calculate its spatial angle, thereby obtaining the required steering amplitude. This method avoids the coordinate conversion error and axial interference that may be caused by the traditional angle decomposition method. At the same time, by modeling the main camera and the auxiliary camera separately, the dual-camera system does not produce coupling interference during the adjustment process. The calculation of the direction angle is based on the vector dot product relationship. All parameters in the calculation are derived from the actual measured spatial coordinates, with clear dimensions and traceability. This control strategy solves the steering error problem caused by different reference directions and positions of each camera in multi-camera collaborative control. It is particularly suitable for application scenarios where two or more cameras have different initial viewing directions and inconsistent installation angles. It significantly improves the consistency of system control and steering accuracy, laying a technical foundation for high-performance collaborative shooting.

[0015] 4. This fully automatic shooting control method based on a stable control algorithm achieves dynamic adjustment of the angular velocity output by constructing a combined control model using angle deviation and angular velocity change rate as inputs. The control function uses a response period parameter as the basis for the adjustment rate, set to 0.03 seconds to match the typical 30 frames per second system refresh cycle, enabling synchronized control of each frame. A feedback period parameter is also introduced to construct an average angular velocity change rate over three frames to filter out transient misjudgments caused by abnormal jitter in a single frame, thereby maintaining the continuity and stability of the control system. This function ensures fast response while effectively suppressing overshoot or oscillation during rapid tracking, preventing camera overshoot or image wobbling caused by rapid angle adjustments. Compared to traditional proportional control methods, this control function uses parameters derived from system time and image acquisition rate. This is deterministic and eliminates the need for fuzzy parameter tuning, improving the algorithm's versatility and deployment efficiency, making it suitable for high-frame-rate, high-sensitivity automatic tracking scenarios.

[0016] 5. This is a fully automatic shooting control method based on a stable control algorithm. It calculates the spatial angle between the current shooting directions of the main camera and the auxiliary camera, and compares it with their respective horizontal field of view angles. When the angle is less than the field of view angle, it is determined that there is an overlapping area. At the same time, through the edge point extraction algorithm, high-response structure edges are extracted from the images of the two cameras. A threshold is set based on 20% of the maximum response value to effectively exclude invalid background and weak feature areas, thereby enhancing the accuracy of overlap detection. When it is determined that an overlap occurs, the system performs distributed fine-tuning on the yaw angles of the two cameras to minimize the intersection of fields of view while maintaining target coverage, thereby improving the information coverage efficiency and resource allocation rationality of the entire system. This design is particularly suitable for multi-viewpoint intelligent shooting scenarios such as multi-camera drones and multi-axis rail vehicles. It solves the problems of image redundancy and information conflict caused by overlapping fields of view, and has practical engineering significance and innovative value. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] Example 1, see Figure 1 , a fully automatic shooting control method based on a stable control algorithm, comprising: Construct a three-dimensional rectangular coordinate system, collect images and calculate the three-dimensional coordinates of the target; Predict the three-dimensional coordinates of the target in the next frame through the three-point trend prediction algorithm; Calculate the steering angles that the main camera and auxiliary camera need to adjust based on the 3D coordinates of the target in the next frame; Calculate the angular velocity of the main camera and auxiliary camera through the stable control function; Calculate angular velocity compensation and optimize the angular velocity of the main camera and auxiliary camera; Monitor the rotation of the main camera and auxiliary camera in real time, and execute adjustment strategies when overlapping images occur.

