An intelligent park inspection system based on drone technology

Abstract

The present invention belongs to the field of park inspection systems, and discloses an intelligent park inspection system based on drone technology, including a path planning device, a flight control device, a drone and a data processing device; the path planning device includes a shooting area acquisition unit, a patrol area acquisition unit and a path planning unit; the shooting area acquisition unit is used to determine the shooting area corresponding to all security cameras; the patrol area acquisition unit is used to acquire the patrol area; the path planning unit is used to acquire patrol points based on the patrol area, and adopts a preset path planning algorithm to acquire a flight path passing through all patrol points; the flight control device is used to control the drone to fly, shoot and transmit data along the flight path; the data processing device is used to identify the patrol image and obtain the patrol result. The present invention can reduce the number of patrol points, so that the drone can patrol the area that the security camera fails to capture more timely.

Description

Technical Field

[0001] The present invention relates to the field of park inspection systems, and in particular to an intelligent park inspection system based on unmanned aerial vehicle technology. Background Art

[0002] In the prior art, when using drones to patrol a park, the park is usually directly divided into multiple patrol areas, and then the flight path is obtained based on a path planning algorithm such as the shortest path algorithm. However, this way of dividing the patrol area does not take into account the overlap between the patrol area of ​​the drone and the shooting area of ​​the security camera of the security system. This will lead to the situation where some overlapping areas are repeatedly patrolled by security cameras and drones, so that when the drone is patrolling, it wastes a lot of time patrolling in the overlapping areas, and fails to patrol the areas that really need drone patrols (areas that the security cameras cannot capture) in time. Summary of the invention

[0003] The purpose of the present invention is to disclose a campus intelligent inspection system based on UAV technology, and to solve the problem of how to improve the flight path generation process so that the UAV can inspect the areas that need to be inspected by the UAV in a timely manner.

[0004] In order to achieve the above object, the present invention provides the following technical solutions:

[0005] The present invention provides a park intelligent inspection system based on UAV technology, including a path planning device, a flight control device, a UAV and a data processing device;

[0006] The path planning device includes a shooting area acquisition unit, a patrol area acquisition unit and a path planning unit;

[0007] The shooting area acquisition unit is used to determine the shooting area corresponding to the shooting range of all security cameras on the park plan based on the shooting range of each security camera in the park;

[0008] The patrol area acquisition unit is used to perform the following calculations:

[0009] Get the minimum enclosing rectangle of the park plan;

[0010] According to the shooting range of the drone, the minimum enclosing rectangle is partitioned into multiple inspection areas of equal area;

[0011] The following judgments are made for each patrol area:

[0012] The first step is to determine whether there is no overlapping area between the inspection area and the park floor plan. If so, delete the inspection area; if so, proceed to the second step;

[0013] The second step is to determine whether the area of ​​the overlapping region between the inspection area and the shooting area is the same as the area of ​​the inspection area. If so, the inspection area is deleted;

[0014] The path planning unit is used to obtain inspection points based on the inspection area, and use a preset path planning algorithm to obtain a flight path passing through all inspection points;

[0015] The flight control device is used to control the UAV to fly along the flight path, to control the UAV to take pictures after the UAV arrives at the inspection point to obtain the inspection image, and to control the UAV to send the inspection image to the data processing device;

[0016] The data processing device is used to identify the inspection image and obtain the inspection result.

[0017] Preferably, the security camera includes a security camera that can rotate along the horizontal direction and a security camera that cannot rotate.

