Unmanned aerial vehicle remote sensing image data monitoring system and method for grassland mouse wasteland

Abstract

This invention relates to the field of remote sensing image monitoring technology, specifically to a UAV remote sensing image data monitoring system and method for grassland rodent wastelands. The system includes a status data acquisition and processing module. This module uses a camera and hyperspectral imager mounted on the UAV to acquire images of ground features in the target area of ​​the grassland, obtaining raw image data of the target area and UAV flight status information during the acquisition of the corresponding raw image data. Based on the UAV flight status information, the acquired raw image data is preprocessed to obtain preprocessed image information data. This invention considers the deviation in shooting angle caused by environmental factors during UAV flight; and determines the interference situation of remote sensing images corresponding to different locations, enabling the screening of areas and locations to be re-captured, and achieving effective monitoring of UAV remote sensing image data.

Description

Technical Field

[0001] This invention relates to the field of remote sensing image monitoring technology, specifically to a UAV remote sensing image data monitoring system and method for prairie mouse wasteland. Background Technology

[0002] With the development of deep learning technology and the increasing prevalence of drones, people's ability to observe the ground has been greatly enhanced. Previously, due to factors such as the resolution, limited information acquisition, and high cost of satellite remote sensing imagery, most scientists faced significant constraints in ground object monitoring. However, drone remote sensing technology, which provides convenient, fast, and low-cost information sources, has become a key carrier for the acquisition, processing, and application of ground object scientific information, playing a vital role in national ecological security and grassland animal husbandry development.

[0003] However, existing UAV-based remote sensing image data is obtained through cameras and hyperspectral imagers equipped on the UAVs. During flight, UAVs are often affected by environmental factors (such as wind affecting their flight status), which can cause deviations in the shooting angle when acquiring remote sensing images. This leads to abnormalities in the acquisition angle (which can cause data gaps, such as grass obscuring adjacent bare ground, reducing the bare ground area in the acquired image) and acquisition range (changes in the acquisition angle can affect the camera's field of view, and the actual distance corresponding to the same pixel spacing in the same remote sensing image may be different). These abnormalities affect the subsequent segmentation results of the remote sensing images. Therefore, existing UAV-based remote sensing image data monitoring systems have significant shortcomings. Summary of the Invention

[0004] The purpose of this invention is to provide a UAV remote sensing image data monitoring system and method for prairie mouse wasteland, in order to solve the problems mentioned in the background art.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a method for monitoring prairie mouse wasteland using unmanned aerial vehicle (UAV) remote sensing image data, the method comprising the following steps:

[0006] S1. Use the camera and hyperspectral imager equipped on the UAV to collect images of ground features in the target area of ​​the grassland, obtain the original image data of the target area and the flight status information of the UAV when collecting the corresponding original image data; preprocess the collected original image data according to the flight status information of the UAV to obtain preprocessed image information data.

[0007] S2. Construct a location region model of the target area, where each location point in the location region model corresponds to a unique location coordinate; construct a grassland feature segmentation model based on deep learning and grassland feature definition; train, verify, and test the constructed grassland feature segmentation model; after comparative analysis, use the optimal segmentation model to segment the preprocessed image information data to obtain the segmentation result; bind each location coordinate corresponding to the segmentation result to the object to which the segmentation result belongs.

[0008] S3. Combine the preprocessed image information data to analyze the information interference deviation of each position coordinate in the position region model; combine the information interference deviation of each position coordinate in the position region model with the bound segmentation result to the object to which it belongs to construct position feature information, form a mapping between position coordinates and position feature information in the position region model, and obtain integrated data of segmented images in the target region.

[0009] S4. Query the set of re-inspection objects through the database preset form, and divide the target area into several re-inspection areas of the same size. Analyze the information interference comprehensive influence value of all elements in the re-inspection object set corresponding to the set of regions to which the re-inspection objects belong in the integrated data of the segmented images in each re-inspection area of ​​the target area, and generate the remote sensing image re-capture coordinate set of the target area based on the analysis results. The remote sensing image re-capture coordinate set includes 0 or 1 or more remote sensing image re-capture coordinates, and one remote sensing image re-capture coordinate corresponds to one re-inspection area.

[0010] Furthermore, the original image data of the target area in S1 includes several remote sensing images of the target area taken by the UAV;

[0011] The UAV flight status information includes the UAV's location, flight altitude, and fuselage tilt vector (obtained by a gyroscope on the UAV); the flight altitude is the difference between the altitude of the UAV and the average altitude of the target area; the fuselage tilt vector represents a vector with a unit length, perpendicular to the bottom surface of the fuselage, and arranged from high to low.

[0012] The method for obtaining preprocessed image information data in S1 includes the following steps:

[0013] S11. Obtain the original image data of the target area, and denote the image data corresponding to the i-th remote sensing image in the original image data as Ai; and denote the flight status information corresponding to the UAV when it obtains Ai as Bi, wherein Bi = {B1bi, B2bi, B3bi}, where B1bi represents the latitude and longitude coordinates of the UAV's location in Bi, B2bi represents the flight altitude of the UAV in Bi, and B3bi represents the fuselage tilt vector in Bi.

