Dynamic modeling viewpoint optimization system based on unmanned aerial vehicle photography

Abstract

The invention discloses a dynamic modeling viewpoint optimization system based on unmanned aerial vehicle photography, relates to the technical field of photography management, and improves the accuracy of a photography result. According to the invention, the unmanned aerial vehicle collects the first scene photography data of the target scene area, and establishes a plurality of dynamic unmanned aerial vehicle flight paths according to the fuzzy scene grid from the scene modeling end, so that each unmanned aerial vehicle collects the second scene photography data of the target scene area along the dynamic unmanned aerial vehicle flight paths. Meanwhile, in the flight process of the unmanned aerial vehicle, according to the position change of the attraction grids in the fuzzy scene grid, dynamically updating the flight path of the dynamic unmanned aerial vehicle, establishing the fuzzy scene grid according to the first scene photography data, and then marking a plurality of attraction grids on the fuzzy scene grid according to the modeling viewpoint information; and meanwhile, establishing a dynamic scene visualization model according to the second scene photography data, and updating the position of the attraction grid on the fuzzy scene grid based on the dynamic scene visualization model.

Description

Technical Field

[0001] The present invention relates to the technical field of photography management, and in particular to a dynamic modeling viewpoint optimization system based on unmanned aerial vehicle photography. Background Art

[0002] Traditional target scene modeling and data collection methods often rely on ground measurement equipment or manual field surveys. These methods are not only inefficient and costly, but also difficult to obtain comprehensive and accurate data when faced with complex terrain, large areas or dangerous environments.

[0003] With the development of drone technology, using drones to collect scene data has gradually become a trend. However, existing drone data collection methods have many shortcomings. On the one hand, the flight path of drones is usually pre-set, and it is difficult to dynamically adjust according to the actual situation of the scene, which may lead to omissions or duplications in the collected data and fail to fully and accurately reflect the characteristics of the target scene. On the other hand, the collected photographic data lacks an effective verification and fusion mechanism, and the data collected at different stages are difficult to effectively integrate, which affects the accuracy and reliability of the final scene modeling. For this reason, a dynamic modeling viewpoint optimization system based on drone photography is provided. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a dynamic modeling viewpoint optimization system based on drone photography.

[0005] In order to achieve the above object, the present invention provides the following technical solutions: A dynamic modeling viewpoint optimization system based on drone photography, comprising a cloud computing platform, wherein the cloud computing platform is communicatively connected to a drone management terminal and a scene modeling terminal; The drone management end is used to collect first scene photography data of the target scene area through the drone, and establish multiple dynamic drone flight paths according to the fuzzy scene grid from the scene modeling end; Then, each drone collects second scene photography data of the target scene area along the dynamic drone flight path, and dynamically updates the dynamic drone flight path according to the change of the attraction grid position in the fuzzy scene grid during the drone flight, and performs difference detection on the first scene photography data and the second scene photography data; The scene modeling end is used to obtain modeling viewpoint information, and establish a fuzzy scene grid according to the first scene photography data, and then mark a number of attraction grids on the fuzzy scene grid according to the modeling viewpoint information, and at the same time establish a dynamic scene visualization model according to the second scene photography data, and update the position of the attraction grid on the fuzzy scene grid based on the dynamic scene visualization model.

[0006] Furthermore, the process of collecting the first scene photography data includes: Set n drones, each of which has a camera range of radius r, and set n one-way drone flight routes according to the scene acquisition range, and the interval distance between each one-way drone flight route is less than r, where n is a natural number greater than 10; Each drone is made to fly over the target scene area along a one-way drone flight route at the same speed, and during the flight, each drone collects first scene photography data within its shooting range; After each drone finishes flying, the first scene photography data collected by each drone is collected.

