Unmanned aerial vehicle-based cultivated land occupation behavior identification early warning system and method

Abstract

The invention relates to the technical field of unmanned aerial vehicles, and discloses an unmanned aerial vehicle-based cultivated land occupation behavior identification early warning system and method, and the method comprises the following steps: S1, dividing a cultivated land region into N monitoring region units through a dividing module; s2, routing inspection positioning points and calibration points are arranged in each monitoring area unit; s3, the image acquisition unmanned aerial vehicle moves towards the inspection positioning point of the monitoring area unit, and the calibration unmanned aerial vehicle moves towards the corresponding calibration point; s4, acquiring and analyzing position information of the image acquisition unmanned aerial vehicle and the calibration unmanned aerial vehicle to judge whether image acquisition can be performed or not; if yes, entering S5; s5, acquiring M pieces of image information data of the monitoring area unit through an image acquisition unmanned aerial vehicle; a specified picture is screened out by analyzing the calibration unmanned aerial vehicle image in each image information data; the quality of the specified picture is ensured; s6, judging whether a cultivated land occupation behavior exists or not; and the accuracy of the judgment result is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles (UAVs), and in particular to a farmland occupation behavior recognition and early warning system and method based on unmanned aerial vehicles (UAVs). Background Art

[0002] The occupation of cultivated land refers to the act of engaging in non-agricultural construction or use on a specific piece of land; these acts may include building houses, opening factories, building gardens, and any other acts that change the original agricultural use of the land; in order to avoid the occurrence of such behavior, cultivated land needs to be regulated.

[0003] At present, in addition to manual supervision, the supervision method of cultivated land can also include the use of drones to collect images of cultivated land areas, and by identifying the collected cultivated land images, it can be determined whether there is any occupation of cultivated land.

[0004] However, the planting conditions of each farmland are different, and when drones are used for photography, the shooting environment will affect the pictures taken, resulting in uneven quality of the pictures. Since there are many plants on the farmland, the image elements taken are relatively complex, and the quality of the pictures is difficult to judge. When the farmland is converted into a garden, the pictures taken of the garden and the farmland are relatively similar. If the judgment of whether there is farmland occupation is based on pictures with low quality, the accuracy of the judgment result will be reduced. Summary of the invention

[0005] The purpose of the present invention is to provide a farmland occupation behavior recognition and early warning system and method based on drones to solve the following technical problems:

[0006] How to improve the accuracy of identifying farmland occupation behavior.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] A method for identifying and warning farmland occupation behavior based on drones, comprising the following steps:

[0009] S1: Divide the cultivated land area into N monitoring area units through the division module and number them in sequence, numbered n;

[0010] S2: Setting inspection positioning points and calibration points in each monitoring area unit through the preset route module, and setting a preset inspection route according to the position of each inspection positioning point;

[0011] S3: The image acquisition drone moves to the inspection positioning point of the monitoring area unit numbered n according to the preset inspection route, and the calibration drone follows the image acquisition drone to move toward the corresponding calibration point;

[0012] S4: obtaining the position information of the image acquisition drone and the calibration drone through the data acquisition module, obtaining a first judgment index by analyzing the position information of the image acquisition drone and the calibration drone, and judging whether image acquisition can be performed according to the first judgment index; if image acquisition can be performed, proceeding to step S5; otherwise, repeating step S4;

[0013] S5: using an image acquisition drone to collect M image information data of the monitoring area unit; using an analysis unit to analyze the calibrated drone image in each image information data, to obtain a quality evaluation index of each image information data, and to analyze and filter out a specified image according to the quality evaluation index of each image information data;

[0014] S6: Identify the designated image through the identification unit to determine whether there is any occupation of cultivated land in the monitoring area unit; if there is any occupation of cultivated land, issue an early warning, otherwise there is no occupation of cultivated land.

[0015] As a further solution of the present invention: by formula:

[0016] Calculate the first judgment index P n ;

[0017] Wherein, f(X) is the first judgment function. When X>0, f(X)=0; when X≤0, f(X)=1; (X A , Y A , Z A ) is the three-dimensional coordinate of the real-time position of the image acquisition drone; (X nx , Y nx , Z nx ) is the preset three-dimensional coordinate of the inspection positioning point of the nth monitoring area unit; (X B , Y B , Z B ) is the three-dimensional coordinate of the real-time position of the calibration drone; (X ny , Y ny , Z ny ) is the preset three-dimensional coordinate of the calibration point of the nth monitoring area unit; W 1 is the permissible inspection error value; W 2 is the allowable calibration error value.