[0014] S12. Take the pixels corresponding to the outer contour of Ai as nodes of Ai; denote the fuselage tilt vector when the UAV is flying horizontally as the standard tilt vector; obtain the latitude and longitude coordinates of the j-th node position in the image captured by the UAV at flight altitude B2bi and in horizontal flight from the database as (T1...). (B2bi，j) T2 (B2bi，j) The latitude and longitude coordinates of the j-th node position in the image captured by the UAV at a flight altitude of B2bi and a fuselage tilt vector of B3bi from the database are denoted as (T1B3bi). (B2bi，j) T2B3bi (B2bi，j) ), which is the deviation (T1B3bi) of the latitude and longitude coordinates of the j-th node position in the image captured when the drone is flying at an altitude of B2bi and the fuselage tilt vector is B3bi. (B2bi，j) -T1 (B2bi，j) T2B3bi (B2bi，j) -T2 (B2bi，j) );

[0015] S13. Obtain the latitude and longitude coordinates Cij corresponding to the j-th node in the preprocessed image information data of Ai.

[0016] The Cij = (B1bi1 + T01 + D1) (B2bi，j，B3bi) B1bi2+T02+D2 (B2bi，j，B3bi) ),

[0017] D1 (B2bi，j，B3bi) =T1B3bi (B2bi，j) -T1 (B2bi，j) ,

[0018] D2 (B2bi，j，B3bi) =T2B3bi (B2bi，j) -T2 (B2bi，j) ,

[0019] Where T01 represents the difference between the longitude of the j-th node and the longitude of the aircraft in the image taken by the drone at a flight altitude of B2bi and in a horizontal state in the database; B1bi1 represents the longitude in the latitude and longitude coordinates B1bi.

[0020] In this invention, B1bi1+T01 represents the longitude of the j-th node position when the longitude of the drone's location is B1bi1 in the image taken by the drone in the database when the drone is flying at an altitude of B2bi and in a horizontal state.

[0021] T02 represents the difference between the latitude of the j-th node and the latitude of the aircraft in the image taken by the drone at a flight altitude of B2bi and in a horizontal state in the database; B1bi2 represents the latitude in the latitude and longitude coordinates B1bi.

[0022] S13. Based on the latitude and longitude coordinates of each node in Ai, scale and adjust Ai in the latitude and longitude coordinate system.

[0023] In this invention, during scaling adjustment, the adjustment coefficient between two nodes on the same axis is first obtained. This coefficient is equal to the ratio between the actual latitude and longitude distance between the two nodes and the pixel distance in the image. After scaling adjustment, the latitude and longitude distance between any two pixels between two corresponding nodes in the image is equal to the product of the distance between the two pixels and the corresponding adjustment coefficient.

[0024] Mark the actual region corresponding to the Ai image in latitude and longitude coordinates, obtain the latitude and longitude coordinates corresponding to different pixel positions in Ai, and bind the image information corresponding to different pixel positions in the preprocessed image information data corresponding to Ai with the corresponding latitude and longitude coordinate points to obtain the preprocessed image information data of Ai.

[0025] In this invention, when shooting images from the same flight altitude using an oblique angle, the actual range of the field of view differs. The larger the oblique angle, the larger the area captured. However, at an oblique angle, grass in grassland wasteland (which has a height difference relative to the wasteland) can obstruct grass-free wasteland and may also obscure mouse burrows, making it impossible to identify mouse burrow information hidden by grass.

[0026] Furthermore, the position coordinates in the location region model in S2 are latitude and longitude coordinates.

[0027] The grassland feature definitions in S2 are pre-defined in the database.

[0028] In S2, when the preprocessed image information data is segmented using the constructed grassland feature segmentation model, the preprocessed image information data corresponding to each remote sensing image in the original image data is segmented separately. The objects to which the segmentation results belong for the same latitude and longitude coordinates corresponding to different remote sensing images in the original image may differ.

[0029] Furthermore, the method for analyzing the information interference deviation affecting the coordinates of each location in the location region model in S3 includes the following steps:

[0030] S301. Obtain the segmentation processing result of the preprocessed image information data corresponding to each remote sensing image in the original image data;

[0031] S302. Denote the latitude and longitude coordinates of any point in the target area as Q in the latitude and longitude coordinate system; obtain the objects to which the segmentation results of Q belong in the preprocessed image information data corresponding to each remote sensing image containing Q, and construct the segmentation object association set GQ = {GQ1, GQ2, ..., GQn} of Q, where GQn represents the object to which the segmentation results of Q belong in the preprocessed image information data corresponding to the nth remote sensing image containing Q;

[0032] S303. Obtain the fuselage tilt vector within the UAV flight status information corresponding to each remote sensing image containing Q, and construct the fuselage tilt association set SQ = {SQ1, SQ2, ..., SQn} of Q, where SQn represents the fuselage tilt vector within the UAV flight status information corresponding to the nth remote sensing image containing Q.

[0033] S304. Obtain the information interference deviation impact value of Q, denoted as PQ, where PQ = ∑ n1＝1 n Un1×cosθ,

[0034] θ is equal to the angle between the resultant vector of each element in SQ and the vertical line.

[0035] Un1 represents the proportion of the abnormal location region in the remote sensing image corresponding to the n1th element in SQ. The proportion of the abnormal location region in each remote sensing image is equal to the quotient of the actual area of ​​the abnormal location region in the latitude and longitude coordinate system of the corresponding remote sensing image divided by the actual area of ​​the corresponding remote sensing image in the latitude and longitude coordinate system. In the segmentation object association set corresponding to each latitude and longitude coordinate in the abnormal location region, there are elements whose respective segmentation results correspond to different objects.

