A method and system for real-time image stitching of unmanned aerial vehicles

Abstract

The present invention provides a method and system for real-time image stitching of a drone, which relates to the field of image stitching technology and includes the following steps: obtaining state parameters of at least two cameras on a drone; analyzing camera state parameters and environmental parameters; obtaining real-time images from the cameras on the drone; segmenting the real-time images frame by frame to obtain frame-by-frame images, and performing stitching processing on the frame-by-frame images to obtain the frame-by-frame images after stitching processing. The present invention provides a method and system for real-time image stitching of a drone, which rapidly processes images captured by the drone camera through frame-by-frame segmentation and stitching processing of real-time images, significantly improves the efficiency of image stitching, and optimizes the errors in the stitching process using a comprehensive energy function and a gradient descent method, thereby reducing stitching errors, improving stitching quality, and generating continuous, seamless, high-quality images.

Description

Technical Field

[0001] The present invention relates to the technical field of image stitching, and in particular to a method and system for stitching unmanned aerial vehicle (UAV) real-time images. Background Art

[0002] As the application of drones continues to expand, the demand for drone real-time image stitching technology is growing. When drones perform tasks, they usually need to obtain real-time images through the cameras they carry. However, the viewing angle of a single camera is limited and cannot provide a wide field of view. In order to obtain a wider field of view and richer information, it is necessary to stitch images captured by multiple cameras to form a continuous, seamless, high-quality image. However, when drones perform tasks such as monitoring, mapping, search and rescue, they need to quickly and accurately acquire and process large amounts of image data.

[0003] Traditional image stitching technology still has some limitations, such as low image stitching efficiency, low stitching quality, and inconsistent parameters such as color and exposure during image stitching, which affect the visual effect. Therefore, it is necessary to propose a real-time UAV image stitching method and system to solve the above problems. Summary of the invention

[0004] The present invention provides a real-time image stitching method and system for unmanned aerial vehicles, which solves the problems of low image stitching efficiency, low stitching quality, and inconsistency of parameters such as color and exposure during the image stitching process.

[0005] In order to solve the above technical problems, the present invention provides a method and system for real-time image stitching of unmanned aerial vehicles, comprising the following steps:

[0006] Step 1: Obtain status parameters of at least two cameras on the drone;

[0007] Step 2: Analyze the camera state parameters and environmental parameters, obtain the posture angle by analyzing the spatial posture of the camera, and determine the camera's viewing angle coverage to obtain the shooting overlapping viewing angle area;

[0008] Step 3: Obtain real-time images from the camera on the drone;

[0009] Step 4: segment the real-time image frame by frame to obtain frame-by-frame images, and perform splicing processing on the frame-by-frame images to obtain spliced ​​frame-by-frame images; the splicing processing includes image registration, image projection and image fusion;

[0010] The splicing process is specifically as follows:

[0011] Image fusion, analyzing the seam lines of each frame of the image to obtain image matching error terms, motion error terms and reconstruction error terms; processing the image matching error terms, motion error terms and reconstruction error terms to obtain a comprehensive energy function;

[0012] The comprehensive energy function in each frame of the image is processed using the gradient descent method to obtain a set of path variables; the satisfactory minimum threshold and the expected number of iterations of the comprehensive energy function are set, and after each iteration, the path variables are updated until the comprehensive energy function is less than its satisfactory minimum threshold; or after the expected number of iterations is reached, the variance of all path variables in the path variable set is calculated to obtain the variable fluctuation value; if the variable fluctuation value is less than the set variable threshold, the iteration ends; if the variable fluctuation value is greater than or equal to the set variable threshold, the result graphs before and after the improvement are analyzed and compared to obtain the evaluation values ​​of the result graphs before and after the improvement; if the evaluation value of the result graph before the improvement is greater than the evaluation value of the result graph after the improvement, the iteration is repeated; otherwise, the improved result graph is directly used;

[0013] Step 5: Combine the stitched frame-by-frame images according to the frame number to generate a high-quality image.