[0018] Preferably, for a security camera that can rotate in the horizontal direction, the process of obtaining the corresponding shooting range is as follows:

[0019] θ1 represents the angle between the main optical axis of the security camera and the vertical direction; h represents the height of the security camera;

[0020] Then the calculation formula for the shooting distance d1 of the security camera in the horizontal direction is:

[0021] d1=h×tanθ1;

[0022] On the park plan, with the horizontal right direction as 0 degrees, when the security camera rotates, the minimum and maximum values ​​of the angles formed between the line segment corresponding to the vertical projection of the main optical axis of the security camera on the park plan and the 0 degree direction are represented as θ2 and θ3 respectively;

[0023] The shooting range of the security camera is centered on the coordinates of the security camera on the park plan, with a radius of d1 and an angle range of The fan-shaped area, θ field is the horizontal field of view of the security camera;

[0024] For security cameras that cannot rotate, d2 is used to represent the shooting distance in the horizontal direction. On the campus plan, the two angles formed by the two sides corresponding to the horizontal field of view of the security camera and the direction of 0 degrees are represented as θ4 and θ5 respectively;

[0025] The shooting range corresponding to the security camera is a sector-shaped area with a radius of d2 and an angle range of [θ4, θ5] centered on the coordinates of the security camera on the park plan.

[0026] Preferably, based on the shooting range of each security camera in the park, the shooting area corresponding to the shooting range of all security cameras on the park plan is determined, including:

[0027] Get the area UA corresponding to the union of all shooting areas;

[0028] On the park planar image, the area corresponding to the area UA is used as the shooting area.

[0029] Preferably, the minimum circumscribed rectangle is partitioned according to the shooting range of the drone, and the minimum circumscribed rectangle is divided into a plurality of patrol areas of the same area, including:

[0030] L1 and L2 are used to represent the length and width of the shooting range of a single shot by the drone on the park plan.

[0031] K1 and K2 are used to represent the length and width of the minimum circumscribed rectangle respectively;

[0032] The following operations are performed on the minimum enclosing rectangle:

[0033] Determine whether the quotient of L1 divided by K1 is an integer. If not, extend the length of the minimum circumscribed rectangle to L1-(L1 mod K1).

[0034] Determine whether the quotient of L2 divided by K2 is an integer. If not, extend the width of the minimum circumscribed rectangle by the length of L2-(L2 mod K2).

[0035] After performing the above calculation on the minimum enclosing rectangle, the number of inspection areas in the length direction is The number of inspection zones in the width direction is

[0036] The area of ​​the inspection area is L1×L2.

[0037] Preferably, controlling the drone to shoot and obtain inspection images includes:

[0038] Calculate the inspection coefficient of each inspection point separately;

[0039] Calculate the dwell time T based on the inspection coefficient;

[0040] The drone is controlled to stay at the inspection point for a period of time T. During the stay, a patrol image is taken every S seconds.

[0041] Preferably, the calculation process of the patrol coefficient is as follows:

[0042]

[0043] inspcoef b Indicates the inspection coefficient of inspection point b, eventN b Indicates that the area with radius D and inspection point b as the center is in the front inspN b The number of events of the preset type found in each inspection, overlap b It represents the area of ​​the part overlapping with the shooting area in the area with radius D and inspection point b as the center. δ1 represents the set event weight, and δ2 represents the set area weight.

[0044] Preferably, the dwell time T is calculated based on the inspection coefficient, including:

[0045] For patrol point b, the corresponding stay time at patrol point b is T b , T b The calculation formula is:

[0046]

[0047] inspcoef std Indicates the set inspection coefficient, T std Indicates the set time length.

[0048] Preferably, the preset types of events include illegal parking events, unclear road marking events, damage to park public facilities events and crowd gathering events.

[0049] Preferably, a patrol image is captured every S seconds, including:

[0050] A patrol image is taken vertically downward every S seconds.

[0051] Beneficial effects:

[0052] Compared with the prior art, the present invention avoids the shooting area of ​​the security camera when planning the flight path of the drone, thereby effectively reducing the range that the drone needs to patrol, and then reducing the number of patrol points, so that the drone can patrol the areas that the security camera cannot capture more timely. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0054] Figure 1 A schematic diagram of an intelligent park inspection system based on drone technology of the present invention.

[0055] Figure 2 A schematic diagram of the process of controlling a drone to take photos according to the present invention. DETAILED DESCRIPTION

[0056] 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 parameters of the protection of the present invention.