[0018] As a further solution of the present invention: the judgment process of whether image acquisition can be performed is:

[0019] When P n =0, the image acquisition drone cannot perform image acquisition;

[0020] When P n=1, the image acquisition drone can perform image acquisition.

[0021] As a further solution of the present invention: by formula:

[0022] δ m =f[|S m -S 0 |-W S ]*f(Q 0 -Q m )*f(C 0 -C m )*f(R 0 -R m ) Calculate the qualified state coefficient δ of the mth image information data m ;

[0023] When δ m =0, the mth image information data is unqualified;

[0024] When δ m =1, the mth image information data is qualified.

[0025] As a further solution of the present invention: by formula:

[0026] Calculate the quality assessment index U of the mth image information data m

[0027] Among them, γ 1 is the first weight coefficient; γ 2 is the second weight coefficient; C 1 is the first preset constant; C 2 is the second preset constant.

[0028] As a further solution of the present invention: after step S6, the shooting status index of the image acquisition drone is obtained by analyzing the quality evaluation index of each image information data; and according to the shooting status index, it is determined whether the image acquisition drone continues to perform the task according to the preset inspection route.

[0029] As a further solution of the present invention: by formula:

[0030] Calculate the shooting state index T of the image acquisition drone state ;

[0031] Where D is the number of qualified states in the M image information data; k is the preset qualified rate; U 0 is the preset quality evaluation index; α 1 is the first weight coefficient of the state; α 2 is the second weight coefficient of the state; C3 is the third preset constant; C 4 is the fourth preset constant; T 0 The basic status index.

[0032] As a further solution of the present invention: the shooting state index T state Compared with the first preset value Z 1 and the second preset comparison value Z 1 Make a comparison;

[0033] When T state <Z 1 When the image acquisition module is seriously abnormal, an early warning is issued and the system is immediately returned for maintenance;

[0034] When Z 1 ≤T state <T 0 When the image acquisition module is slightly abnormal, an early warning is issued and the task is continued according to the preset inspection route before maintenance;

[0035] When T 0 ≤T state <Z 2 The image acquisition module is in normal state and continues to perform tasks according to the preset inspection route without warning.

[0036] When Z 2 ≤T state When the image acquisition module is in good condition, it performs tasks according to the preset inspection route without early warning.

[0037] A farmland occupation behavior recognition and early warning system based on an unmanned aerial vehicle, the recognition and early warning system comprising:

[0038] The division module divides the cultivated land area into N monitoring area units and numbers them in sequence, numbered n;

[0039] The preset route module is used to set inspection positioning points and calibration points in each monitoring area unit, and set the preset inspection route according to the position of each inspection positioning point;

[0040] Image acquisition drones are used to collect image information data of each monitoring area according to the preset inspection route;

[0041] The calibration drone is used to follow the image acquisition drone to the calibration points corresponding to each monitoring area unit;

[0042] A data acquisition module for collecting position information of the image acquisition drone and the calibration drone;

[0043] The analysis module includes an analysis unit and an identification unit; the analysis unit is used to analyze each image information data, obtain the quality assessment index of each image information data, and analyze and screen out the designated pictures of each monitoring area unit according to the quality assessment index of each image information data; the identification unit is a trained neural network model, which is used to identify the designated picture and determine whether there is any cultivated land occupation behavior in the monitoring area unit; if there is cultivated land occupation behavior, an early warning is issued, otherwise there is no cultivated land occupation behavior.

[0044] Beneficial effects of the present invention:

[0045] The present invention collects M image information data of the monitoring area unit by an image acquisition drone; analyzes the calibrated drone image in each image information data by an analysis unit to obtain a quality evaluation index of each image information data, and analyzes and screens out a specified picture according to the quality evaluation index of each image information data; ensures the quality of the specified picture, and finally identifies the specified image by an identification unit to determine whether there is cultivated land occupation in the monitoring area unit; if there is cultivated land occupation, an early warning is issued, and if not, the accuracy of the judgment result is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The present invention will be further described below in conjunction with the accompanying drawings.