[0014] As a preferred embodiment of the present invention, based on step four, before image registration, unified processing is performed on each frame of the image, and the specific method is: obtain image information of each frame of the image, the image information includes resolution, color information, exposure and illumination information, the color information includes information of three RGB color channels, and the exposure and illumination information includes light intensity and color temperature; perform mean calculation on any parameter in the image information of the same frame number to obtain the mean value YG1 corresponding to the parameter; perform variance calculation on any parameter in the image information within the frame number range set by the current frame number to obtain the parameter fluctuation value YG2; set multiple groups of frame number ranges, perform variance calculation on the fluctuation values ​​of multiple groups of frame number ranges, and obtain the range fluctuation value YG4;

[0015] Set the conventional parameter of any parameter in the image information, perform difference calculation between any parameter in the image information and its conventional parameter, and obtain the difference value YG3;

[0016] The mean, fluctuation value, difference value and range fluctuation value corresponding to the parameter are processed, and the adjustment function YG corresponding to the parameter is obtained using the formula YG=[(YG1-YG3)*g1] / (YG2*g2+YG4*g3+1); wherein g1 represents the weight of the difference between the mean and difference corresponding to the parameter, and g2 and g3 represent the weights corresponding to the fluctuation value and range fluctuation value of the parameter respectively; the adjustment function is applied to perform image adjustment on all frame-by-frame images.

[0017] As a preferred embodiment of the present invention, the splicing process includes:

[0018] Image registration, using scale-invariant feature transformation to detect feature points in each frame of the image; using the best binary tree matching to match feature points to obtain feature point pairs; using random sampling consensus algorithm to remove mismatched feature point pairs;

[0019] Image projection, converting frame-by-frame images from different cameras into images with the same coordinate system and scale.

[0020] As a preferred embodiment of the present invention, overlapping images A and B of any two cameras A1 and B1 in the overlapping viewing angle area are obtained, and the pixel values ​​of any same position in the overlapping viewing angle area are represented as I A (m,n) and I B (m,n);

[0021] Match the error term between the two overlapping viewing angles, and the formula is expressed as: The image matching error term ER is obtained; wherein N represents the total number of pixel values ​​at all positions in the overlapping viewing angle area;

[0022] Get the predicted motion vector as The actual motion vector is Measures the difference between the predicted motion vector and the actual motion vector, expressed as: Get the motion compensation error term EP;

[0023] To evaluate the smoothness of the reconstructed image, the function of the reconstructed image is set to U(q,p), which is expressed as The second-order derivatives are calculated as

[0024] The sum of the second-order derivatives calculated by the Laplace operator is used for approximation, and the formula is expressed as: Among them, U(q, p) represents the function of reconstructing the image, (q, p) represents the coordinate position in the image, represents the second-order partial derivative of the U function with respect to the coordinate q, Represents the second-order partial derivative of the U function with respect to the coordinate p; the average value of the sum of squares of the second-order derivatives is calculated as the measure of smoothness S, and the formula is expressed as: Among them, (qi,pi) represents the coordinate position of the i-th pixel in the image;

[0025] The difference between the smoothness S of the reconstructed image and the smoothness S0 of the original image is calculated, and the reconstruction error value SE1 is obtained using the formula SE1=|S-S0|;

[0026] The mean square error of the smoothness S of the reconstructed image and the smoothness S0 of the original image is calculated using the formula The mean square reconstruction error value SE2 is obtained; wherein Si and S0i are the smoothness measures of the i-th pixel in the image in the reconstructed image and the original image respectively;

[0027] The reconstruction error value and the mean square reconstruction error value are weighted, and the reconstruction error term SE is obtained using the formula SE=SE1×e1+SE2×e2; wherein e1 and e2 represent the weights corresponding to the reconstruction error value and the mean square reconstruction error value, respectively.

[0028] As a preferred implementation of the present invention, after the high-quality image is generated, the high-quality image is transmitted to a terminal or a remote control terminal corresponding to the drone via a wired or wireless network.

[0029] As a preferred embodiment of the present invention, the present invention also includes a storage unit, a registration unit, and a display unit; the registration unit is used to submit the registration information of the control end for registration, and mark the successfully registered control end as a remote control end; the storage unit is used to receive and store high-quality images; the display unit is used to access and display the high-quality image data in the storage unit.

[0030] As a preferred embodiment of the present invention, the present invention further comprises a data collection unit; the data collection unit is used to coordinate the acquisition of the state parameters, the environmental parameters and the real-time image of the camera.

