Unmanned aerial vehicle visual inspection alignment method

Abstract

The invention discloses an unmanned aerial vehicle visual inspection alignment method. The method comprises the steps of image annular division, annular area identification, peripheral feature point positioning, center feature point positioning and lifting alignment positioning. The beneficial effects of the invention are that the detection of the initial image of the shooting area is combined with the detection of the current situation of the shooting area, the precise detection of the detection shooting area is realized, and the initial image of the detection shooting area can be used as a reference, so that the unmanned aerial vehicle can carry out short-distance flight adjustment. The current situation picture corresponding to the initial image can be shot, so that the detection result is more accurate; when the unmanned aerial vehicle body is aligned, distance measurement is performed on the feature points arranged on the initial image of the detection shooting area, and the lift-off height of the corner of the detection shooting area shot by the unmanned aerial vehicle body is calculated in combination with an equal proportion formula, so that the alignment shooting detection of the detection shooting area is realized.

Description

Technical Field

[0001] The invention relates to unmanned aerial vehicle visual detection, in particular to an unmanned aerial vehicle visual detection alignment method, and belongs to the technical field of unmanned aerial vehicle detection. Background Art

[0002] A drone is an unmanned aircraft controlled by a radio remote control device and a self-contained program control device. A drone is mainly composed of a flight component, a camera component, and a flight control component. Therefore, when inspecting some more dangerous areas or areas that are inconvenient for people to pass through, the drone's flight advantages are often used for drone visual inspection.

[0003] For example, in the prior art, a method for detecting bridge surface defects using a drone based on the fusion of laser and machine vision is disclosed in the announcement number CN114778551A. The method detects the surface of the bridge using a drone. In view of the problem that the existing drones do not adapt to the different working modes of the drones at different positions of the bridge when performing bridge surface defect detection, which causes excessive power consumption of the drone, the drone is creatively combined with a guide rail structure to design a drone that can adapt to the guide rail structure, so that when the drone is performing image acquisition, the camera position of the camera can be adapted to the bridge surface. At the same time, the rotor is no longer the power to provide the drone with suspension, but is converted into the power to drive the drone to move on the guide rail, making the drone's endurance stronger and further ensuring the drone's adaptability to the environment. However, in the actual detection process, although the prior art realizes that the camera position of the camera is adapted to the bridge surface for detection, since the drone needs to be connected to the surface of the bridge through the guide rail, the range of the picture that can be captured is small, that is, the drone needs to work continuously for a long time during the detection to ensure a comprehensive detection of the bridge surface. Detection, although the power of the drone's suspension can be converted into driving force to increase the drone's standby time, it takes a long time to conduct a comprehensive inspection, which is also not conducive to rapid detection. For some situations that require overhead detection, such as some urban planning or lakes and forests and other areas with few people, it is necessary to use high-altitude overhead shooting for detection, and the existing detection method cannot be used for high-altitude overhead shooting, so there are great limitations. Secondly, when performing visual inspection, an initial reference image is required for comparative detection. For example, for bridge inspection, the initial state of the bridge image will be compared with the current state of the bridge image, or for lake and forest inspection under overhead shooting, the initial image of the lake and forest will be compared with the current state image, and then the detection result can be accurately and quickly obtained according to the different area points of the two images. Therefore, when performing visual inspection on the current state of the area to be tested, it is necessary to make the visual shooting range of the drone just the shooting atmosphere area of ​​the initial image to obtain two sets of visual images that can be aligned to make the detection result more accurate, and the existing technology often only has visual detection function, and cannot align the shooting range of the drone well. Summary of the invention

[0004] The purpose of the present invention is to provide a UAV visual detection alignment method in order to solve at least one of the above technical problems.

[0005] The present invention achieves the above-mentioned purpose through the following technical solutions: a method for visual detection and alignment of unmanned aerial vehicles, comprising an unmanned aerial vehicle body and a detection and shooting area, wherein the unmanned aerial vehicle body is suspended and arranged directly above the detection and shooting area, the unmanned aerial vehicle body stores an initial image of the detection and shooting area, and a visual camera is connected to the fuselage of the unmanned aerial vehicle body. The unmanned aerial vehicle body performs alignment and shooting detection of the detection and shooting area, comprising the following steps: Step 1: Image annular division: perform region division on the initial image of the detection shooting area, divide the detection shooting area into multiple areas and mark the feature points of each divided area, and also mark the feature points in the central area of ​​the image; Step 2: annular area recognition: the drone body is raised to a certain height, and the height is recorded as L1. The drone body roughly scans the current status of the detection and shooting area at the raised height. Combined with the regional division information of the initial image of the detection and shooting area, the visual camera is used to identify each divided area of ​​the current detection and shooting area, and the feature points in each annular area and the feature points in the central area are identified; Step 3: Positioning the peripheral feature points: Select the closest feature point as the base point, and hover the drone body 1 just above the base point to achieve positioning of the base point; Step 4: Position the central feature point, adjust the flight trajectory of the drone body, so that the drone body flies from the above-positioned base point along a straight line to the feature point in the central area, and record the flight distance of the drone body as S1; Step 5: Lift and align positioning. Based on the shooting angle of the visual camera, the drone body is adjusted to a certain height so that the video shooting range of the visual camera covers the entire detection shooting area, thus forming an aligned visual detection of the detection shooting area.