[0036] Furthermore, the method for obtaining the integrated data of the segmented image within the target region in S3 includes the following steps:

[0037] S311, The impact of information interference and deviation on each location coordinate in the location region model and the object to which the bound segmentation result belongs;

[0038] S312. Select all latitude and longitude coordinate points in the target area whose corresponding information interference deviation influence value is less than or equal to the first preset value, and mark the selected latitude and longitude coordinate points. The first preset value is a constant preset in the database.

[0039] S313. Construct location feature information. The location feature information of unlabeled latitude and longitude coordinate points is the best matching result of the segmented object of the corresponding latitude and longitude coordinate points. The location feature information of labeled latitude and longitude coordinate points is an empty set.

[0040] The best matching result of the segmented object is obtained by obtaining the priority of each element in the segmented object association set of the corresponding unmarked latitude and longitude coordinate points, and selecting the segmented object with the highest priority. The priority of the segmented object is obtained by querying a pre-set form in the database.

[0041] S314. Map the location coordinates and location feature information in the location region model to obtain integrated data of the segmented image within the target region.

[0042] Furthermore, the method in S4 for analyzing the comprehensive impact value of information interference of all elements in the set of re-examined objects corresponding to the set of regions to which the re-examined objects belong in the integrated data of the segmented images in each re-examined region of the target region includes the following steps:

[0043] S41. Obtain the intersection result of the set of associated segments and the set of re-examined objects in the integrated data for each re-examined area corresponding to the target area. Record the markers where the intersection is an empty set as first-type markers and the markers where the intersection is not an empty set as second-type markers.

[0044] S42. Obtain the reshooting requirement feature value of each re-inspection area after being affected by the re-inspection objects corresponding to all elements in the re-inspection object set, denoted as R. The R is equal to the ratio of the area occupied by the second type of marker points in the corresponding re-inspection area to the total area of ​​the re-inspection area.

[0045] S43. Obtain the best matching result of the segmentation object of the corresponding latitude and longitude coordinate point among the first type of marker points and the corresponding second marker points whose R is less than the second preset value in the re-inspection area, and update the position feature information of the corresponding marker points in the integrated data;

[0046] Obtain the latitude and longitude coordinates of all second marker points in the re-inspection area whose corresponding R is greater than or equal to the second preset value, determine that the corresponding re-inspection area needs to be re-shot with remote sensing images, and obtain the remote sensing image re-shot coordinates of the corresponding re-inspection area. The remote sensing image re-shot coordinates are the latitude and longitude coordinates of the center point of the area enclosed by all second marker points in the corresponding re-inspection area.

[0047] Each re-inspection area with an R value greater than or equal to the second preset value corresponds to a remote sensing image re-capture coordinate.

[0048] The purpose of this invention to retake remote sensing images of the re-inspection area is to update the integrated data later, fill in the location feature information of the second type of marker points in the integrated data, and make the segmentation results of the updated integrated data more accurate when the grassland feature segmentation model is used to segment the data.

[0049] A drone-based remote sensing imagery data monitoring system for prairie vole habitats, comprising the following modules:

[0050] The status data acquisition and processing module uses the camera and hyperspectral imager equipped on the UAV to acquire images of ground features in the target area of ​​the grassland, obtaining the original image data of the target area and the flight status information of the UAV when acquiring the corresponding original image data; and preprocesses the acquired original image data according to the UAV flight status information to obtain preprocessed image information data.

[0051] The model construction and analysis module constructs a location region model of the target area, where different location points in the location region model correspond to unique location coordinates. Based on deep learning and grassland feature definitions, a grassland feature segmentation model is constructed. This model is trained, validated, and tested. After comparative analysis, the optimal segmentation model is used to segment the preprocessed image information data, obtaining the segmentation results. The location coordinates corresponding to each segmentation result are then bound to the object to which the segmentation result belongs.

[0052] The segmentation data integration module combines preprocessed image information data to analyze the information interference deviation of each position coordinate in the location region model; combining the information interference deviation of each position coordinate in the location region model and the bound segmentation result to its corresponding object, it constructs position feature information, forming a mapping between position coordinates and position feature information in the location region model, and obtains integrated data of segmented images within the target region.

[0053] The supplementary image data analysis and management module queries the set of re-inspection objects through a pre-set form in the database, evenly divides the target area into several re-inspection areas of the same size, analyzes the information interference comprehensive influence value of all elements in the re-inspection object set corresponding to the set of regions to which the re-inspection objects belong in the integrated data of the segmented images in each re-inspection area of ​​the target area, and generates a set of remote sensing image supplementary coordinates for the target area based on the analysis results.

[0054] Furthermore, the segmented data integration module includes a deviation impact analysis unit, a location feature analysis unit, and a data integration analysis unit.

[0055] The deviation impact analysis unit combines preprocessed image information data to analyze the information interference deviation impact on the coordinates of each location in the location region model.

[0056] The location feature analysis unit combines the information interference and deviation effects on each location coordinate in the location region model with the object to which the bound segmentation result belongs to construct location feature information.

[0057] The data integration and analysis unit combines the constructed location feature information to form a mapping between location coordinates and location feature information in the location region model, thereby obtaining integrated data of segmented images within the target region.

[0058] The remote sensing image re-capture coordinate set includes 0, 1, or more remote sensing image re-capture coordinates, and one remote sensing image re-capture coordinate corresponds to one re-inspection area.