[0007] Furthermore, the process of performing the first scene photography data includes: Establish a two-dimensional coordinate system, map the first scene photography data collected by each UAV on the two-dimensional coordinate system according to its flight route, perform grayscale processing on all the first scene photography data, and set a pixel difference threshold; Determine the magnitude relationship between the pixel difference between the grayscale pixels in the overlapping state in the two-dimensional coordinate system and the pixel difference threshold; If the pixel difference is less than or equal to the pixel difference threshold, no operation is performed; If the pixel difference is greater than the pixel difference threshold, it is determined whether the adjacent grayscale values ​​of the corresponding grayscale pixels have the same size relationship at the same time. If so, the overlapping positions of the two grayscale pixels are spliced, otherwise the pixel values ​​of the two grayscale pixels are directly accumulated and averaged; When all the grayscale pixels in the overlapping state are judged, any layer of grayscale pixels in the overlapping position of the grayscale pixels in the splicing state are displaced and translated in four directions in the plane direction by a distance of a grayscale pixel; The relationship between the pixel difference between the grayscale pixels in the overlapping state in the grayscale pixel overlapping position after each translation and the pixel difference threshold is judged in turn, the number of grayscale pixels in each direction whose pixel difference is less than or equal to the pixel difference threshold is counted, and then the direction with the largest number of grayscale pixels and more than half of the total number of grayscale pixels is selected as the correction direction, and the grayscale pixel overlapping position in the splicing state is displaced, and then the grayscale pixels in which the pixel difference is less than or equal to the pixel difference threshold are subjected to pixel averaging processing; If there is no direction in which the number of grayscale pixels is the largest and exceeds half of the total number of grayscale pixels, pixel averaging processing is performed on the grayscale pixels in the overlapping state at the overlapping positions of the original grayscale pixels.

[0008] Furthermore, the process of establishing the fuzzy scene grid includes: Performing image stitching on the first scene photography data, and establishing a fuzzy scene grid according to the image stitching result, wherein the fuzzy scene grid is composed of a plurality of two-dimensional image grids; According to the modeling viewpoint information uploaded by the user, multiple viewpoint feature models are marked on the fuzzy scene grid. At the same time, according to the modeling viewpoint information priority set by the user, the corresponding photography priority is set for each viewpoint feature model.

[0009] Furthermore, the process of establishing the dynamic UAV flight path includes: According to the position distribution of the viewpoint feature model in the fuzzy scene grid, a plurality of attraction grids are set in the fuzzy scene grid, and according to the photographic priority of each viewpoint feature model, an attraction degree is set for each attraction grid; n flight starting point grids with different spatial positions are set in the fuzzy scene grid, and then starting from each flight starting point grid at the same time, an attraction value β between each flight starting point grid and each attraction grid is obtained; According to the direction of the attraction grid with the largest attraction value β, a two-dimensional image grid adjacent to the flight starting point grid is selected as the next flight grid, and it is determined whether the flight grids associated with each flight starting point grid overlap. If so, the maximum attraction values ​​β associated with the two flight starting point grids are compared, and then the flight starting point grid with the smallest attraction value β selects the direction of the attraction grid with the second largest attraction value β, reselects a flight grid, and repeats the process of whether the flight grids overlap again until the flight grids associated with each flight starting point grid do not overlap; Then, the attraction value β between each attraction grid is re-obtained according to each flight starting point grid, and the above process of setting the flight grid and determining whether there is a flight grid overlap is repeated until each attraction grid is selected as a flight grid. The corresponding dynamic UAV flight path is generated according to the flight grid associated with each flight starting point grid.

[0010] Furthermore, according to the photography priority from high to low, the attraction degree of the corresponding attraction grid decreases successively.

[0011] Furthermore, a data uploading cycle is set for each drone, and then when each drone flies along the dynamic drone flight path, the second scene photography data within the shooting range of the drone is collected.

[0012] Furthermore, the process of establishing the dynamic scene visualization model includes: fusing each second scene photography data to obtain a scene visualization image of a corresponding data upload period, establishing or updating a dynamic scene visualization model according to the scene visualization image, and marking a plurality of scene tracking grids in the dynamic scene visualization model through a viewpoint feature model; Whenever a data upload cycle starts, the drone management end updates the distribution of each attraction grid in the fuzzy scene grid according to the scene tracking grid in the dynamic scene visualization model, and then replans the dynamic drone flight path of each drone and sends it to each drone synchronously.

[0013] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention establishes a fuzzy scene grid based on the first scene photography data and marks the attraction grid, establishes a dynamic scene visualization model based on the second scene photography data, and updates the attraction grid position. This realizes the organic combination of data collected at different stages, and dynamically and accurately models the target scene, more accurately reflecting the actual changes in the scene.

[0014] 2. The present invention establishes multiple dynamic UAV flight paths based on the fuzzy scene grid, and dynamically updates the paths according to the changes in the attraction grid positions, so that the UAV can comprehensively and accurately collect photographic data of the target scene area while adapting to the actual situation of the target scene, thereby effectively avoiding the data omission or duplication problems that occur in the traditional fixed path collection method. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a system structure diagram of the present invention. DETAILED DESCRIPTION

[0016] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation mode, structure, characteristics and effects of the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments.