[0006] As a further solution of the present invention: the detection and shooting area is divided into a plurality of outer annular areas and an inner central area, and the areas divided into the detection and shooting area correspond one to one with the area divisions made by the initial image of the detection and shooting area.

[0007] As a further solution of the present invention: each outer annular area divided by the detection and shooting area is provided with a plurality of annular area feature points, the inner central area divided by the detection and shooting area is provided with a central area feature point, and the annular area feature points and the central area feature points are arranged at selected positions of the detection and shooting area in one-to-one correspondence with the feature points on the initial image of the detection and shooting area.

[0008] As a further solution of the present invention: each outer annular area divided by the detection and shooting area is provided with four annular area feature points distributed in a cross shape.

[0009] As a further solution of the present invention: in step 2, when the annular area is identified, the height position of the drone body rises to ensure that the visual camera coverage area is larger than the area of ​​the detection and shooting area.

[0010] As a further solution of the present invention: in step three, when selecting peripheral feature points, two adjacent outer annular areas are used as reference planes, and two adjacent annular area feature points on the outer annular area are used as feature points to be selected, that is, the quadrilateral formed by the four annular area feature points can be used as the reference plane, and the four annular area feature points are all feature points to be selected.

[0011] As a further solution of the present invention: in step 5, the conditions for the lifting height of the drone body specifically include: Combined with the distance S1 from the positioning base point to the feature point in the central area of ​​the drone body, the height position of the circular area where the positioning base point is located captured by the visual camera is calculated; The positioning base point selected in step 3 and the central area feature point selected in step 4 are placed in correspondence in the initial image of the detection shooting area, and the distance between the two points on the initial image is measured and recorded as S2; The distance between the feature point in the central area of ​​the initial image and the corner of the initial image is measured and recorded as S3, and the lift-off height L2 of the corner of the detection shooting area taken by the drone body is determined by using the proportional calculation formula; According to the difference between L2 and L1, the height of the drone body is adjusted to the corresponding difference.

[0012] The beneficial effects of the present invention are as follows: accurate detection of the inspection shooting area is achieved by combining the initial image of the inspection shooting area with the current status of the inspection shooting area, and the initial image of the inspection shooting area can be used as a reference to enable the drone to perform short-distance flight adjustments and capture the current status image corresponding to the initial image, so that the detection result is more accurate; When aligning the drone body, the present invention makes the areas divided by the detection and shooting area correspond to the areas divided by the initial image of the detection and shooting area one by one, so that the initial image of the detection and shooting area and the current situation of the detection and shooting area are equally divided, so that the initial image of the detection and shooting area can reflect the general situation of the current situation of the detection and shooting area, that is, it is convenient for the drone body to perform regional recognition of the detection and shooting area and feature point recognition of each area; and further set a plurality of feature points, which can point to the flight control trajectory of the drone body through the corresponding feature points on the initial image, and also facilitate the measurement of the flight distance of the drone body through the set feature points, and can measure the distance of the feature points set on the initial image of the detection and shooting area, so as to calculate the lift-off height of the drone body at the corner of the detection and shooting area in combination with the proportional formula, so as to realize the alignment shooting detection of the detection and shooting area. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a schematic diagram of the alignment method flow of the present invention; Figure 2 This is a schematic diagram of the state of positioning peripheral feature points of the drone body of the present invention; Figure 3 This is a schematic diagram of the central feature point positioning state of the drone body of the present invention; Figure 4 This is a schematic diagram of the structure of the drone body of the present invention in the lifting, alignment and positioning state; Figure 5 Schematic diagram of the feature point division of the shooting area detected by the present invention.