[0059] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: When acquiring remote sensing image data based on UAVs, the present invention takes into account the deviation caused by the shooting angle when acquiring remote sensing images after the UAV is affected by environmental factors during flight; and considers the acquisition angle and acquisition range of the obtained remote sensing images to determine the interference situation of remote sensing images corresponding to different locations, thereby achieving the screening of areas and locations to be re-shot and realizing effective supervision of UAV remote sensing image data. Attached Figure Description

[0060] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0061] Figure 1 This is a flowchart illustrating the method for monitoring unmanned aerial vehicle (UAV) remote sensing image data in grassland rat wasteland according to the present invention.

[0062] Figure 2 This is a schematic diagram of the structure of the UAV remote sensing image data monitoring system for grassland rat wasteland according to the present invention. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] Please see Figure 1 The present invention provides a technical solution: a method for monitoring prairie rat wasteland using unmanned aerial vehicle (UAV) remote sensing image data, the method comprising the following steps:

[0065] S1. Use the camera and hyperspectral imager equipped on the UAV to collect images of ground features in the target area of ​​the grassland, obtain the original image data of the target area and the flight status information of the UAV when collecting the corresponding original image data; preprocess the collected original image data according to the flight status information of the UAV to obtain preprocessed image information data.

[0066] The original image data of the target area in S1 includes several remote sensing images of the target area taken by the UAV;

[0067] The UAV flight status information includes the UAV's location, flight altitude, and fuselage tilt vector;

[0068] In this invention, the fuselage tilt vector is obtained by monitoring the gyroscope mounted on the UAV.

[0069] In this embodiment, the camera on the drone points to a plane perpendicular to the bottom of the drone;

[0070] The flight altitude is the difference between the altitude of the UAV and the average altitude of the target area, and the fuselage tilt vector represents a vector with a unit length, perpendicular to the bottom surface of the fuselage, and arranged from high to low.

[0071] The method for obtaining preprocessed image information data in S1 includes the following steps:

[0072] S11. Obtain the original image data of the target area, and denote the image data corresponding to the i-th remote sensing image in the original image data as Ai; and denote the flight status information corresponding to the UAV when it obtains Ai as Bi, wherein Bi = {B1bi, B2bi, B3bi}, where B1bi represents the latitude and longitude coordinates of the UAV's location in Bi, B2bi represents the flight altitude of the UAV in Bi, and B3bi represents the fuselage tilt vector in Bi.

[0073] S12. Take the pixels corresponding to the outer contour of Ai as nodes of Ai; denote the fuselage tilt vector when the UAV is flying horizontally as the standard tilt vector; obtain the latitude and longitude coordinates of the j-th node position in the image captured by the UAV at flight altitude B2bi and in horizontal flight from the database as (T1...). (B2bi，j) T2 (B2bi，j) The latitude and longitude coordinates of the j-th node position in the image captured by the UAV at a flight altitude of B2bi and a fuselage tilt vector of B3bi from the database are denoted as (T1B3bi). (B2bi，j) T2B3bi (B2bi，j) ), which is the deviation (T1B3bi) of the latitude and longitude coordinates of the j-th node position in the image captured when the drone is flying at an altitude of B2bi and the fuselage tilt vector is B3bi. (B2bi，j) -T1 (B2bi，j) T2B3bi (B2bi，j) -T2 (B2bi，j) );

[0074] S13. Obtain the latitude and longitude coordinates Cij corresponding to the j-th node in the preprocessed image information data of Ai.

[0075] The Cij = (B1bi1 + T01 + D1) (B2bi，j，B3bi) B1bi2+T02+D2 (B2bi，j，B3bi) ),

[0076] D1 (B2bi，j，B3bi) =T1B3bi (B2bi，j) -T1 (B2bi，j) ,

[0077] D2 (B2bi，j，B3bi) =T2B3bi (B2bi，j) -T2 (B2bi，j) ,

[0078] Where T01 represents the difference between the longitude of the j-th node and the longitude of the aircraft in the image taken by the drone at a flight altitude of B2bi and in a horizontal state in the database; B1bi1 represents the longitude in the latitude and longitude coordinates B1bi.

[0079] T02 represents the difference between the latitude of the j-th node and the latitude of the aircraft in the image taken by the drone at a flight altitude of B2bi and in a horizontal state in the database; B1bi2 represents the latitude in the latitude and longitude coordinates B1bi.

[0080] In this embodiment, when the latitude and longitude coordinates are outside the range of values, they are automatically adjusted to obtain the latitude and longitude coordinates of the corresponding position within the range of values. For example, if the longitude value in ZAi is 181 degrees when calculating the reference point, then the longitude in the adjusted latitude and longitude coordinates of the same position point is -179 degrees.

[0081] S13. Based on the latitude and longitude coordinates of each node in Ai, scale and adjust Ai in the latitude and longitude coordinate system, mark the actual area corresponding to the Ai image in the latitude and longitude coordinate system, obtain the latitude and longitude coordinates corresponding to different pixel positions in Ai, and bind the image information corresponding to different pixel positions in the preprocessed image information data of Ai with the corresponding latitude and longitude coordinate points to obtain the preprocessed image information data of Ai.

[0082] In this invention, when shooting images from the same flight altitude using an oblique angle, the actual range of the field of view is different. The larger the oblique angle, the larger the area captured. Under an oblique angle, grass in grassland wasteland (which has a height difference relative to the wasteland) will obstruct the grassless wasteland and may also cover mouse holes, making it impossible to identify mouse hole information covered by grass.