[0047] Figure 1 A method flow chart of an embodiment of the present invention;

[0048] Figure 2 A system module framework diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] See also Figure 1 As shown, in one embodiment, a method for identifying and warning farmland occupation behavior based on a drone is provided, comprising the following steps:

[0051] S1: Divide the cultivated land area into N monitoring area units through the division module and number them in sequence, numbered n;

[0052] S2: Setting inspection positioning points and calibration points in each monitoring area unit through the preset route module, and setting a preset inspection route according to the position of each inspection positioning point;

[0053] S3: The image acquisition drone moves to the inspection positioning point of the monitoring area unit numbered n according to the preset inspection route, and the calibration drone follows the image acquisition drone to move toward the corresponding calibration point;

[0054] S4: obtaining the position information of the image acquisition drone and the calibration drone through the data acquisition module, obtaining a first judgment index by analyzing the position information of the image acquisition drone and the calibration drone, and judging whether image acquisition can be performed according to the first judgment index; if image acquisition can be performed, proceeding to step S5; otherwise, repeating step S4;

[0055] S5: using an image acquisition drone to collect M image information data of the monitoring area unit; using an analysis unit to analyze the calibrated drone image in each image information data, to obtain a quality evaluation index of each image information data, and to analyze and filter out a specified image according to the quality evaluation index of each image information data;

[0056] S6: Identify the designated image through the identification unit to determine whether there is farmland occupation in the monitoring area unit; if there is farmland occupation, issue an early warning; if there is no farmland occupation;

[0057] Through the above technical scheme, this embodiment divides the cultivated land area into N monitoring area units through the division module and numbers them in sequence, sets inspection positioning points and calibration points in each monitoring area unit through the preset route module, and sets a preset inspection route according to the position of each inspection positioning point; enables the image acquisition drone to collect images of the entire monitoring area unit at the inspection positioning point, and the calibration point is located at a preset position and preset distance of the inspection positioning point; the preset position and preset distance are preset values, which are obtained based on experience and are not described in detail here; the image acquisition drone moves to the inspection positioning point of the monitoring area unit numbered n according to the preset inspection route, and the calibration drone follows the image acquisition drone toward the corresponding calibration point; the image acquisition drone and the calibration drone are obtained through the data acquisition module. Position information, by analyzing the position information of the image acquisition drone and the calibration drone to obtain a first judgment index, and judging whether image acquisition can be performed according to the first judgment index; if image acquisition can be performed, the image acquisition drone collects M image information data of the monitoring area unit; the calibration drone image in each image information data is analyzed by the analysis unit to obtain the quality evaluation index of each image information data, and the specified image is analyzed and screened out according to the quality evaluation index of each image information data; the quality of the specified image is ensured, and finally the specified image is identified by the recognition unit to judge whether there is cultivated land occupation behavior in the monitoring area unit; if there is cultivated land occupation behavior, an early warning is issued, and if there is no cultivated land occupation behavior, the accuracy of the judgment result is improved;

[0058] It should be noted that the qualified image information data with the highest quality evaluation index is the designated picture.

[0059] As an implementation mode of the present invention, by formula:

[0060]

[0061] Calculate the first judgment index P n ;

[0062] Wherein, f(X) is the first judgment function. When X>0, f(X)=0; when X≤0, f(X)=1; (X A , Y A , Z A ) is the three-dimensional coordinate of the real-time position of the image acquisition drone; (X nx , Y nx , Z nx ) is the preset three-dimensional coordinate of the inspection positioning point of the nth monitoring area unit; (X B , Y B , Z B ) is the three-dimensional coordinate of the real-time position of the calibration drone; (X ny , Y ny , Z ny ) is the preset three-dimensional coordinate of the calibration point of the nth monitoring area unit; W 1 is the permissible inspection error value; W 2 is the allowable calibration error value;