[0057] like Figure 1 In one embodiment shown, the present invention provides a park intelligent inspection system based on UAV technology, including a path planning device, a flight control device, a UAV and a data processing device;

[0058] The path planning device includes a shooting area acquisition unit, a patrol area acquisition unit and a path planning unit;

[0059] The shooting area acquisition unit is used to determine the shooting area corresponding to the shooting range of all security cameras on the park plan based on the shooting range of each security camera in the park;

[0060] The patrol area acquisition unit is used to perform the following calculations:

[0061] Get the minimum enclosing rectangle of the park plan;

[0062] According to the shooting range of the drone, the minimum enclosing rectangle is partitioned into multiple inspection areas of equal area;

[0063] The following judgments are made for each patrol area:

[0064] The first step is to determine whether there is no overlapping area between the inspection area and the park floor plan. If so, delete the inspection area; if so, proceed to the second step;

[0065] The second step is to determine whether the area of ​​the overlapping region between the inspection area and the shooting area is the same as the area of ​​the inspection area. If so, the inspection area is deleted;

[0066] The path planning unit is used to obtain inspection points based on the inspection area, and use a preset path planning algorithm to obtain a flight path passing through all inspection points;

[0067] The flight control device is used to control the UAV to fly along the flight path, to control the UAV to take pictures after the UAV arrives at the inspection point to obtain the inspection image, and to control the UAV to send the inspection image to the data processing device;

[0068] The data processing device is used to identify the inspection image and obtain the inspection result.

[0069] In the above-mentioned path planning process, since the minimum circumscribed rectangle is obtained, when the shape of the park is not a rectangle, some areas that do not belong to the park will be introduced into the minimum circumscribed rectangle. Therefore, after obtaining the inspection area, the present invention determines whether there is no overlapping area between the inspection area and the park plan. When there is no overlapping area, it means that the inspection area is not within the scope of the park. Therefore, the present invention deletes the inspection area, thereby avoiding invalid inspections by drones. In addition, the present invention also deletes the inspection area that completely falls into the shooting area. In this way, it can avoid repeated inspections of the shooting area by drones, which is conducive to improving the timely inspection of the range outside the shooting area by drones and improving the effectiveness of drone inspections.

[0070] Specifically, the path planning device may be a local computer or a cloud server.

[0071] Specifically, the park floor plan is a map that only includes various landforms within the park.

[0072] Preferably, the security camera includes a security camera that can rotate along the horizontal direction and a security camera that cannot rotate.

[0073] In order to expand the shooting range of security cameras, in the prior art, many cameras can rotate in the horizontal direction. Therefore, the present invention adopts different determination methods to determine the shooting range of security cameras that can rotate and security cameras that cannot rotate, so as to more accurately determine the shooting area.

[0074] Preferably, for a security camera that can rotate in the horizontal direction, the process of obtaining the corresponding shooting range is as follows:

[0075] θ1 represents the angle between the main optical axis of the security camera and the vertical direction; h represents the height of the security camera;

[0076] Since security cameras are used to shoot contents on the ground, they are usually set to shoot the ground diagonally downward, so that the main optical axis of the security camera forms an angle with the vertical direction, that is, the direction perpendicular to the ground plane.

[0077] Then the calculation formula for the shooting distance d1 of the security camera in the horizontal direction is:

[0078] d1=h×tanθ1;

[0079] Based on the angle and height, the shooting distance of the security camera on the ground can be calculated.

[0080] On the park plan, with the horizontal right direction as 0 degrees, when the security camera rotates, the minimum and maximum values ​​of the angles formed between the line segment corresponding to the vertical projection of the main optical axis of the security camera on the park plan and the 0 degree direction are represented as θ2 and θ3 respectively;

[0081] After setting the direction of 0 degrees, the shooting range of the security camera can be more clearly indicated. The present invention takes the horizontal right direction as the direction of 0 degrees, and in actual use, the direction of 0 degrees can also be set according to actual needs.

[0082] Preferably, in the process of determining θ2 and θ3, what is obtained is the angle formed with the line segment in the clockwise direction.

[0083] In order to more accurately represent the shooting range, the present invention uses the projection of the main optical axis on the park plan, that is, the ground, as an indication of the shooting distance in the horizontal direction of the security camera during operation.