[0083] S2. Construct a location region model of the target area, where each location point in the location region model corresponds to a unique location coordinate; construct a grassland feature segmentation model based on deep learning and grassland feature definition; train, verify, and test the constructed grassland feature segmentation model; after comparative analysis, use the optimal segmentation model to segment the preprocessed image information data to obtain the segmentation result; bind each location coordinate corresponding to the segmentation result to the object to which the segmentation result belongs.

[0084] The location coordinates in the location region model in S2 are latitude and longitude coordinates.

[0085] The grassland feature definitions in S2 are pre-defined in the database.

[0086] In this embodiment, grassland features are defined as vegetation, bare soil, mounds, and mouse burrows.

[0087] In S2, when the preprocessed image information data is segmented using the constructed grassland feature segmentation model, the preprocessed image information data corresponding to each remote sensing image in the original image data is segmented separately. The objects to which the segmentation results belong for the same latitude and longitude coordinates corresponding to different remote sensing images in the original image may differ.

[0088] S3. Combine the preprocessed image information data to analyze the information interference deviation of each position coordinate in the position region model; combine the information interference deviation of each position coordinate in the position region model with the bound segmentation result to the object to which it belongs to construct position feature information, form a mapping between position coordinates and position feature information in the position region model, and obtain integrated data of segmented images in the target region.

[0089] The method for analyzing the impact of information interference deviations on the coordinates of each location in the location region model in S3 includes the following steps:

[0090] S301. Obtain the segmentation processing result of the preprocessed image information data corresponding to each remote sensing image in the original image data;

[0091] S302. Denote the latitude and longitude coordinates of any point in the target area as Q in the latitude and longitude coordinate system; obtain the objects to which the segmentation results of Q belong in the preprocessed image information data corresponding to each remote sensing image containing Q, and construct the segmentation object association set GQ = {GQ1, GQ2, ..., GQn} of Q, where GQn represents the object to which the segmentation results of Q belong in the preprocessed image information data corresponding to the nth remote sensing image containing Q;

[0092] S303. Obtain the fuselage tilt vector within the UAV flight status information corresponding to each remote sensing image containing Q, and construct the fuselage tilt association set SQ = {SQ1, SQ2, ..., SQn} of Q, where SQn represents the fuselage tilt vector within the UAV flight status information corresponding to the nth remote sensing image containing Q.

[0093] S304. Obtain the information interference deviation impact value of Q, denoted as PQ, where PQ = ∑ n1＝1 n Un1×cosθ,

[0094] θ is equal to the angle between the resultant vector of each element in SQ and the vertical line.

[0095] Un1 represents the proportion of the abnormal location region in the remote sensing image corresponding to the n1th element in SQ. The proportion of the abnormal location region in each remote sensing image is equal to the quotient of the actual area of ​​the abnormal location region in the latitude and longitude coordinate system of the corresponding remote sensing image divided by the actual area of ​​the corresponding remote sensing image in the latitude and longitude coordinate system. In the segmentation object association set corresponding to each latitude and longitude coordinate in the abnormal location region, there are elements whose respective segmentation results correspond to different objects.

[0096] The method for obtaining the integrated data of the segmented image within the target region in S3 includes the following steps:

[0097] S311, The impact of information interference and deviation on each location coordinate in the location region model and the object to which the bound segmentation result belongs;

[0098] S312. Select all latitude and longitude coordinate points in the target area whose corresponding information interference deviation influence value is less than or equal to the first preset value, and mark the selected latitude and longitude coordinate points. The first preset value is a constant preset in the database.

[0099] S313. Construct location feature information. The location feature information of unlabeled latitude and longitude coordinate points is the best matching result of the segmented object of the corresponding latitude and longitude coordinate points. The location feature information of labeled latitude and longitude coordinate points is an empty set.

[0100] The best matching result of the segmented object is obtained by obtaining the priority of each element in the segmented object association set of the corresponding unmarked latitude and longitude coordinate points, and selecting the segmented object with the highest priority. The priority of the segmented object is obtained by querying a pre-set form in the database.

[0101] In this embodiment, the priority of bare soil in the segmentation object is higher than that of vegetation in the segmentation object;

[0102] S314. Map the location coordinates and location feature information in the location region model to obtain integrated data of the segmented image within the target region.

[0103] S4. Query the set of re-inspection objects through the database preset form, and divide the target area into several re-inspection areas of the same size. Analyze the information interference comprehensive influence value of all elements in the re-inspection object set corresponding to the set of regions to which the re-inspection objects belong in the integrated data of the segmented images in each re-inspection area of ​​the target area, and generate the remote sensing image re-capture coordinate set of the target area based on the analysis results. The remote sensing image re-capture coordinate set includes 0 or 1 or more remote sensing image re-capture coordinates, and one remote sensing image re-capture coordinate corresponds to one re-inspection area.

[0104] The method for analyzing the comprehensive impact value of information interference of all elements in the set of re-examined objects corresponding to the set of regions to which the re-examined objects belong in the integrated data of the segmented images in each re-examined region of the target region in S4 includes the following steps:

[0105] S41. Obtain the intersection result of the set of associated segments and the set of re-examined objects in the integrated data for each re-examined area corresponding to the target area. Record the markers where the intersection is an empty set as first-type markers and the markers where the intersection is not an empty set as second-type markers.