[0063] Through the above technical solution, this embodiment The real-time position of the image acquisition drone and the real-time distance between the inspection location point; is the difference between the real-time position of the image acquisition drone and the real-time distance between the inspection positioning point and the allowable inspection error value; in the formula In the first judgment function f(X), X refers to when When , it means that the real-time position of the image acquisition drone and the real-time distance of the inspection positioning point exceed the allowable inspection error value; therefore, the image acquisition drone has not reached the inspection positioning point; when When , it indicates that the real-time position of the image acquisition UAV and the real-time distance of the inspection positioning point are within the allowable inspection error value; To calibrate the real-time position of the drone and the real-time distance of the calibration point; is the difference between the real-time position of the calibrated drone and the real-time distance of the calibration point and the allowable calibration error value; in the formula In the first judgment function f(X), X refers to when , it means that the real-time position of the calibration drone and the real-time distance of the calibration point exceed the allowable calibration error value; therefore, the calibration drone has not reached the calibration point; when When , it means that the real-time position of the calibration drone and the real-time distance of the calibration point are within the allowable calibration error value;

[0064]

[0065] It should be noted that the preset three-dimensional coordinates (X nx , Y nx , Z nx ), the preset three-dimensional coordinates (X ny , Y ny , Z ny ), allowable inspection error value W 1 And the allowable calibration error value W 2 It is a preset value obtained based on experience and will not be described in detail here.

[0066] As an implementation mode of the present invention, the process of determining whether image acquisition can be performed is as follows:

[0067] When P n =0, the image acquisition drone cannot perform image acquisition;

[0068] When P n =1, the image acquisition drone can perform image acquisition;

[0069] Through the above technical solution, this embodiment and When , it means that both the image acquisition UAV and the calibration UAV have arrived at the designated position of the monitoring area unit numbered n; therefore, the image acquisition UAV can perform image acquisition; otherwise, at least one of the image acquisition UAV and the calibration UAV has not arrived at the designated position, so the image acquisition UAV cannot perform image acquisition.

[0070] As an implementation mode of the present invention, by formula:

[0071] δ m =f[|S m -S 0 |-W S ]*f(Q 0 -Q m )*f(C 0 -C m )*f(R 0 -R m )

[0072] Calculate the qualified state coefficient δ of the mth image information data m ;

[0073] When δ m =0, the mth image information data is unqualified;

[0074] When δ m =1, the mth image information data is qualified;

[0075] Among them, S m is the area size of the calibration drone in the mth image information data; S 0 is the preset area size; W S Q is the preset area allowable error value; m is the clarity of the mth image information data; Q 0 is the preset definition; R 0 is the preset image similarity; R m is the image similarity of the calibration drone in the mth image information data; C 0 is the preset color accuracy; C m is the color accuracy of the calibration drone in the mth image information data;