[0031] The present invention comprises a data collection unit, a state pre-analysis unit, a splicing processing unit and an integration unit;

[0032] A data collection unit, used to obtain status parameters of at least two cameras on the drone and to obtain real-time images from the cameras on the drone;

[0033] The state pre-analysis unit is used to analyze the state parameters and environmental parameters of the camera, obtain the posture angle by analyzing the spatial posture of the camera, and determine the viewing angle coverage of the camera to obtain the shooting overlapping viewing angle area;

[0034] A splicing processing unit is used to divide the real-time image frame by frame to obtain frame-by-frame images, and to splice the frame-by-frame images to obtain spliced ​​frame-by-frame images; the splicing processing includes image registration, image projection and image fusion;

[0035] The integration unit is used to combine the spliced ​​frame-by-frame images according to the frame number to generate high-quality images.

[0036] Compared with the related art, the real-time image stitching method and system of UAV provided by the present invention has the following beneficial effects:

[0037] 1. The present invention can quickly process the images captured by the drone camera through frame-by-frame segmentation and splicing of real-time images, significantly improving the efficiency of image splicing. It also uses a comprehensive energy function and gradient descent method to optimize the error in the splicing process, thereby reducing the splicing error, improving the splicing quality, and generating continuous, seamless, high-quality images.

[0038] 2. The present invention can continuously adjust and optimize the path variables in the splicing process through the gradient descent method until a lower comprehensive energy function value is reached, indicating the minimization of the splicing error, thereby improving the splicing quality. At the same time, by setting the satisfactory minimum threshold and the expected number of iterations of the comprehensive energy function, as well as the variance calculation and the setting of the variable fluctuation value, the system can determine when to stop the iteration, avoiding endless iterations, and can judge whether the current iteration is stable enough to decide whether to end the iteration, thereby increasing the stability and reliability of the algorithm, reducing the calculation amount and resource consumption of the algorithm, and making the entire splicing process more efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A flowchart of a real-time image stitching method for an unmanned aerial vehicle provided by the present invention.

[0040] Figure 2 This is a principle block diagram of a UAV real-time image stitching system provided by the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0042] The terms used in this disclosure are for the purpose of describing specific embodiments only and are not intended to limit the disclosure. The singular forms of "class", "group" and "the" used in this disclosure and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0043] It should be understood that although the terms first, second, third, etc. may be used in the present disclosure to describe various information, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present disclosure, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0044] Please refer to Figure 1-2 A method and system for real-time image stitching of unmanned aerial vehicles, comprising the following steps:

[0045] Step 1: Obtain status parameters of at least two cameras on the drone;

[0046] Step 2: Analyze the camera state parameters and environmental parameters, obtain the posture angle by analyzing the spatial posture of the camera, and determine the camera's viewing angle coverage to obtain the shooting overlapping viewing angle area;

[0047] Step 3: Obtain real-time images from the camera on the drone;

[0048] Step 4: segment the real-time image frame by frame to obtain frame-by-frame images, and perform splicing processing on the frame-by-frame images to obtain spliced ​​frame-by-frame images; the splicing processing includes image registration, image projection and image fusion;

[0049] The splicing process is specifically as follows:

[0050] Image fusion, analyze the seam lines of each frame of the image, obtain the overlapping images A and B of any two cameras A1 and B1 in the overlapping viewing area, and represent the pixel values ​​of any identical position in the overlapping viewing area as I A (m,n) and I B (m,n);

[0051] Match the error term between the two overlapping viewing angles, and the formula is expressed as: The image matching error term ER is obtained; wherein N represents the total number of pixel values ​​at all positions in the overlapping viewing angle area;

[0052] Get the predicted motion vector as The actual motion vector is Measures the difference between the predicted motion vector and the actual motion vector, expressed as: Get the motion compensation error term EP;

[0053] To evaluate the smoothness of the reconstructed image, the function of the reconstructed image is set to U(q,p), which is expressed as The second-order derivatives are calculated as

[0054] The sum of the second-order derivatives calculated by the Laplace operator is used for approximation, and the formula is expressed as: Among them, U(q, p) represents the function of reconstructing the image, (q, p) represents the coordinate position in the image, represents the second-order partial derivative of the U function with respect to the coordinate q, Represents the second-order partial derivative of the U function with respect to the coordinate p; the average value of the sum of squares of the second-order derivatives is calculated as the measure of smoothness S, and the formula is expressed as: Among them, (qi,pi) represents the coordinate position of the i-th pixel in the image;