[0084] When the main optical axis is tilted downward, the straight line where the main optical axis is located will intersect the ground at a point, and the projection of the security camera on the ground is also a point. Connecting these two points, we can get the line segment corresponding to the vertical projection of the main optical axis of the security camera on the ground. The length of this line segment is the length of the shooting radius. This line segment will form a fan-shaped area on the ground during the rotation of the camera. The angle corresponding to this fan-shaped area plus the angle of the horizontal field of view is the complete angle of the shooting range.

[0085] The shooting range of the security camera is centered on the coordinates of the security camera on the park plan, with a radius of d1 and an angle range of The fan-shaped area, θ field is the horizontal field of view of the security camera;

[0086] For security cameras that cannot rotate, d2 is used to represent the shooting distance in the horizontal direction. On the campus plan, the two angles formed by the two sides corresponding to the horizontal field of view of the security camera and the direction of 0 degrees are represented as θ4 and θ5 respectively;

[0087] The shooting range corresponding to the security camera is a sector-shaped area with a radius of d2 and an angle range of [θ4, θ5] centered on the coordinates of the security camera on the park plan.

[0088] When the security camera cannot rotate, the shooting range of the security camera will still form a fan-shaped shooting area on the ground. The two straight lines forming the horizontal field of view will form corresponding angles with the 0 degree direction.

[0089] Preferably, based on the shooting range of each security camera in the park, the shooting area corresponding to the shooting range of all security cameras on the park plan is determined, including:

[0090] Get the area UA corresponding to the union of all shooting areas;

[0091] On the park planar image, the area corresponding to the area UA is used as the shooting area.

[0092] Since different security cameras may have overlapping shooting ranges, the present invention can avoid repeated statistics of overlapping shooting ranges by obtaining a union, and can obtain an accurate shooting area.

[0093] Preferably, the minimum circumscribed rectangle is partitioned according to the shooting range of the drone, and the minimum circumscribed rectangle is divided into a plurality of patrol areas of the same area, including:

[0094] L1 and L2 are used to represent the length and width of the shooting range of a single shot by the drone on the park plan.

[0095] The images taken by current cameras are generally rectangular. Therefore, after determining the shooting height and focal length of the drone, the length and width of the area actually corresponding to the image taken on the floor plan of the park can be determined.

[0096] Preferably, the shooting height of the drone is equal to the maximum height of the buildings in the park plus the vertical safe shooting distance of the drone. The shooting distance can be 5 meters.

[0097] K1 and K2 represent the length and width of the minimum circumscribed rectangle respectively; the length and width here are the values ​​in the park plan;

[0098] The following operations are performed on the minimum enclosing rectangle:

[0099] Determine whether the quotient of L1 divided by K1 is an integer. If not, extend the length of the minimum circumscribed rectangle to L1-(L1 mod K1).

[0100] Determine whether the quotient of L2 divided by K2 is an integer. If not, extend the width of the minimum circumscribed rectangle by the length of L2-(L2 mod K2).

[0101] After performing the above calculation on the minimum enclosing rectangle, the number of inspection areas in the length direction is The number of inspection zones in the width direction is

[0102] The area of ​​the inspection area is L1×L2.

[0103] The area here is the area in the park plan. Dividing L1 and L2 by the scale will give the length in real space.

[0104] Preferably, if Figure 2 As shown, the drone is controlled to shoot and obtain patrol images, including:

[0105] Calculate the inspection coefficient of each inspection point separately;

[0106] Calculate the dwell time T based on the inspection coefficient;

[0107] The drone is controlled to stay at the inspection point for a period of time T. During the stay, a patrol image is taken every S seconds.

[0108] Since the number of times a preset type of event is encountered is not the same in different patrol areas, in order to improve the effectiveness of patrols, the present invention calculates the dwell time based on the patrol coefficient, so that more patrol images can be captured for areas with a higher probability of the occurrence of the preset type of event.

[0109] Preferably, the value of S is 30 seconds, which can be adjusted according to actual needs.