[0076] Through the above technical solution, this embodiment |S m -S 0 | is the absolute value of the difference between the area size of the calibrated drone in the mth image information data and the preset area size; |S m -S 0 |-W S is the difference between the absolute value of the difference between the area size of the calibrated drone and the preset area size in the mth image information data and the preset area allowable error value; in the formula f[|S m -S 0 |-W S ], X in the first judgment function f(X) refers to |S m -S 0 |-W S ; When |S m -S 0 |-W S When ≤0, it means that the absolute value of the difference between the area size of the calibration drone in the mth image information data and the preset area size is within the range of the preset area allowable error value, so it means that the area size of the calibration drone in the mth image information data is qualified, f[|S m -S 0 |-W S ]=1; when |S m -S 0 |-W S>0, it means that the absolute value of the difference between the area size of the calibration drone in the mth image information data and the preset area size exceeds the preset area allowable error value, so it means that the area size of the calibration drone in the mth image information data is unqualified, f[|S m -S 0 |-W S ]=0; Q in this embodiment 0 -Q m is the difference between the preset definition and the definition of the mth image information data; in the formula f(Q 0 -Q m ), X in the first judgment function f(X) refers to Q m -Q 0 ; When Q 0 -Q m ≤0, the clarity of the mth image information data is not lower than the preset clarity, so the clarity of the mth image information data is qualified, f(Q 0 -Q m )=1; when Q 0 -Q m >0, the clarity of the mth image information data is lower than the preset clarity, so the clarity of the mth image information data is unqualified, f(Q 0 -Q m )=0;This embodiment C 0 -C m is the difference between the preset color accuracy and the color accuracy of the calibration drone in the mth image information data; in the formula f(C 0 -C m ), X in the first judgment function f(X) refers to C 0 -C m ; When C 0 -C m ≤0, the color accuracy of the mth image information data is not lower than the preset color accuracy, so the color accuracy of the mth image information data is qualified, f(C 0 -C m )=1; when C 0 -C m >0, the color accuracy of the mth image information data is lower than the preset color accuracy, so the color accuracy of the mth image information data is unqualified, f(C 0 -C m )=0;This embodiment R 0 -R m is the difference between the preset image similarity and the image similarity of the calibration drone in the mth image information data; in the formula f(R 0 -R m ), X in the first judgment function f(X) refers to R0 -R m ; When R 0 -R m ≤0, the image similarity of the mth image information data is not less than the preset image similarity, so the image similarity of the mth image information data is qualified, f(R 0 -R m )=1; when R 0 -R m >0, the image similarity of the mth image information data is lower than the preset image similarity, so the image similarity of the mth image information data is unqualified, f(R 0 -R m )=0; before working, the image acquisition drone moves to the inspection positioning point, the calibration drone moves and stays at the corresponding calibration point with a preset posture, and the image acquisition drone collects the preset image information data of the corresponding monitoring area unit with a preset focal length, and the preset area size S 0 The preset image information data is obtained; the preset image similarity is obtained by comparing the calibration drone image data in the mth image information data with the calibration drone image data in the preset image information data, and the image similarity R of the mth image information data m The higher the value, the closer the posture of the calibrated drone is to the preset posture. The color accuracy C of the calibrated drone in the mth image information data m The color of the calibration drone in the mth image information data is compared with the actual color of the drone part at the calibration drone shooting position in the preset image information data, which is the prior art and will not be described in detail here;

[0077] It should be noted that the preset area allowable error value W S , the preset definition Q of the mth image information data 0 , preset image similarity R 0 It is a preset value obtained based on experience and will not be described in detail here.

[0078] As an implementation mode of the present invention, by formula:

[0079]

[0080] Calculate the quality assessment index U of the mth image information data m

[0081] Among them, γ 1 is the first weight coefficient; γ 2 is the second weight coefficient; C 1 is the first preset constant; C 2 is the second preset constant;

[0082] Through the above technical solution, this embodiment uses the qualified state coefficient δ of the mth image information data m , so that the quality assessment index U of the unqualified m-th image information data m =0; Quality evaluation index of the qualified mth image information data The definition Q of the qualified mth image information data m The larger the value, the quality assessment index U of the mth image information data m The larger the value, the higher the color accuracy C of the qualified m-th image information data. m The larger the value, the quality assessment index U of the mth image information data m The larger the value, the better the image quality and the more accurate the assessed farmland occupation behavior.

[0083] It should be noted that the first weight coefficient γ 1 , the second weight coefficient γ 2 , the first preset constant C 1 and the second preset constant C 2 It is a preset value obtained based on experience and will not be described in detail here.

[0084] As an implementation mode of the present invention, after step S6, the shooting status index of the image acquisition drone is obtained by analyzing the quality evaluation index of each image information data; and according to the shooting status index, it is determined whether the image acquisition drone continues to perform the task according to the preset inspection route;

[0085] As an implementation mode of the present invention, by formula:

[0086]

[0087] Calculate the shooting state index T of the image acquisition drone state ;

[0088] Where D is the number of qualified states in the M image information data; k is the preset qualified rate; U 0 is the preset quality evaluation index; α 1 is the first weight coefficient of the state; α 2 is the second weight coefficient of the state; C 3 is the third preset constant; C 4 is the fourth preset constant; T 0 is the basic status index;

[0089] Through the above technical solution, this embodiment is the actual pass rate; is the difference between the actual qualified rate and the preset qualified rate; when When the actual qualified rate is greater than the preset qualified rate, the better the drone's shooting status is, the higher the drone's shooting status index T is. state The bigger; when When the actual qualified rate is less than the preset qualified rate, the worse the drone's shooting status is, the higher the drone's shooting status index T is. state The smaller; is the average quality assessment index of M image information data; is the difference between the average quality assessment index and the preset quality assessment index; when When the average quality evaluation index is greater than the preset quality evaluation index, the better the drone's shooting status, the higher the drone's shooting status index T state The bigger; when When the average quality evaluation index is less than the preset quality evaluation index, the worse the drone's shooting status is, the higher the drone's shooting status index T is. state The smaller;