[0055] The difference between the smoothness S of the reconstructed image and the smoothness S0 of the original image is calculated, and the reconstruction error value SE1 is obtained using the formula SE1=|S-S0|;

[0056] The mean square error of the smoothness S of the reconstructed image and the smoothness S0 of the original image is calculated using the formula The mean square reconstruction error value SE2 is obtained; wherein Si and S0i are the smoothness measures of the i-th pixel in the image in the reconstructed image and the original image respectively;

[0057] The reconstruction error value and the mean square reconstruction error value are weighted, and the reconstruction error term SE is obtained using the formula SE=SE1×e1+SE2×e2; wherein e1 and e2 represent the weights corresponding to the reconstruction error value and the mean square reconstruction error value, respectively;

[0058] The image matching error term, motion error term and reconstruction error term are processed, and the comprehensive energy function ED is obtained using the formula ED=ER×s1+EP×s2+ET×s3; where s1, s2, s3 represent the weights corresponding to the image matching error term, motion compensation error term and reconstruction error term, respectively, and ED represents the comprehensive energy function of any position in the frame-by-frame image;

[0059] The gradient descent method is used to process the comprehensive energy function in each frame of the image, using the formula Get the path variable set; where Pt represents the path variable set in the tth iteration; Pt+1 represents the path variable set in the t+1th iteration, α is the set learning rate, represents the gradient of the integrated energy function ED at Pt;

[0060] Set the satisfactory minimum threshold of the comprehensive energy function and the expected number of iterations r. After each iteration, update the path variable P to Pt+1 until the comprehensive energy function is less than its satisfactory minimum threshold; or after reaching the expected number of iterations, perform variance calculation on all path variables in the path variable set to obtain the variable fluctuation value; if the variable fluctuation value is less than its set variable threshold le-r, the iteration ends; if the variable fluctuation value is greater than or equal to its set variable threshold le-r, analyze and compare the result graphs before and after the improvement to obtain the evaluation values ​​of the result graphs before and after the improvement; if the evaluation value of the result graph before the improvement is greater than the evaluation value of the result graph after the improvement, re-iterate; otherwise, directly use the improved result graph;

[0061] Step 5: Combine the stitched frame-by-frame images according to the frame number to generate a high-quality image.

[0062] In the present application, based on step 4, before image registration, unified processing is performed on each frame of the image, and the specific method is as follows: image information of each frame of the image is obtained, the image information includes resolution, color information, and exposure and illumination information, the color information includes information of three color channels of RGB, and the exposure and illumination information includes the intensity and color temperature of the light; mean calculation is performed on any parameter in the image information of the same frame number to obtain the mean value YG1 corresponding to the parameter; variance calculation is performed on any parameter in the image information within the frame number range set by the current frame number to obtain the parameter fluctuation value YG2; multiple groups of frame number ranges are set, and variance calculation is performed on the fluctuation values ​​of multiple groups of frame number ranges to obtain the range fluctuation value YG4;

[0063] Set the conventional parameters of any parameter in the image information, perform difference calculation between any parameter in the image information and its conventional parameters, and obtain the difference value YG3;

[0064] The mean, fluctuation value, difference value and range fluctuation value corresponding to the parameter are processed, and the adjustment function YG corresponding to the parameter is obtained using the formula YG=[(YG1-YG3)*g1] / (YG2*g2+YG4*g3+1); wherein g1 represents the weight of the difference between the mean and difference corresponding to the parameter, and g2 and g3 represent the weights corresponding to the fluctuation value and range fluctuation value of the parameter respectively; the adjustment function is applied to perform image adjustment on all frame-by-frame images.

[0065] It should be noted that the present invention ensures the consistency of the stitched image in terms of color and exposure by performing unified processing on each frame of the image before image registration, including mean calculation and variance analysis of resolution, color information, and exposure and illumination information.

[0066] In this application, the splicing process includes:

[0067] Image registration, using scale-invariant feature transformation to detect feature points in each frame of the image; using the best binary tree matching to match feature points to obtain feature point pairs; using random sampling consensus algorithm to remove mismatched feature point pairs;

[0068] Image projection, converting frame-by-frame images from different cameras into images with the same coordinate system and scale.

[0069] In the present application, after the high-quality image is generated, the high-quality image is transmitted to the terminal or remote control terminal corresponding to the drone via a wired or wireless network.