[0110] Preferably, the calculation process of the patrol coefficient is as follows:

[0111]

[0112] inspcoef b Indicates the inspection coefficient of inspection point b, eventN b Indicates that the area with radius D and inspection point b as the center is in the front inspN b The number of events of the preset type found in each inspection, overlap bIt represents the area of ​​the part overlapping with the shooting area in the area with radius D and inspection point b as the center. δ1 represents the set event weight, and δ2 represents the set area weight.

[0113] When calculating the patrol coefficient, in addition to considering the number of times the preset type of events are discovered, the overlapping area of ​​the area with a radius of D and the shooting area with the patrol point as the center is also considered. The more times and the smaller the overlapping area, the larger the patrol coefficient, which makes the calculated stay time longer, which is conducive to more timely detection of abnormal situations in the park.

[0114] eventN b The calculation rule is as follows: eventN b The initial value is 0, in total inspN b During a patrol, if a patrol finds that there is at least one event of the preset type whose location is within the patrol area, eventN b The value of is added by 1, that is, in the process of a patrol, as long as there are events of the preset type, no matter how many events there are, only eventN b The value of eventN is increased by 1. Therefore, eventN b The maximum value is equal to inspN b .

[0115] Preferably, inspN b The value of is 10.

[0116] Preferably, the total number of inspections is less than inspN b The patrol coefficient is set to 1.

[0117] Preferably, the event weight is set to 0.7, and the area weight is set to 0.3.

[0118] Preferably, the value of D is scale indicates the scale of the park plan.

[0119] Preferably, the dwell time T is calculated based on the inspection coefficient, including:

[0120] For patrol point b, the corresponding stay time at patrol point b is T b , T b The calculation formula is:

[0121]

[0122] inspcoef std Indicates the set inspection coefficient, T std Indicates the set time length.

[0123] Tb with inspcoef b Positive correlation, inspcoef b The larger the T b The larger the value, the more inspection time can be set for time zones where preset types of accidents are more likely to occur.

[0124] Preferably, the patrol coefficient is set to 1.

[0125] Preferably, the set time length is 10 minutes.

[0126] Preferably, the preset types of events include illegal parking events, unclear road marking events, damage to park public facilities events and crowd gathering events.

[0127] Preferably, a patrol image is captured every S seconds, including:

[0128] A patrol image is taken vertically downward every S seconds.

[0129] Preferably, obtaining inspection points based on the inspection area includes:

[0130] If there is no overlapping area between the inspection area and the shooting area, the center of the inspection area is used as the inspection point;

[0131] If there is an area in the patrol area that overlaps with the shooting area, the patrol point is determined in the following way;

[0132] Acquire an area Z in the inspection area except for the area overlapping with the shooting area;

[0133] The center of area Z is used as the inspection point.

[0134] In the process of obtaining the inspection points, the present invention does not directly use the center of all inspection areas as the inspection points, because if there is a large overlapping area between the inspection area and the shooting area, the effectiveness of the drone's inspection at the inspection point will be reduced, because most areas can be monitored by security cameras. Therefore, by obtaining area Z, the present invention can make the inspection point fall into the non-overlapping area, thereby improving the effectiveness of the inspection.

[0135] Preferably, a preset path planning algorithm is used to obtain a flight path passing through all inspection points, including:

[0136] Calculate the path planning coefficient for each inspection point respectively;

[0137] Obtain a flight path passing through all inspection points based on the path planning coefficients.

[0138] Preferably, the process of obtaining the path planning coefficient is as follows:

[0139] Store all inspection points into the set patU;

[0140] For patrol point b, the path planning coefficient plancoef corresponding to patrol point b b The calculation formula is:

[0141]

[0142] inspcoef i Represents the inspection coefficient of inspection point i.

[0143] Specifically, the path planning coefficient is obtained by the ratio of the inspection coefficient to the sum of the inspection coefficients of all inspection points. The larger the path planning coefficient is, the more priority the inspection point needs to be inspected.