[0090] It should be noted that the preset pass rate k and the preset quality assessment index U 0 、State first weight coefficient α 1 、State second weight coefficient α 2 , the third preset constant C 3 , the fourth preset constant C 4 and the basic state index T 0 It is a preset value obtained based on experience and will not be described in detail here.

[0091] As an implementation mode of the present invention, the shooting state index T state Compared with the first preset value Z 1 and the second preset comparison value Z 1 Make a comparison;

[0092] When T state <Z 1 When the image acquisition module is seriously abnormal, an early warning is issued and the system is immediately returned for maintenance;

[0093] When Z 1 ≤T state <T 0 When the image acquisition module is slightly abnormal, an early warning is issued and the task is continued according to the preset inspection route before maintenance;

[0094] When T 0 ≤T state <Z 2 The image acquisition module is in normal state and continues to perform tasks according to the preset inspection route without warning.

[0095] When Z 2 ≤Tstate The image acquisition module is in good condition and performs tasks according to the preset inspection route without early warning.

[0096] It should be noted that the first preset contrast value Z 1 and the second preset comparison value Z 1 It is a preset value obtained based on experience and will not be described in detail here.

[0097] See also Figure 2 As shown, a farmland occupation behavior recognition and early warning system based on drones includes:

[0098] The division module divides the cultivated land area into N monitoring area units and numbers them in sequence, numbered n;

[0099] The preset route module is used to set inspection positioning points and calibration points in each monitoring area unit, and set the preset inspection route according to the position of each inspection positioning point;

[0100] Image acquisition drones are used to collect image information data of each monitoring area according to the preset inspection route;

[0101] The calibration drone is used to follow the image acquisition drone to the calibration points corresponding to each monitoring area unit;

[0102] A data acquisition module for collecting position information of the image acquisition drone and the calibration drone;

[0103] The analysis module includes an analysis unit and an identification unit; the analysis unit is used to analyze each image information data, obtain the quality assessment index of each image information data, and analyze and screen out the designated pictures of each monitoring area unit according to the quality assessment index of each image information data; the identification unit is a trained neural network model, which is used to identify the designated picture and determine whether there is farmland occupation in the monitoring area unit; if there is farmland occupation, an early warning is issued, otherwise there is no farmland occupation. The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equal changes and improvements made according to the scope of application of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for identifying and warning farmland occupation behavior based on drones, characterized in that: The following steps are involved: S1: Divide the cultivated land area into N monitoring area units through the division module and number them in sequence, numbered n; S2: Setting inspection positioning points and calibration points in each monitoring area unit through the preset route module, and setting a preset inspection route according to the position of each inspection positioning point; S3: The image acquisition drone moves to the inspection positioning point of the monitoring area unit numbered n according to the preset inspection route, and the calibration drone follows the image acquisition drone to move toward the corresponding calibration point; S4: obtaining the position information of the image acquisition drone and the calibration drone through the data acquisition module, obtaining a first judgment index by analyzing the position information of the image acquisition drone and the calibration drone, and judging whether image acquisition can be performed according to the first judgment index; if image acquisition can be performed, proceeding to step S5; Otherwise, repeat step S4; S5: using an image acquisition drone to collect M image information data of the monitoring area unit; using an analysis unit to analyze the calibrated drone image in each image information data, to obtain a quality evaluation index of each image information data, and to analyze and filter out a specified image according to the quality evaluation index of each image information data; S6: Identify the designated image through the identification unit to determine whether there is farmland occupation behavior in the monitoring area unit; If there is any occupation of cultivated land, an early warning will be issued; if there is no occupation of cultivated land, an early warning will be issued.