[0014] In the figure: 1. UAV body, 11. Visual camera, 2. Detection and shooting area, 21. Outer annular area, 22. Inner central area, 23. Feature points of the annular area, 24. Feature points of the central area. DETAILED DESCRIPTION

[0015] 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. Embodiment 1

[0016] like Figures 1 to 5As shown, a method for visual detection and alignment of a drone includes a drone body 1 and a detection and shooting area 2. The drone body 1 is suspended and arranged directly above the detection and shooting area 2. The drone body 1 stores an initial image of the detection and shooting area 2. A visual camera 11 is connected to the fuselage of the drone body 1. The alignment and shooting detection of the detection and shooting area 2 by the drone body 1 includes the following steps: Step 1: image annular division: divide the initial image of the detection shooting area 2 into multiple areas and mark the feature points of each divided area. Also mark the feature points in the central area of ​​the image. Step 2: annular area recognition: the drone body 1 is raised to a certain height, and the height is recorded as L1. The drone body 1 roughly scans the current status of the detection and shooting area 2 at the raised height. Combined with the area division information of the initial image of the detection and shooting area 2, the visual camera 11 recognizes each divided area of ​​the current detection and shooting area 2, and recognizes the feature points in each annular area and the feature points in the central area; Step 3: Positioning the peripheral feature points: Select the closest feature point as the base point, and hover the drone body 1 just above the base point to achieve positioning of the base point; Step 4: Position the central feature point, adjust the flight trajectory of the drone body 1, so that the drone body 1 flies from the above-positioned base point along a straight line to the feature point in the central area, and record the flight distance of the drone body 1 as S1; Step 5, lifting and alignment positioning, based on the shooting lens angle of the visual camera 11, and adjusting the lifting height position of the drone body 1, so that the video shooting range of the visual camera 11 just covers the entire detection and shooting area 2, forming an aligned visual detection of the detection and shooting area 2. Embodiment 2

[0017] In addition to all the technical features of the first embodiment, this embodiment also includes: The detection and shooting area 2 is divided into multiple outer annular areas 21 and an inner central area 22, and the areas divided by the detection and shooting area 2 correspond one to one with the area divisions performed by the initial image of the detection and shooting area 2, that is, the initial image of the detection and shooting area 2 and the current status of the detection and shooting area 2 can be equally divided, and thus can have the same distribution of divided areas, so that the general situation of the current status of the detection and shooting area 2 can be reflected through the initial image of the detection and shooting area 2, that is, it is convenient for the drone body 1 to perform area recognition on the detection and shooting area 2 and feature point recognition of each area.

[0018] Furthermore, each outer annular area 21 divided by the detection and shooting area 2 is provided with a plurality of annular area feature points 23, and the inner central area 22 divided by the detection and shooting area 2 is provided with a central area feature point 24, and the annular area feature points 23 and the central area feature points 24 are set in a one-to-one correspondence with the feature points on the initial image of the detection and shooting area 2 at the selected positions of the detection and shooting area 2, so that the flight control trajectory of the drone body 1 can be pointed through the corresponding feature points on the initial image, and it is also convenient to measure the flight distance of the drone body 1 through the set feature points, and it is possible to measure the distance of the feature points set on the initial image of the detection and shooting area 2, and to calculate the lift-off height of the drone body 1 at the corner of the detection and shooting area 2 in combination with the proportional formula, so as to realize the aligned shooting detection of the detection and shooting area 2.

[0019] Furthermore, each outer annular area 21 divided by the detection and shooting area 2 is provided with four annular area feature points 23 distributed in a cross shape, so that the annular area feature points 23 are evenly distributed around the central area feature point 24, thereby facilitating the selection of the closest feature point as the base point when locating the peripheral feature point. Embodiment 3

[0020] In addition to all the technical features of the first embodiment, this embodiment also includes: In step two, when performing annular area identification, the height of the rising position of the drone body 1 ensures that the shooting coverage area of ​​the visual camera 11 is larger than the area of ​​the detection and shooting area 2, so that a comprehensive shooting of the detection and shooting area 2 can be achieved, that is, the full picture of the detection and shooting area 2 can be captured, so as to facilitate the regional division of the current status of the detection and shooting area 2.

[0021] Furthermore, in step three, when selecting peripheral feature points, two adjacent outer annular areas 21 are used as reference planes, and two adjacent annular area feature points 23 on the outer annular area 21 are used as feature points to be selected, that is, the quadrilateral formed by the four annular area feature points 23 can be used as the reference plane, and the four annular area feature points 23 are all feature points to be selected, which makes it easy to determine the nearest annular area feature point 23 as the base point, so that the drone body 1 can find the base point with the shortest flying distance, which greatly reduces the flight time of the drone body 1.

[0022] Furthermore, in step 5, the conditions for the lifting height of the drone body 1 specifically include: Combined with the distance S1 from the positioning base point to the characteristic point in the central area of ​​the drone body 1, the height position of the annular area where the positioning base point is located captured by the visual camera 11 is calculated; The positioning base point selected in step 3 and the central area feature point selected in step 4 are placed in the initial image of the detection shooting area 2, and the distance between the two points on the initial image is measured and recorded as S2; The distance between the feature point in the central area of ​​the initial image and the corner of the initial image is measured and recorded as S3, and the lift-off height L2 of the corner of the detection shooting area 2 captured by the drone body 1 is determined by using the proportional calculation formula; According to the difference between L2 and L1, the height of the drone body 1 is adjusted up or down by the corresponding difference.