[0106] S42. Obtain the reshooting requirement feature value of each re-inspection area after being affected by the re-inspection objects corresponding to all elements in the re-inspection object set, denoted as R. The R is equal to the ratio of the area occupied by the second type of marker points in the corresponding re-inspection area to the total area of ​​the re-inspection area.

[0107] S43. Obtain the best matching result of the segmentation object of the corresponding latitude and longitude coordinate point among the first type of marker points and the corresponding second marker points whose R is less than the second preset value in the re-inspection area, and update the position feature information of the corresponding marker points in the integrated data;

[0108] Obtain the latitude and longitude coordinates of all second marker points in the re-inspection area whose corresponding R is greater than or equal to the second preset value, determine that the corresponding re-inspection area needs to be re-shot with remote sensing images, and obtain the remote sensing image re-shot coordinates of the corresponding re-inspection area. The remote sensing image re-shot coordinates are the latitude and longitude coordinates of the center point of the area enclosed by all second marker points in the corresponding re-inspection area.

[0109] Each re-inspection area with an R value greater than or equal to the second preset value corresponds to a remote sensing image re-capture coordinate.

[0110] like Figure 2 As shown, a UAV remote sensing image data monitoring system for prairie rat wasteland includes the following modules:

[0111] The status data acquisition and processing module uses the camera and hyperspectral imager equipped on the UAV to acquire images of ground features in the target area of ​​the grassland, obtaining the original image data of the target area and the flight status information of the UAV when acquiring the corresponding original image data; and preprocesses the acquired original image data according to the UAV flight status information to obtain preprocessed image information data.

[0112] The model construction and analysis module constructs a location region model of the target area, where different location points in the location region model correspond to unique location coordinates. Based on deep learning and grassland feature definitions, a grassland feature segmentation model is constructed. This model is trained, validated, and tested. After comparative analysis, the optimal segmentation model is used to segment the preprocessed image information data, obtaining the segmentation results. The location coordinates corresponding to each segmentation result are then bound to the object to which the segmentation result belongs.

[0113] The segmentation data integration module combines preprocessed image information data to analyze the information interference deviation of each position coordinate in the location region model; combining the information interference deviation of each position coordinate in the location region model and the bound segmentation result to its corresponding object, it constructs position feature information, forming a mapping between position coordinates and position feature information in the location region model, and obtains integrated data of segmented images within the target region.

[0114] The supplementary image data analysis and management module queries the set of re-inspection objects through a pre-set form in the database, evenly divides the target area into several re-inspection areas of the same size, analyzes the information interference comprehensive influence value of all elements in the re-inspection object set corresponding to the set of regions to which the re-inspection objects belong in the integrated data of the segmented images in each re-inspection area of ​​the target area, and generates a set of remote sensing image supplementary coordinates for the target area based on the analysis results.

[0115] The segmented data integration module includes a deviation impact analysis unit, a location feature analysis unit, and a data integration analysis unit.

[0116] The deviation impact analysis unit combines preprocessed image information data to analyze the information interference deviation impact on the coordinates of each location in the location region model.

[0117] The location feature analysis unit combines the information interference and deviation effects on each location coordinate in the location region model with the object to which the bound segmentation result belongs to construct location feature information.

[0118] The data integration and analysis unit combines the constructed location feature information to form a mapping between location coordinates and location feature information in the location region model, thereby obtaining integrated data of segmented images within the target region.

[0119] The remote sensing image re-capture coordinate set includes 0, 1, or more remote sensing image re-capture coordinates, and one remote sensing image re-capture coordinate corresponds to one re-inspection area.

[0120] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0121] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for monitoring unmanned aerial vehicle remote sensing image data for rangeland mouse infestation, characterized by, The method comprises the following steps: S1, using the camera and hyperspectral imager equipped in the unmanned aerial vehicle to collect the ground feature image of the target area in the grassland, to obtain the original image data of the target area and the unmanned aerial vehicle flight state information when collecting the corresponding original image data; according to the unmanned aerial vehicle flight state information, the collected original image data is preprocessed to obtain preprocessed image information data; S2, a location area model of the target area is constructed, and different position points in the location area model correspond to unique position coordinates; a grassland feature segmentation model is constructed based on deep learning and grassland feature definition, and the constructed grassland feature segmentation model is trained, verified and tested, and after comparative analysis, the optimal segmentation model is used to segment the preprocessed image information data to obtain a segmentation result; each position coordinate corresponding to the segmentation result is bound with the object to which the segmentation result belongs; S3, combined with the preprocessed image information data, the influence of information interference deviation on each position coordinate in the location area model is analyzed; combined with the influence of information interference deviation on each position coordinate in the location area model and the bound segmentation result belonging object, a location feature information is constructed, a mapping of the position coordinates and the location feature information in the location area model is formed, and integrated data of the segmented image in the target area is obtained; S4, the target area is uniformly divided into several reinspection areas with the same specification by querying the reinspection object set through the database preset form, the integrated data of the segmented image in each reinspection area of the target area is analyzed, the information interference comprehensive influence value of all elements in the reinspection object set on the reinspection object belonging area set is analyzed, and the remote sensing image retake coordinate set of the target area is generated according to the analysis result; the remote sensing image retake coordinate set includes 0 or 1 or more remote sensing image retake coordinates, and one remote sensing image retake coordinate corresponds to one reinspection area.