[0070] It should be noted that the ability to transmit high-quality images through wired or wireless networks allows multiple users or teams to access image data simultaneously, facilitating collaboration and information sharing.

[0071] In the present application, the present invention also includes a storage unit, a registration unit, and a display unit; the registration unit is used to submit the registration information of the control end for registration, and mark the successfully registered control end as a remote control end; the storage unit is used to receive and store high-quality images; the display unit is used to access and display the high-quality image data in the storage unit.

[0072] It should be noted that the present invention realizes efficient management and use of drone real-time image stitching through the cooperation of the storage unit, the registration unit and the display unit.

[0073] In the present application, the present invention also includes a data collection unit; the data collection unit is used to coordinate the acquisition of camera status parameters, environmental parameters and real-time images.

[0074] The present invention comprises a data collection unit, a state pre-analysis unit, a splicing processing unit and an integration unit;

[0075] A data collection unit, used to obtain status parameters of at least two cameras on the drone and to obtain real-time images from the cameras on the drone;

[0076] The state pre-analysis unit is used to analyze the state parameters and environmental parameters of the camera, obtain the posture angle by analyzing the spatial posture of the camera, and determine the viewing angle coverage of the camera to obtain the shooting overlapping viewing angle area;

[0077] A splicing processing unit is used to divide the real-time image frame by frame to obtain frame-by-frame images, and to splice the frame-by-frame images to obtain spliced ​​frame-by-frame images; the splicing processing includes image registration, image projection and image fusion;

[0078] The integration unit is used to combine the spliced ​​frame-by-frame images according to the frame number to generate high-quality images.

[0079] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed in this disclosure. The specification and examples are to be considered exemplary only, and the true scope and spirit of the present invention are indicated by the following claims.

[0080] It should be understood that the present invention is not limited to the exact construction that has been described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A method for real-time image stitching of unmanned aerial vehicles, characterized in that: The following steps are involved: Step 1: Obtain status parameters of at least two cameras on the drone; Step 2: Analyze the camera state parameters and environmental parameters, obtain the posture angle by analyzing the spatial posture of the camera, and determine the camera's viewing angle coverage to obtain the shooting overlapping viewing angle area; Step 3: Obtain real-time images from the camera on the drone; Step 4: segment the real-time image frame by frame to obtain frame-by-frame images, and perform splicing processing on the frame-by-frame images to obtain spliced ​​frame-by-frame images; the splicing processing includes image registration, image projection and image fusion; The splicing process is specifically as follows: Image fusion, analyze the seam lines of each frame of the image, and obtain the image matching error term, motion error term and reconstruction error term, specifically: Get the overlapping images A and B of any two cameras A1 and B1 in the overlapping viewing area, and express the pixel values ​​of any identical position in the overlapping viewing area as and ; Match the error term between the two overlapping viewing angles, and the formula is expressed as: , the image matching error term ER is obtained; where N represents the total number of pixel values ​​at all positions in the overlapping viewing angle area; Get the predicted motion vector as , the actual motion vector is , which measures the difference between the predicted motion vector and the actual motion vector, and the formula is expressed as: , get the motion compensation error term EP; To evaluate the smoothness of the reconstructed image, the function of the reconstructed image is set to U(q,p), which is expressed as , the second-order derivatives are calculated as , ; The sum of the second-order derivatives calculated by the Laplace operator is used for approximation, and the formula is expressed as: ; Where U(q, p) represents the function of reconstructing the image, (q, p) represents the coordinate position in the image, represents the second-order partial derivative of the U function with respect to the coordinate q, Represents the second-order partial derivative of the U function with respect to the coordinate p; the average value of the sum of squares of the second-order derivatives is calculated as the measure of smoothness S, and the formula is expressed as: ;in, Represents the coordinate position of the i-th pixel in the image; The difference between the smoothness S of the reconstructed image and the smoothness S0 of the original image is calculated, and the reconstruction error value SE1 is obtained using the formula SE1=|S-S0|; The mean square error of the smoothness S of the reconstructed image and the smoothness S0 of the original image is calculated using the formula , and obtain the mean square reconstruction error value SE2; where Si and S0i are the smoothness measures of the i-th pixel in the image in the reconstructed image and the original image respectively; The reconstruction error value and the mean square reconstruction error value are weighted, and the reconstruction error term SE is obtained using the formula SE=SE1×e1+SE2×e2; where e1 and e2 represent the weights corresponding to the reconstruction error value and the mean square reconstruction error value, respectively; The comprehensive energy function in each frame of the image is processed using the gradient descent method to obtain a set of path variables; the satisfactory minimum threshold and the expected number of iterations of the comprehensive energy function are set, and after each iteration, the path variables are updated until the comprehensive energy function is less than its satisfactory minimum threshold; or after the expected number of iterations is reached, the variance of all path variables in the path variable set is calculated to obtain the variable fluctuation value; if the variable fluctuation value is less than the set variable threshold, the iteration ends; if the variable fluctuation value is greater than or equal to the set variable threshold, the result graphs before and after the improvement are analyzed and compared to obtain the evaluation values ​​of the result graphs before and after the improvement; if the evaluation value of the result graph before the improvement is greater than the evaluation value of the result graph after the improvement, the iteration is repeated; otherwise, the improved result graph is directly used; Step 5: Combine the stitched frame-by-frame images according to the frame number to generate a high-quality image.