[0144] Preferably, obtaining a flight path passing through all inspection points based on the path planning coefficients includes:

[0145] Generate all flight paths starting from each inspection point;

[0146] Calculate the inspection event coefficient for each flight path separately:

[0147]

[0148] insevecoef k represents the inspection event coefficient of flight path k, npatU represents the total number of inspection points, pathcoef j represents the inspection event value of the jth inspection point in the flight path k, plancoef j Represents the path planning coefficient of the jth inspection point in the inspection path k; pathcoef j ∈pathcoefU, pathcoefU contains npatU patrol event values. pathcoefU is an ordered set. The patrol event values ​​in pathcoefU are sorted from large to small. The first patrol event value in pathcoefU has the largest value, and the npatUth patrol event value has the smallest value. The difference between two adjacent patrol event values ​​in pathcoefU is 1.

[0149] Calculate the cumulative flight distance for each flight path separately:

[0150]

[0151] flidist k represents the cumulative flight distance of flight path k, nodedist j,j+1represents the distance between the jth inspection point and the j+1th inspection point in the flight path k;

[0152] Calculate the competitive value of each flight path separately:

[0153]

[0154] The flight path with the largest competition value is taken as the final flight path.

[0155] The flight path planning process of the present invention is different from the existing commonly used shortest path planning method. In addition to considering the distance, the present invention also considers the timeliness of the inspection of different inspection points, so that in the final flight path, the inspection points with large path planning coefficients are inspected with priority. The smaller the total flight distance, the greater the competition value of the flight path. This is conducive to making the total flight distance as short as possible while giving priority to inspecting the inspection points with large path planning coefficients as much as possible, so that the drone of the present invention can inspect the areas that need to be inspected by the drone in a timely manner.

[0156] When calculating the inspection event coefficient, in the flight path, the later the inspection point is in the inspection order, the smaller the corresponding inspection event value is, so that the earlier the inspection point with a large path planning coefficient is inspected, the greater the competition value of the flight path is. In addition, the present invention does not directly generate the flight path according to the order of path planning coefficients from large to small, because the total flight distance of such a flight path may be very large, which leads to a relatively low actual inspection timeliness.

[0157] Preferably, all flight paths are generated starting from each inspection point, including:

[0158] The flight path is generated by permutation and combination. The first inspection point in the flight path has npatU possibilities, the second inspection point has npatU-1 possibilities, and so on. flight paths.

[0159] Preferably, the minimum and maximum values ​​of the patrol event values ​​in pathcoefU are 1 and npatU, respectively.

[0160] Preferably, the inspection image is identified to obtain the inspection result, including:

[0161] The inspection images are recognized using a pre-trained neural network recognition model to obtain the inspection results.

[0162] The inspection result includes the type of events present in the inspection image.

[0163] When planning the flight path of the drone, the present invention avoids the shooting area of ​​the security camera, thereby effectively reducing the range that the drone needs to patrol, and further reducing the number of patrol points, so that the drone can patrol the areas that the security camera fails to capture more promptly.

[0164] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0165] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A park intelligent inspection system based on drone technology, characterized in that: It includes a path planning device, a flight control device, a drone and a data processing device; The path planning device includes a shooting area acquisition unit, a patrol area acquisition unit and a path planning unit; The shooting area acquisition unit is used to determine the shooting area corresponding to the shooting range of all security cameras on the park plan based on the shooting range of each security camera in the park; The patrol area acquisition unit is used to perform the following calculations: Get the minimum enclosing rectangle of the park plan; According to the shooting range of the drone, the minimum enclosing rectangle is partitioned into multiple inspection areas of equal area; The following judgments are made for each patrol area: The first step is to determine whether there is no overlapping area between the inspection area and the park floor plan. If so, delete the inspection area; if so, proceed to the second step; The second step is to determine whether the area of ​​the overlapping region between the inspection area and the shooting area is the same as the area of ​​the inspection area. If so, the inspection area is deleted; The path planning unit is used to obtain inspection points based on the inspection area, and use a preset path planning algorithm to obtain a flight path passing through all inspection points; The flight control device is used to control the UAV to fly along the flight path, to control the UAV to take pictures after the UAV arrives at the inspection point to obtain the inspection image, and to control the UAV to send the inspection image to the data processing device; The data processing device is used to identify the inspection image and obtain the inspection result; Control the drone to shoot and obtain inspection images, including: Calculate the inspection coefficient of each inspection point separately; Calculate the dwell time T based on the inspection coefficient; Control the drone to stay at the inspection point for a period of time T. During the stay, take an inspection image every S seconds. The calculation process of the inspection coefficient is as follows: inspcoef b Indicates the inspection coefficient of inspection point b, eventN b Indicates that the area with radius D and inspection point b as the center is in the front inspN b The number of events of the preset type found in each inspection, overlap b It represents the area of ​​the part overlapping with the shooting area in the area with radius D and inspection point b as the center. δ1 represents the set event weight, and δ2 represents the set area weight. The dwell time T is calculated based on the inspection coefficient, including: For patrol point b, the corresponding stay time at patrol point b is T b , T b The calculation formula is: inspcoef std Indicates the set inspection coefficient, T std Indicates the set time length.