2. The method for monitoring the UAV remote sensing image data of the grassland mouse wasteland according to claim 1, characterized in that: The original image data of the target area in S1 includes several remote sensing images of the target area taken by the unmanned aerial vehicle; The unmanned aerial vehicle flight state information includes the position of the unmanned aerial vehicle, the flight height and the body inclination vector; the flight height is the difference between the altitude of the unmanned aerial vehicle and the average altitude of the target area, and the body inclination vector is a vector with a unit length, which is perpendicular to the bottom surface of the body and from high to low; The method for obtaining preprocessed image information data in S1 comprises the following steps: S11, obtaining the original image data of the target area, recording the image data corresponding to the i-th remote sensing image in the original image data as Ai; and recording the flight state information corresponding to Ai obtained by the unmanned aerial vehicle as Bi, Bi={B1bi, B2bi, B3bi}, wherein B1bi represents the longitude and latitude coordinates of the position of the unmanned aerial vehicle in Bi, B2bi represents the flight height of the unmanned aerial vehicle in Bi, and B3bi represents the body inclination vector in Bi; S12, the pixel points corresponding to the outer ring profile in Ai are taken as each node of Ai; the body inclination vector of the unmanned aerial vehicle in the horizontal state flight is recorded as a standard inclination vector; the longitude and latitude coordinates of the jth node position in the picture taken when the unmanned aerial vehicle flies at a height of B2bi and flies in the horizontal state are recorded as (T1 （B2bi，j） , T2 （B2bi，j） ); the longitude and latitude coordinates of the jth node position in the picture taken when the unmanned aerial vehicle flies at a height of B2bi and the body inclination vector is B3bi are recorded as (T1B3bi （B2bi，j） , T2B3bi （B2bi，j） ); and the deviation amount of the longitude and latitude coordinates of the jth node position in the picture taken when the unmanned aerial vehicle flies at a height of B2bi and the body inclination vector is B3bi is recorded as (T1B3bi （B2bi，j） -T1 （B2bi，j） , T2B3bi （B2bi，j） -T2 （B2bi，j） ). S13, obtaining the longitude and latitude coordinates Cij corresponding to the j-th node in the preprocessed image information data of Ai, The Cij = (B1bi1+T01+D1 （B2bi，j，B3bi） , B1bi2+T02+D2 （B2bi，j，B3bi） ), D1 （B2bi，j，B3bi） = T1B3bi （B2bi，j） - T1 （B2bi，j） , D2 （B2bi，j，B3bi） = T2B3bi （B2bi，j） - T2 （B2bi，j） , Wherein, T01 represents the difference between the longitude of the jth node position and the longitude of the position where the aircraft is located in the picture taken by the unmanned aerial vehicle at the flight height B2bi and in the horizontal state in the database; B1bi1 represents the longitude in the latitude and longitude coordinate B1bi, T02 represents the difference between the latitude of the jth node position and the latitude of the position where the aircraft is located in the picture taken by the unmanned aerial vehicle at the flight height B2bi and in the horizontal state in the database; B1bi2 represents the latitude in the latitude and longitude coordinate B1bi; S13, according to the latitude and longitude coordinates of each node in Ai, the Ai is scaled and adjusted in the latitude and longitude coordinate system, the actual area corresponding to the Ai image is marked in the latitude and longitude coordinate, the latitude and longitude coordinates corresponding to different pixel positions in Ai are obtained, and the image information corresponding to different pixel positions in the preprocessed image information data corresponding to Ai is bound with the corresponding latitude and longitude coordinate points, to obtain the preprocessed image information data of Ai. 3.The method for monitoring the rangeland and wasteland using UAV remote sensing image data according to claim 1, characterized in that: The position coordinates in the position area model in S2 are latitude and longitude coordinates, The grassland feature definition in S2 is preset in the database in advance, When the preprocessed image information data is segmented by the constructed grassland feature segmentation model in S2, the preprocessed image information data corresponding to each remote sensing image in the original image data is segmented and processed respectively, and the segmentation results of the same latitude and longitude coordinates corresponding to different remote sensing images in the original image belong to different objects. 4.The method for monitoring the UAV remote sensing image data of the grassland mouse wasteland according to claim 2, characterized in that: The method for analyzing the influence of information interference deviation on each position coordinate in the position area model in S3 includes the following steps: S301, obtaining the segmentation processing result of the preprocessed image information data corresponding to each remote sensing image in the original image data; S302, marking the latitude and longitude coordinates corresponding to any point in the target area as Q in the latitude and longitude coordinate system; obtaining the objects to which the segmentation results corresponding to Q belong in the preprocessed image information data corresponding to each remote sensing image containing Q, and constructing the segmentation object association set GQ of Q, wherein GQn represents the object to which the segmentation result corresponding to Q belongs in the preprocessed image information data corresponding to the nth remote sensing image containing Q; S303, obtaining the fuselage inclination vector in the unmanned aerial vehicle flight state information corresponding to each remote sensing image containing Q, and constructing the fuselage inclination association set SQ of Q, wherein SQn represents the fuselage inclination vector in the unmanned aerial vehicle flight state information corresponding to the nth remote sensing image containing Q; S304. Obtain the information interference deviation impact value of Q, denoted as PQ, where PQ = ∑ n1=1 n Un1×cosθ, θ is equal to the included angle between the resultant vector of each element in SQ and the vertical perpendicular line, Un1 represents the proportion of the abnormal position area in the nth1 element in SQ, and the proportion of the abnormal position area in each remote sensing image is equal to the quotient of the actual area of the abnormal position area in the corresponding remote sensing image in the latitude and longitude coordinate system divided by the actual area of the corresponding remote sensing image in the latitude and longitude coordinate system, and each latitude and longitude coordinate in the abnormal position area corresponds to an element in the segmentation object association set, and the element has different objects corresponding to the segmentation result.