[0017] like Figure 1 As shown, a dynamic modeling viewpoint optimization system based on drone photography includes a cloud computing platform, wherein the cloud computing platform is communicatively connected to a drone management terminal and a scene modeling terminal; The drone management terminal is provided with a drone flight management module and a photography data verification module; The UAV flight management module is used to collect first scene photography data of the target scene area through the UAV, and establish multiple dynamic UAV flight paths according to the fuzzy scene grid from the scene modeling end, and then collect second scene photography data of the target scene area according to the dynamic UAV flight path, and dynamically update the dynamic UAV flight path according to the change of the attraction grid position in the fuzzy scene grid during the UAV flight; The photographic data verification module is used to detect differences between the first scene photographic data and the second scene photographic data; The scene modeling end is provided with a photographic data fusion module and a scene modeling module; The photographic data fusion module is used to obtain modeling viewpoint information, and to establish a fuzzy scene grid according to the first scene photographic data, and then to mark a number of attraction grids on the fuzzy scene grid according to the modeling viewpoint information; The scene modeling module is used to establish a dynamic scene visualization model according to the second scene photography data, and update the position of the attraction grid on the fuzzy scene grid based on the dynamic scene visualization model.

[0018] Further, the working principle of the present invention is described below by way of examples: The user uploads the modeling viewpoint information and target scene information to the photography data fusion module through the scene modeling terminal. The modeling viewpoint information includes clouds, people, vehicles, plants, etc., and the target scene information includes the scene coordinate position and scene acquisition range; It should be noted that after the user uploads the modeling viewpoint information, the priority of the modeling viewpoint information can be set in the photographic data fusion module, that is, when collecting image data of the target scene, the target with a high level of photographic focus is collected first and updated data analysis resources are allocated according to the level of photographic focus set according to the priority of the modeling viewpoint information; When the user has completed uploading the modeling viewpoint information and target scene information, the target scene information is sent to the drone management terminal; The drone flight management module in the drone management terminal locates the target scene area according to the scene coordinate position of the target scene information and the scene acquisition range; Set n drones, each of which has a camera range of radius r, and set n one-way drone flight routes according to the scene acquisition range, and the interval distance between each one-way drone flight route is less than r, where n is a natural number greater than 10; Each drone is made to fly over the target scene area along a one-way drone flight route at the same speed, and during the flight, each drone collects first scene photography data within its shooting range; When each drone finishes flying, the drone flight management module collects the first scene photography data collected by each drone and sends it to the photography data verification module; Since the interval between each unidirectional UAV flight route is less than r, the first scene photography data collected by the UAVs at adjacent spatial positions have the same part; Then, the photographic data verification module establishes a two-dimensional coordinate system, maps the first scene photographic data collected by each drone on the two-dimensional coordinate system according to its flight route, performs grayscale processing on all the first scene photographic data, and sets a pixel difference threshold; Determine the magnitude relationship between the pixel difference between the grayscale pixels in the overlapping state in the two-dimensional coordinate system and the pixel difference threshold; If the pixel difference is less than or equal to the pixel difference threshold, no operation is performed; If the pixel difference is greater than the pixel difference threshold, it is determined whether the adjacent grayscale values ​​of the corresponding grayscale pixels have the same size relationship at the same time. If so, the overlapping positions of the two grayscale pixels are spliced, otherwise the pixel values ​​of the two grayscale pixels are directly accumulated and averaged; When all the grayscale pixels in the overlapping state are judged, any layer of grayscale pixels in the overlapping position of the grayscale pixels in the splicing state are displaced and translated in four directions in the plane direction by a distance of a grayscale pixel; The relationship between the pixel difference between the grayscale pixels in the overlapping state in the grayscale pixel overlapping position after each translation and the pixel difference threshold is judged in turn, the number of grayscale pixels in each direction whose pixel difference is less than or equal to the pixel difference threshold is counted, and then the direction with the largest number of grayscale pixels and more than half of the total number of grayscale pixels is selected as the correction direction, and the grayscale pixel overlapping position in the splicing state is displaced, and then the grayscale pixels in which the pixel difference is less than or equal to the pixel difference threshold are subjected to pixel averaging processing; If there is no direction in which the number of grayscale pixels is the largest and exceeds half of the total number of grayscale pixels, pixel averaging processing is performed on the grayscale pixels in the overlapping state at the overlapping positions of the original grayscale pixels.