[0020] The process of constructing a three-dimensional rectangular coordinate system, acquiring images, and calculating the three-dimensional coordinates of the target includes: Get the main viewpoint camera and auxiliary viewpoint cameras ; pass and Real-time image acquisition; Get and The images in the tth frame are respectively recorded as ; Where t is the frame number; Establish a three-dimensional rectangular coordinate system, specifically: Take the optical center of the main camera as the coordinate origin ; Starting from the coordinate origin, draw a horizontal ray to the right on the main camera image as the horizontal axis; Starting from the coordinate origin, draw a vertical downward ray on the main camera image as the vertical axis; Starting from the coordinate origin, along the direction of the main camera pointing to the scene, draw a ray perpendicular to the image plane extending outward as the depth direction; Construct a feature point extraction algorithm as follows: in: is the main photographic response graph; is the auxiliary response diagram; and They are the pixel points in the main response map and the auxiliary response map respectively Gray value of and They are The first-order gradient of the inner pixel in the horizontal and vertical directions; and They are The first-order gradient of the inner pixel in the horizontal and vertical directions; The first-order gradient is used to measure the change in grayscale value of a pixel in the horizontal and vertical directions. The faster the change, the larger the value. Extract separately The maximum response value is taken as the target feature point coordinate ; Construct a standard binocular disparity formula and calculate the three-dimensional coordinates of the target based on the t-th frame image, as follows: in: F is the main view camera and auxiliary viewpoint cameras focal length; D is the main view camera and auxiliary viewpoint cameras the spacing between; They are and The center pixel coordinates of is the three-dimensional coordinate of the target when the t-th frame image is collected.

[0021] Through three-dimensional coordinate construction and image acquisition methods, a high-precision, real-time solution for the target's position in three-dimensional space is achieved. A dual-camera primary and secondary viewpoint system is constructed, with the optical center of the primary camera as the coordinate origin. A spatial rectangular coordinate system is constructed in the horizontal, vertical, and depth directions, ensuring that the spatial solution of the target in the image has a clear physical meaning. Edge extraction is performed using a proprietary response point difference operator. This operator constructs a response value function based on the horizontal and vertical differences in pixel grayscale within the image, and extracts highly responsive points from each of the primary and secondary images. By setting the response threshold to 20% of the maximum value, structurally significant edge regions in the image are effectively retained, eliminating background and invalid feature interference, and providing stable keypoint input for three-dimensional coordinate calculation. This method, based on accessible parameters and a realistic physical model, effectively improves the stability and solvability of image features. It can accurately identify the target's three-dimensional position, especially in scenes with complex backgrounds or drastic lighting changes, resolving the practical problem of unstable target position accuracy caused by background interference in existing technologies.

[0022] The three-point trend prediction algorithm is used to predict the three-dimensional coordinates of the target in the next frame, including: Construct a three-point trend prediction algorithm as follows: The lateral acceleration of the target at frame t : in: and are the speed of the target in the horizontal direction at the t-th frame image and the t-1-th frame image respectively; is the interval between image acquisitions; The longitudinal acceleration of the target at frame t : in: and are the speed of the target in the vertical direction at the t-th frame image and the t-1-th frame image respectively; The acceleration of the target in the depth direction at frame t : in: and are the speed of the target in the depth direction at the t-th frame image and the t-1-th frame image respectively; Predict the position of the target on the horizontal axis when the t+1 frame image is collected: Predict the target's position on the vertical axis when the t+1th frame image is collected: Predict the position of the target in the depth direction when the t+1th frame image is collected: in, Represents the three-dimensional coordinates of the target at the t+1th frame image.

[0023] This method uses the target's coordinate changes across three consecutive image frames to construct a velocity and acceleration model, enabling advance prediction of the target's position in the next frame. This approach abandons the traditional approach of relying solely on differential calculations between the first two frames and instead uses a joint model based on data from all three frames. By extracting instantaneous acceleration values ​​in the lateral, longitudinal, and depth directions and combining them with the current velocity vector, the predicted target coordinates at the next moment are calculated. This algorithm offers two significant advantages: First, it accounts for nonlinear motion characteristics such as non-uniform speed or sudden changes in direction, making it widely applicable to tracking targets with dramatic or erratic movements. Second, all required data is derived from spatial coordinate changes within the image frames, allowing parameters to be directly derived without introducing fuzzy factors, ensuring the determinism and stability of the computational model. By precalculating the target's potential position, the system provides "pre-aiming" capability, enabling adjustments to be made before the target moves rapidly or is obscured. This significantly reduces jitter or target deviation caused by tracking delay, effectively improving the smoothness and robustness of tracking.