2. The method for real-time image stitching of a drone according to claim 1, characterized in that: Based on step 4, before image registration, the frame-by-frame images are uniformly processed, and the specific method is as follows: image information of the frame-by-frame images is obtained, the image information includes resolution, color information, and exposure and illumination information, the color information includes information of the three RGB color channels, and the exposure and illumination information includes light intensity and color temperature; Perform mean calculation on any parameter in the image information of the same frame number to obtain the mean value corresponding to the parameter; perform variance calculation on any parameter in the image information within the frame number range set by the current frame number to obtain the fluctuation value of the parameter; set multiple sets of frame number ranges, perform variance calculation on the fluctuation values ​​of multiple sets of frame number ranges to obtain the range fluctuation value; Setting a conventional parameter of any parameter in the image information, performing a difference calculation between any parameter in the image information and its conventional parameter to obtain a difference value; The mean value, fluctuation value, difference value and range fluctuation value corresponding to the parameter are processed to obtain the adjustment function corresponding to the parameter; and the adjustment function is applied to perform image adjustment on all frame-by-frame images.

3. The method and system for real-time image stitching of unmanned aerial vehicles according to claim 1, characterized in that: in, The stitching process includes: Image registration, using scale-invariant feature transformation to detect feature points in each frame of the image; using the best binary tree matching to match feature points to obtain feature point pairs; using random sampling consensus algorithm to remove mismatched feature point pairs; Image projection, converting frame-by-frame images from different cameras into images with the same coordinate system and scale.

4. The method and system for real-time image stitching of unmanned aerial vehicles according to claim 1, characterized in that: After high-quality images are generated, they are transmitted to the corresponding terminal or remote control terminal of the drone via a wired or wireless network.

5. The method for real-time image stitching of unmanned aerial vehicles according to claim 4, characterized in that: It also includes a storage unit, a registration unit, and a display unit; the registration unit is used to submit the registration information of the control terminal for registration, and mark the successfully registered control terminal as a remote control terminal; the storage unit is used to receive and store high-quality images; the display unit is used to access and display the high-quality image data in the storage unit.

6. The method for real-time image stitching of unmanned aerial vehicles according to claim 1, characterized in that: It also includes a data collection unit; the data collection unit is used to coordinate the acquisition of camera status parameters, environmental parameters and real-time images.

7. A drone real-time image stitching system, using a drone real-time image stitching method according to any one of claims 1 to 6, characterized in that: It includes a data collection unit, a state pre-analysis unit, a splicing processing unit and an integration unit; A data collection unit, used to obtain status parameters of at least two cameras on the drone and to obtain real-time images from the cameras on the drone; The state pre-analysis unit is used to analyze the state parameters and environmental parameters of the camera, obtain the posture angle by analyzing the spatial posture of the camera, and determine the viewing angle coverage of the camera to obtain the shooting overlapping viewing angle area; A splicing processing unit is used to divide the real-time image frame by frame to obtain frame-by-frame images, and to splice the frame-by-frame images to obtain spliced ​​frame-by-frame images; the splicing processing includes image registration, image projection and image fusion; The integration unit is used to combine the spliced ​​frame-by-frame images according to the frame number to generate high-quality images.

Images