[0019] Further, after all the first scene photography data have completed the difference detection, the drone management terminal sends all the first scene photography data to the photography data fusion module; The photographic data fusion module first performs image stitching on the first scene photographic data, and establishes a fuzzy scene grid according to the image stitching result, wherein the fuzzy scene grid is composed of a plurality of two-dimensional image grids; According to the modeling viewpoint information uploaded by the user, multiple viewpoint feature models are marked on the fuzzy scene grid. At the same time, according to the modeling viewpoint information priority set by the user, the corresponding photography priority is set for each viewpoint feature model, and the fuzzy scene grid is sent to the drone management end.

[0020] Furthermore, the drone flight management module in the drone management terminal sets a plurality of attraction grids in the fuzzy scene grid according to the position distribution of the viewpoint feature model in the fuzzy scene grid, and sets the attraction degree for each attraction grid according to the photography priority of each viewpoint feature model; It should be noted that, according to the photography priority from high to low, the attraction degree of the corresponding attraction grid decreases in sequence. For example, the attraction degree of the attraction grid with the highest photography priority is 100, and the attraction degree of the attraction grid with the lowest photography priority is 10; Set n flight starting grids with different spatial positions in the fuzzy scene grid, and then start from each flight starting grid at the same time to obtain the attraction value β between each flight starting grid and each attraction grid. The calculation formula of the attraction value β is: ; in Represents the attraction value between the i-th flight starting point grid and the j-th attraction grid, To correct the parameters, , , as well as Represents the coordinate values ​​of the i-th flight starting point grid and the j-th attraction grid, represents the attraction degree of the jth attraction grid; According to the direction of the attraction grid with the largest attraction value β, a two-dimensional image grid adjacent to the flight starting point grid is selected as the next flight grid, and it is determined whether the flight grids associated with each flight starting point grid overlap. If so, the maximum attraction values ​​β associated with the two flight starting point grids are compared, and then the flight starting point grid with the smallest attraction value β selects the direction of the attraction grid with the second largest attraction value β, reselects a flight grid, and repeats the process of whether the flight grids overlap again until the flight grids associated with each flight starting point grid do not overlap; Then, the attraction value β between each attraction grid is re-obtained according to each flight starting grid, and the above process of setting the flight grid and determining whether there is a flight grid overlap is repeated until all the attraction grids are selected as the flight grid; The corresponding dynamic UAV flight path is generated according to the flight grid associated with each flight starting point grid, and then the UAV flight management module sends the dynamic UAV flight path and the viewpoint feature model corresponding to the attraction grid on it to each UAV.

[0021] Furthermore, the UAV flight management module sets a data upload cycle for each UAV, and then collects the second scene photography data within the shooting range of each UAV during its flight along the dynamic UAV flight path, and sends the collected second scene photography data to the photography data verification module at the end of each data upload cycle; The photographic data verification module performs difference detection on each second scene photographic data according to the difference detection process on the first scene photographic data, and sends all the second scene photographic data to the scene modeling end after the difference detection is completed; The photographic data fusion module in the scene modeling end fuses each second scene photographic data to obtain a scene visualization image of the corresponding data upload period, and sends the scene visualization image and the viewpoint feature model to the scene modeling module.

[0022] The scene modeling module establishes or updates a dynamic scene visualization model according to the scene visualization image, and marks a plurality of scene tracking grids in the dynamic scene visualization model through the viewpoint feature model, and then sends the dynamic scene visualization model to the drone management terminal and the user; Whenever a data upload cycle starts, the drone flight management module in the drone management terminal updates the distribution of each attraction grid in the fuzzy scene grid according to the scene tracking grid in the dynamic scene visualization model, and then replans the dynamic drone flight path of each drone and sends it to each drone synchronously; The above process of establishing or updating the dynamic scene visualization model is repeated until the user sends a data collection completion prompt to the scene modeling end.

[0023] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any indirect modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A dynamic modeling viewpoint optimization system based on drone photography, including a cloud computing platform, characterized in that: The cloud computing platform is communicatively connected to a drone management terminal and a scene modeling terminal; The drone management end is used to collect first scene photography data of the target scene area through the drone, and establish multiple dynamic drone flight paths according to the fuzzy scene grid from the scene modeling end; Then, each drone collects second scene photography data of the target scene area along the dynamic drone flight path, and dynamically updates the dynamic drone flight path according to the change of the attraction grid position in the fuzzy scene grid during the drone flight, and performs difference detection on the first scene photography data and the second scene photography data; The scene modeling end is used to obtain modeling viewpoint information, and establish a fuzzy scene grid according to the first scene photography data, and then mark a number of attraction grids on the fuzzy scene grid according to the modeling viewpoint information, and at the same time establish a dynamic scene visualization model according to the second scene photography data, and update the position of the attraction grid on the fuzzy scene grid based on the dynamic scene visualization model.