[0023] Working principle: divide the initial image of the detection and shooting area 2 into multiple areas and mark the feature points of each divided area, identify the current detection and shooting area 2 through the drone body 1, and identify the feature points in each annular area and the feature points in the central area; select the closest feature point as the base point, and hover the drone body 1 directly above the base point to achieve the positioning of the base point; adjust the flight trajectory of the drone body 1 so that the drone body 1 flies from the above-mentioned located base point along a straight line to the feature point in the central area, and record the flight distance of the drone body 1 as S1; based on the shooting lens angle of the visual camera 11, adjust the lifting height position of the drone body 1 so that the video shooting range of the visual camera 11 just covers the entire detection and shooting area 2, forming an aligned visual detection of the detection and shooting area 2.

[0024] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0025] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for visual detection and alignment of a drone, comprising a drone body (1) and a detection and shooting area (2), characterized in that: The drone body (1) is suspended and arranged directly above the detection and shooting area (2); the drone body (1) stores an initial image of the detection and shooting area (2); a visual camera (11) is connected to the fuselage of the drone body (1); and the alignment shooting detection of the detection and shooting area (2) by the drone body (1) comprises the following steps: Step 1: annular division of the image: performing region division on the initial image of the detection and shooting area (2), dividing the detection and shooting area (2) into a plurality of areas and marking the feature points of each of the divided areas, and also marking the feature points in the central area of ​​the image; Step 2: annular area recognition: the drone body (1) is raised to a certain height, and the height is recorded as L1. The drone body (1) roughly scans the current status of the detection and shooting area (2) at the raised height. Combined with the area division information of the initial image of the detection and shooting area (2), the visual camera (11) recognizes each divided area of ​​the current detection and shooting area (2), and recognizes the feature points in each annular area and the feature points in the central area; Step 3: Positioning the peripheral feature points, selecting the closest feature point as the base point, and hovering the drone body (1) directly above the base point to achieve positioning of the base point; Step 4: Position the central feature point, adjust the flight trajectory of the drone body (1), so that the drone body (1) flies from the above-positioned base point along a straight line to the feature point in the central area, and record the flight distance of the drone body (1) as S1; Step 5: lifting and alignment positioning. Based on the shooting lens angle of the visual camera (11), the height position of the drone body (1) is adjusted so that the video shooting range of the visual camera (11) just covers the entire detection shooting area (2), thereby forming an aligned visual detection of the detection shooting area (2).

2. The UAV visual detection alignment method according to claim 1, characterized in that: The detection and shooting area (2) is divided into a plurality of outer annular areas (21) and an inner central area (22), and the areas divided into the detection and shooting area (2) correspond one to one with the area divisions made in the initial image of the detection and shooting area (2).

3. The UAV visual detection alignment method according to claim 2, characterized in that: A plurality of annular region feature points (23) are provided in each outer annular region (21) divided by the detection and shooting region (2), a central region feature point (24) is provided in the inner central region (22) divided by the detection and shooting region (2), and the annular region feature points (23) and the central region feature points (24) are provided at selected positions of the detection and shooting region (2) in one-to-one correspondence with feature points on an initial image of the detection and shooting region (2).

4. The UAV visual detection alignment method according to claim 3 is characterized in that: Each outer annular area (21) divided by the detection and shooting area (2) is provided with four annular area characteristic points (23) distributed in a cross shape.

5. The UAV visual detection alignment method according to claim 1, characterized in that: In the second step, when the annular area is identified, the height of the drone body (1) is raised to ensure that the shooting coverage area of ​​the visual camera (11) is larger than the area of ​​the detection shooting area (2).

6. The UAV visual detection alignment method according to claim 1, characterized in that: In the step three, when selecting the peripheral feature points, two adjacent outer annular areas (21) are used as reference surfaces, and two adjacent annular area feature points (23) on the outer annular area (21) are used as feature points to be selected.

7. The UAV visual detection alignment method according to claim 1, characterized in that: In step 5, the conditions for the lifting height of the drone body (1) specifically include: Combined with the distance S1 from the positioning base point to the feature point in the central area of ​​the drone body (1), the height position of the annular area where the positioning base point is located captured by the visual camera (11) is calculated; The positioning base point selected in step 3 and the central area feature point selected in step 4 are placed in correspondence in the initial image of the detection shooting area (2), and the distance between the two points on the initial image is measured and recorded as S2; The distance between the feature point in the central area of ​​the initial image and the corner of the initial image is measured and recorded as S3, and the lift-off height L2 of the corner of the detection shooting area (2) captured by the drone body (1) is determined by using a proportional calculation formula; According to the difference between L2 and L1, the height of the drone body (1) is adjusted to the corresponding difference.

Images