5. The method for monitoring the rangeland mouse disaster area by using the UAV remote sensing image data according to claim 4, characterized in that: The method for obtaining the integrated data of the segmented image in the target region in S3 comprises the following steps: S311, obtaining the information interference deviation influence on each position coordinate in the position region model and the bound segmented result of the object; S312, selecting all the longitude and latitude coordinate points in the target region with the information interference deviation influence value less than or equal to a first preset value, and marking the selected longitude and latitude coordinate points, wherein the first preset value is a constant preset in the database; S313, constructing position feature information, wherein the position feature information of the unmarked longitude and latitude coordinate point is the best matching result of the segmented object of the corresponding longitude and latitude coordinate point, and the position feature information of the marked longitude and latitude coordinate point is an empty set, the best matching result of the segmented object is obtained by obtaining the priority of the segmented object corresponding to each element in the segmented object association set of the corresponding unmarked longitude and latitude coordinate point, and selecting the segmented object with the highest priority, and the priority of the segmented object is obtained by querying a database preset table; S314, mapping the position coordinates in the position region model and the position feature information to obtain the integrated data of the segmented image in the target region. 6.The method for monitoring the rangeland and wasteland using UAV remote sensing image data according to claim 1, characterized in that: The method for analyzing the integrated data of the segmented image in each review region in the target region in S4 comprises the following steps: S41, obtaining the intersection result of the segmented object association set and the review object set in the integrated data in each review region in the target region, marking the points with an empty intersection set as first type marking points, and marking the points with a non-empty intersection set as second type marking points; S42, obtaining the rephotographing demand characteristic value of each review region affected by all the elements in the review object set, denoted as R, wherein the R is equal to the ratio of the area of the region occupied by the second type marking points in the corresponding review region to the total area of the review region; S43, obtaining the segmented object best matching result of the corresponding longitude and latitude coordinate point in the second marking points with R less than a second preset value in the review region, and updating the position feature information of the corresponding marking points in the integrated data; obtaining the longitude and latitude coordinates of all the second marking points with R greater than or equal to the second preset value in the review region, determining that the corresponding review region needs to be rephotographed, and obtaining the rephotographing coordinates of the corresponding review region, wherein the rephotographing coordinates are the longitude and latitude coordinates of the center point of the region surrounded by all the second marking points in the corresponding review region; each review region corresponding to the R value greater than or equal to the second preset value has a rephotographing coordinate of a remote sensing image.

7. The unmanned aerial vehicle remote sensing image data monitoring system for grassland mouse land used in the unmanned aerial vehicle remote sensing image data monitoring method for grassland mouse land according to any one of claims 1-6, characterized in that, The system comprises the following modules: a state data acquisition and processing module, which acquires the ground feature image of the target region in the grassland by using the camera and the hyperspectral imager equipped in the unmanned aerial vehicle, obtains the original image data of the target region and the unmanned aerial vehicle flight state information when the corresponding original image data is acquired, and pre-processes the acquired original image data according to the unmanned aerial vehicle flight state information to obtain pre-processed image information data; The model construction and analysis module constructs a position area model of the target area, and different position points in the position area model correspond to unique position coordinates; a grassland feature segmentation model is constructed based on deep learning and grassland feature definition, the constructed grassland feature segmentation model is trained, verified and tested, the optimal segmentation model is adopted after comparative analysis to segment the preprocessed image information data to obtain a segmentation result; each position coordinate corresponding to the segmentation result is bound with the object to which the segmentation result belongs; The segmentation data integration module analyzes the information interference deviation influence on each position coordinate in the position area model in combination with the preprocessed image information data; position characteristic information is constructed in combination with the information interference deviation influence on each position coordinate in the position area model and the bound object to which the segmentation result belongs, a mapping of the position coordinates and the position characteristic information in the position area model is formed, and integrated data of the segmented image in the target area is obtained; The retake data analysis management module queries the review object set through a database preset form, divides the target area into several review areas with the same specification, analyzes the information interference comprehensive influence value of all elements in the review object set on the region set to which the review object belongs in the integrated data of the segmented image in each review area of the target area, and generates a remote sensing image retake coordinate set of the target area according to the analysis result.

8. The unmanned aerial vehicle remote sensing image data monitoring system for grassland mouse wasteland according to claim 7, characterized in that: The segmentation data integration module includes a deviation influence analysis unit, a position characteristic analysis unit and a data integration analysis unit, The deviation influence analysis unit analyzes the information interference deviation influence on each position coordinate in the position area model in combination with the preprocessed image information data; The position characteristic analysis unit constructs position characteristic information in combination with the information interference deviation influence on each position coordinate in the position area model and the bound object to which the segmentation result belongs, The data integration analysis unit forms a mapping of the position coordinates and the position characteristic information in the position area model in combination with the constructed position characteristic information, and obtains integrated data of the segmented image in the target area; the remote sensing image retake coordinate set includes 0 or 1 or more remote sensing image retake coordinates, and one remote sensing image retake coordinate corresponds to one review area.

Images