2. According to claim 1, the intelligent park inspection system based on drone technology is characterized in that: Security cameras include those that can rotate horizontally and those that cannot rotate.

3. According to claim 2, the intelligent park inspection system based on drone technology is characterized in that: For a security camera that can rotate horizontally, the process of obtaining the corresponding shooting range is as follows: θ1 represents the angle between the main optical axis of the security camera and the vertical direction; h represents the height of the security camera; Then the calculation formula for the shooting distance d1 of the security camera in the horizontal direction is: d1=h×tanθ1; On the park plan, with the horizontal right direction as 0 degrees, when the security camera rotates, the minimum and maximum values ​​of the angles formed between the line segment corresponding to the vertical projection of the main optical axis of the security camera on the park plan and the 0 degree direction are represented as θ2 and θ3 respectively; The shooting range of the security camera is centered on the coordinates of the security camera on the park plan, with a radius of d1 and an angle range of The fan-shaped area, θ field is the horizontal field of view of the security camera; For security cameras that cannot rotate, d2 is used to represent the shooting distance in the horizontal direction. On the campus plan, the two angles formed by the two sides corresponding to the horizontal field of view of the security camera and the direction of 0 degrees are represented as θ4 and θ5 respectively; The shooting range corresponding to the security camera is a sector-shaped area with a radius of d2 and an angle range of [θ4, θ5] centered on the coordinates of the security camera on the park plan.

4. According to claim 3, the intelligent park inspection system based on drone technology is characterized in that: Based on the shooting range of each security camera in the park, determine the shooting area corresponding to the shooting range of all security cameras on the park plan, including: Get the area UA corresponding to the union of all shooting areas; On the park planar image, the area corresponding to the area UA is used as the shooting area.

5. The intelligent park inspection system based on drone technology according to claim 1 is characterized in that: According to the shooting range of the drone, the minimum enclosing rectangle is divided into multiple inspection areas of equal area, including: L1 and L2 are used to represent the length and width of the shooting range of a single shot by the drone on the park plan. K1 and K2 are used to represent the length and width of the minimum circumscribed rectangle respectively; The following operations are performed on the minimum enclosing rectangle: Determine whether the quotient of K1 divided by L1 is an integer. If not, extend the length of the minimum circumscribed rectangle to L1-(K1 mod L1). Determine whether the quotient of K2 divided by L2 is an integer. If not, extend the width of the minimum circumscribed rectangle by the length of L2-(K2 mod L2). After performing the above calculation on the minimum enclosing rectangle, the number of inspection areas in the length direction is The number of inspection zones in the width direction is The area of ​​the inspection area is L1×L2.

6. The intelligent park inspection system based on drone technology according to claim 1 is characterized in that: The preset types of events include illegal parking, unclear road markings, damage to park public facilities and crowd gathering.

7. The intelligent park inspection system based on drone technology according to claim 1 is characterized in that: A patrol image is taken every S seconds, including: A patrol image is taken vertically downward every S seconds.

Images