2. The dynamic modeling viewpoint optimization system based on drone photography according to claim 1 is characterized in that: The process of collecting the first scene photography data includes: Set n drones, each of which has a camera range of radius r, and set n one-way drone flight routes according to the scene acquisition range, and the interval distance between each one-way drone flight route is less than r, where n is a natural number greater than 10; Each drone is made to fly over the target scene area along a one-way drone flight route at the same speed, and during the flight, each drone collects the first scene photography data within its shooting range.

3. The dynamic modeling viewpoint optimization system based on drone photography according to claim 2 is characterized in that: The process performed on the first scene photographic data includes: Establish a two-dimensional coordinate system, map the first scene photography data collected by each UAV on the two-dimensional coordinate system according to its flight route, perform grayscale processing on all the first scene photography data, and set a pixel difference threshold; Determine the size relationship between the pixel difference between the grayscale pixels in the overlapping state in the two-dimensional coordinate system and the pixel difference threshold, and determine again whether the adjacent grayscale pixels of the corresponding grayscale pixels have the same size relationship at the same time according to the determination result, and perform pixel averaging processing on the grayscale pixels according to the determination result; Any layer of grayscale pixels in the overlapping position of grayscale pixels in the splicing state is displaced and translated in four directions in sequence along the plane direction by a distance of a grayscale pixel; The magnitude relationship between the pixel difference between the grayscale pixels in the overlapping state in the grayscale pixel overlapping position after each translation and the pixel difference threshold is judged in turn, and pixel averaging processing is performed on the grayscale pixels in the overlapping state according to the judgment result.

4. The dynamic modeling viewpoint optimization system based on drone photography according to claim 3 is characterized in that: The process of establishing the fuzzy scene grid includes: Performing image stitching on the first scene photography data, and establishing a fuzzy scene grid according to the image stitching result, wherein the fuzzy scene grid is composed of a plurality of two-dimensional image grids; A plurality of viewpoint feature models are marked on the fuzzy scene grid according to the modeling viewpoint information, and a photography priority is set for each viewpoint feature model according to the modeling viewpoint information priority level set by the user.

5. The dynamic modeling viewpoint optimization system based on drone photography according to claim 4 is characterized in that: The process of establishing the dynamic UAV flight path includes: According to the position distribution of the viewpoint feature model in the fuzzy scene grid, a plurality of attraction grids are set in the fuzzy scene grid, and according to the photographic priority of each viewpoint feature model, an attraction degree is set for each attraction grid; n flight starting point grids with different spatial positions are set in the fuzzy scene grid, and then starting from each flight starting point grid at the same time, an attraction value β between each flight starting point grid and each attraction grid is obtained; According to the direction of the attraction grid with the largest attraction value β, a two-dimensional image grid adjacent to the flight starting point grid is selected as the next flight grid, and it is determined whether the flight grids associated with each flight starting point grid overlap, and a flight grid is reselected according to the determination result, and the determination of whether the flight grids overlap is repeated again until the flight grids associated with each flight starting point grid do not overlap; Then, the attraction value β between each attraction grid is re-obtained according to each flight starting point grid, and the above process of setting the flight grid and determining whether there is a flight grid overlap is repeated until each attraction grid is selected as a flight grid. The corresponding dynamic UAV flight path is generated according to the flight grid associated with each flight starting point grid.

6. The dynamic modeling viewpoint optimization system based on drone photography according to claim 5, characterized in that: According to the photography priority from high to low, the attraction degree of the corresponding attraction grid decreases successively.

7. The dynamic modeling viewpoint optimization system based on drone photography according to claim 6, characterized in that: The process of collecting the second scene photographic data includes: A data uploading cycle is set for each drone, and then the second scene photography data within the shooting range of each drone is collected while the drone is flying along the dynamic drone flight path.

8. The dynamic modeling viewpoint optimization system based on drone photography according to claim 7, characterized in that: The process of establishing a dynamic scene visualization model includes: fusing each second scene photography data to obtain a scene visualization image of a corresponding data upload period, establishing or updating a dynamic scene visualization model according to the scene visualization image, and marking a plurality of scene tracking grids in the dynamic scene visualization model through a viewpoint feature model; Whenever a data upload cycle starts, the drone management end updates the distribution of each attraction grid in the fuzzy scene grid according to the scene tracking grid in the dynamic scene visualization model, and then replans the dynamic drone flight path of each drone.

Images