[0024] The calculation of the steering angles that the main camera and the auxiliary camera need to adjust according to the three-dimensional coordinates of the target in the next frame includes: Adjust the main view camera at the same time and auxiliary viewpoint cameras Steering angle; Define the current moment The direction vector is , the predicted target direction vector is ; Construct the camera steering angle calculation formula as follows: in: is the three-dimensional coordinate of the camera; for and The angle between them is the steering angle; They are and The vector modulus of .

[0025] By introducing the three-dimensional direction vector angle as a control variable, the camera adjustment angle is uniformly incorporated into the spatial vector framework for processing. Specifically, the vector formed by the current camera's line of sight and the predicted target position is used as input to calculate its spatial angle, thereby obtaining the required steering amplitude. This method avoids the coordinate conversion error and axial interference that may be caused by the traditional angle decomposition method. At the same time, by modeling the main camera and the auxiliary camera separately, the dual-camera system does not produce coupling interference during the adjustment process. The calculation of the direction angle is based on the vector dot product relationship. All parameters in the calculation are derived from the actual measured spatial coordinates, with clear dimensions and traceability. This control strategy solves the steering error problem caused by different reference directions and positions of each camera in multi-camera collaborative control. It is particularly suitable for application scenarios where two or more cameras have different initial viewing directions and inconsistent installation angles. It significantly improves the consistency of system control and steering accuracy, laying a technical foundation for high-performance collaborative shooting.

[0026] The calculation of the angular velocity of the main camera and the auxiliary camera by the stabilization control function includes: Construct a stable control function and calculate the target angular velocity , as follows: in: Indicates the response cycle; The common frame rate of modern video capture systems is 30 frames per second, and the interval between each frame is about 30 = 0.033 seconds. Therefore, setting A value of 0.03 seconds ensures that the system performs a complete response calculation for each frame, achieving frame synchronization control. At the same time, this value is small enough to ensure fast response without overshoot, thus balancing stability and real-time performance. is the number of feedback increment cycles, which is 3 frames in this scheme; Directly using the current frame angular velocity change rate will cause allergic response or oscillation, while It means calculating the average angular velocity change rate between three frames, which is equivalent to short-term inertial filtering. This value can not only maintain fast response, but also eliminate the influence of abnormal jitter in a single frame. is the rate of change of angle.

[0027] By constructing a combined control model that uses angle deviation and angular velocity rate as inputs, dynamic adjustment of the angular velocity output is achieved. The control function uses a response period parameter as the basis for the adjustment rate, set to 0.03 seconds to match the typical 30-frame-per-second system refresh cycle, enabling frame-by-frame synchronous control. A feedback period parameter is also introduced to construct an average angular velocity rate over three frames. This is used to filter out transient misjudgments caused by abnormal jitter in a single frame, thereby maintaining the continuity and stability of the control system. This function ensures fast response while effectively suppressing overshoot or oscillation during rapid tracking, preventing camera overshoot or image wobbling caused by rapid angle adjustments. Compared to traditional proportional control methods, this control function uses parameters derived from system time and image acquisition rate. This is deterministic and eliminates the need for fuzzy parameter tuning, improving the algorithm's versatility and deployment efficiency, making it suitable for high-frame-rate, high-sensitivity automatic tracking scenarios.

[0028] The calculation of angular velocity compensation to optimize the angular velocity of the main camera and the auxiliary camera includes: Set the acceleration vector ; in, are the inertial disturbance acceleration of the target in the horizontal, vertical and depth directions respectively. The specific calculation process is as follows: Construct the compensation correction function as follows: in: is angular velocity compensation; The inertial parameters of the camera response; The device responds to inertial parameters It can be obtained through actual control response testing. It is the feedback coefficient of the motor's response angular velocity to the target acceleration change. This parameter is a fixed value, unique to the device, and can be calibrated at the factory. is the acceleration vector The specific expression is ; calculate The final angular velocity , specifically: .