2. The method for identifying and warning farmland occupation behavior based on drones according to claim 1 is characterized in that: By formula: Calculate the first judgment index P n ; Wherein, f(X) is the first judgment function. When X>0, f(X)=0; when X≤0, f(X)=1; (X A , Y A , Z A ) is the three-dimensional coordinate of the real-time position of the image acquisition drone; (X nx , Y nx , Z nx ) is the preset three-dimensional coordinate of the inspection positioning point of the nth monitoring area unit; (X B , Y B , Z B ) is the three-dimensional coordinate of the real-time position of the calibration drone; (X ny , Y ny , Z ny ) is the preset three-dimensional coordinate of the calibration point of the nth monitoring area unit; W1 is the allowable inspection error value; W2 is the allowable calibration error value.

3. The method for identifying and warning farmland occupation behavior based on drones according to claim 2 is characterized in that: The process of determining whether image acquisition can be performed is as follows: When P n =0, the image acquisition drone cannot perform image acquisition; When P n =1, the image acquisition drone can perform image acquisition.

4. The method for identifying and warning farmland occupation behavior based on drones according to claim 3 is characterized in that: By formula: δ m =f[|S m -S0|-W S ]*f(Q0-Q m )*f(C0-C m )*f(R0-R m ) Calculate the qualified state coefficient δ of the mth image information data m ; When δ m =0, the mth image information data is unqualified; When δ m =1, the mth image information data is qualified.

5. The method for identifying and warning farmland occupation behavior based on drones according to claim 4 is characterized in that: By formula: Calculate the quality assessment index U of the mth image information data m Among them, γ1 is the first weight coefficient; γ2 is the second weight coefficient; C1 is the first preset constant; C2 is the second preset constant.

6. The method for identifying and warning farmland occupation behavior based on drones according to claim 5 is characterized in that: After step S6, the shooting status index of the image acquisition drone is obtained by analyzing the quality evaluation index of each image information data; and according to the shooting status index, it is determined whether the image acquisition drone continues to perform the task according to the preset inspection route.

7. The method for identifying and warning farmland occupation behavior based on drones according to claim 6 is characterized in that: By formula: Calculate the shooting state index T of the image acquisition drone state ; Among them, D is the number of qualified states in M ​​image information data; k is the preset qualified rate; U0 is the preset quality assessment index; α1 is the first weight coefficient of the state; α2 is the second weight coefficient of the state; C3 is the third preset constant; C4 is the fourth preset constant; T0 is the basic state index.

8. The method for identifying and warning farmland occupation behavior based on drones according to claim 7 is characterized in that: The shooting status index T state Compare with the first preset comparison value Z1 and the second preset comparison value Z1; When T state <When it is less than Z1, the image acquisition module has a serious abnormality, issues a warning, and immediately returns for maintenance; When Z1 ≤ T state When < T0, the image acquisition module has a slight anomaly, issues a warning, continues to execute tasks according to the preset inspection route, and then performs maintenance; When T0 ≤ T state <When Z2; the image acquisition module is in normal state, continue to execute tasks according to the preset inspection route without warning; When Z2≤T state When the image acquisition module is in good condition, it performs tasks according to the preset inspection route without early warning.

9. A farmland occupation behavior recognition and early warning system based on drones, applicable to a farmland occupation behavior recognition and early warning method based on drones as claimed in any one of claims 1 to 8, characterized in that: The identification and early warning system comprises: The division module divides the cultivated land area into N monitoring area units and numbers them in sequence, numbered n; The preset route module is used to set inspection positioning points and calibration points in each monitoring area unit, and set the preset inspection route according to the position of each inspection positioning point; Image acquisition drones are used to collect image information data of each monitoring area according to the preset inspection route; The calibration drone is used to follow the image acquisition drone to the calibration points corresponding to each monitoring area unit; A data acquisition module for collecting position information of the image acquisition drone and the calibration drone; The analysis module includes an analysis unit and an identification unit; the analysis unit is used to analyze each image information data, obtain the quality assessment index of each image information data, and analyze and screen out the designated pictures of each monitoring area unit according to the quality assessment index of each image information data; the identification unit is a trained neural network model, which is used to identify the designated picture and determine whether there is any cultivated land occupation behavior in the monitoring area unit; if there is cultivated land occupation behavior, an early warning is issued, otherwise there is no cultivated land occupation behavior.

Images