[0029] The real-time monitoring of the rotation of the main camera and the auxiliary camera, and when image overlap occurs, executing an adjustment strategy, include: Get the current time Direction vector ; Define the current moment The direction vector is ; Calculate the field of view angle between the main camera and the auxiliary camera , as follows: Get the horizontal field of view angle of the main camera and auxiliary camera ; The horizontal field of view angle refers to the angle of the field of view that the camera can cover in the horizontal direction; like , it is determined that there is overlap between the field of view of the main camera and the auxiliary camera, and the adjustment strategy is executed; The adjustment strategy is as follows: Extract separately and The edge point set in , specifically: in, and Respectively and The maximum value of the corresponding responsivity of all pixels in ; Since only a small number of pixels in the image response map are located in the high-gradient edge area and the response values ​​are concentrated, the threshold is set to 20% of the maximum response value. This can effectively eliminate background interference and retain structural edge points. It also has scene adaptability and noise suppression capabilities, making it suitable as the response lower limit standard for edge point set extraction. Determine whether the images from the main camera and the auxiliary camera overlap: in, Represents the edge point set Coordinates within the intersection; Get the yaw angle of the current main camera and the yaw angle of the auxiliary camera ; The yaw angle of the main camera and the yaw angle of the auxiliary camera Make adjustments as follows: .

[0030] The yaw angle describes the angle at which the camera rotates around the depth direction and is used to represent the rotational posture of the lens in the horizontal direction.

[0031] By calculating the spatial angle between the current shooting directions of the main camera and the auxiliary camera, and comparing it with their respective horizontal field of view angles, when the angle is smaller than the field of view angle, it is determined that there is an overlapping area. At the same time, through the edge point extraction algorithm, high-response structure edges are extracted from the images of the two cameras, and a threshold is set based on 20% of the maximum response value to effectively exclude invalid background and weak feature areas, thereby enhancing the accuracy of overlap detection. When it is determined that an overlap occurs, the system performs distributed fine-tuning on the yaw angles of the two cameras to minimize the intersection of fields of view while maintaining target coverage, thereby improving the information coverage efficiency and resource allocation rationality of the entire system. This design is particularly suitable for multi-viewpoint intelligent shooting scenarios such as multi-camera drones and multi-axis rail vehicles. It solves the problems of image redundancy and information conflict caused by overlapping fields of view, and has practical engineering significance and innovative value. It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0032] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A fully automatic shooting control method based on a stable control algorithm, characterized in that: include: Construct a three-dimensional rectangular coordinate system, collect images and calculate the three-dimensional coordinates of the target; Predict the three-dimensional coordinates of the target in the next frame through the three-point trend prediction algorithm; Calculate the steering angles that the main camera and auxiliary camera need to adjust based on the 3D coordinates of the target in the next frame; Calculate the angular velocity of the main camera and auxiliary camera through the stable control function; Calculate angular velocity compensation and optimize the angular velocity of the main camera and auxiliary camera; Monitor the rotation of the main camera and auxiliary camera in real time, and execute adjustment strategies when overlapping images occur.

2. The fully automatic shooting control method based on a stable control algorithm according to claim 1, characterized in that: The process of constructing a three-dimensional rectangular coordinate system, acquiring images, and calculating the three-dimensional coordinates of the target includes: Get the main viewpoint camera and auxiliary viewpoint cameras ; pass and Real-time image acquisition; Get and The images in the tth frame are respectively recorded as ; Where t is the frame number; Establish a three-dimensional rectangular coordinate system, specifically: Take the optical center of the main camera as the coordinate origin ; Starting from the coordinate origin, draw a horizontal ray to the right on the main camera image as the horizontal axis; Starting from the coordinate origin, draw a vertical downward ray on the main camera image as the vertical axis; Starting from the coordinate origin, along the direction of the main camera pointing to the scene, draw a ray perpendicular to the image plane extending outward as the depth direction; Construct a feature point extraction algorithm as follows: in: is the main photographic response graph; is the auxiliary response diagram; and They are the pixel points in the main response map and the auxiliary response map respectively Gray value of and They are The first-order gradient of the inner pixel in the horizontal and vertical directions; and They are The first-order gradient of the inner pixel in the horizontal and vertical directions; Extract separately The maximum response value is taken as the target feature point coordinate ; Construct a standard binocular disparity formula and calculate the three-dimensional coordinates of the target based on the t-th frame image, as follows: in: F is the main view camera and auxiliary viewpoint cameras focal length; D is the main view camera and auxiliary viewpoint cameras the spacing between; They are and The center pixel coordinates of is the three-dimensional coordinate of the target when the t-th frame image is collected.

3. The fully automatic shooting control method based on a stable control algorithm according to claim 1, characterized in that: The three-point trend prediction algorithm is used to predict the three-dimensional coordinates of the target in the next frame, including: Construct a three-point trend prediction algorithm as follows: The lateral acceleration of the target at frame t : in: and are the speed of the target in the horizontal direction at the t-th frame image and the t-1-th frame image respectively; is the interval between image acquisitions; The longitudinal acceleration of the target at frame t : in: and are the speed of the target in the vertical direction at the t-th frame image and the t-1-th frame image respectively; The acceleration of the target in the depth direction at frame t : in: and are the speed of the target in the depth direction at the t-th frame image and the t-1-th frame image respectively; Predict the position of the target on the horizontal axis when the t+1 frame image is collected: Predict the target's position on the vertical axis when the t+1th frame image is collected: Predict the position of the target in the depth direction when the t+1th frame image is collected: in, Represents the three-dimensional coordinates of the target at the t+1th frame image.

4. The fully automatic shooting control method based on a stable control algorithm according to claim 1, characterized in that: The calculation of the steering angles that the main camera and the auxiliary camera need to adjust according to the three-dimensional coordinates of the target in the next frame includes: Adjust the main view camera at the same time and auxiliary viewpoint cameras Steering angle; Define the current moment The direction vector is , the predicted target direction vector is ; Construct the camera steering angle calculation formula as follows: in: is the three-dimensional coordinate of the camera; for and The angle between them is the steering angle; They are and The vector modulus of .

5. The fully automatic shooting control method based on a stable control algorithm according to claim 1, characterized in that: The calculation of the angular velocity of the main camera and the auxiliary camera by the stabilization control function includes: Construct a stable control function and calculate the target angular velocity , as follows: in: Indicates the response cycle; is the number of feedback increment cycles, which is 3 frames in this scheme; is the rate of change of angle.

6. The fully automatic shooting control method based on a stable control algorithm according to claim 1, characterized in that: The calculation of angular velocity compensation to optimize the angular velocity of the main camera and the auxiliary camera includes: Set the acceleration vector ; in, are the inertial disturbance acceleration of the target in the horizontal, vertical and depth directions respectively. The specific calculation process is as follows: Construct the compensation correction function as follows: in: is angular velocity compensation; The inertial parameters of the camera response; is the acceleration vector The specific expression is ; calculate The final angular velocity , specifically: .

7. The fully automatic shooting control method based on a stable control algorithm according to claim 1, characterized in that: The real-time monitoring of the rotation of the main camera and the auxiliary camera, and when image overlap occurs, executing an adjustment strategy, include: Get the current time Direction vector ; Define the current moment The direction vector is ; Calculate the field of view angle between the main camera and the auxiliary camera , as follows: Get the horizontal field of view angle of the main camera and auxiliary camera ; The horizontal field of view angle refers to the angle of the field of view that the camera can cover in the horizontal direction; like , it is determined that there is overlap between the field of view of the main camera and the auxiliary camera, and the adjustment strategy is executed; The adjustment strategy is as follows: Extract separately and The edge point set in , specifically: in, and Respectively and The maximum value of the corresponding responsivity of all pixels in ; Determine whether the images from the main camera and the auxiliary camera overlap: in, Represents the edge point set Coordinates within the intersection; Get the yaw angle of the current main camera and the yaw angle of the auxiliary camera ; The yaw angle of the main camera and the yaw angle of the auxiliary camera Make adjustments as follows: 。

Images