Unmanned aerial vehicle aerial photography system and method based on transportation engineering

Abstract

The invention discloses an unmanned aerial vehicle aerial photography system and method based on traffic transportation engineering, and belongs to the technical field of traffic transportation engineering facility state detection, and the system comprises a flight parameter setting module, an image shooting and preprocessing module, a pit detection module, a side abrasion detection module and a state evaluation module. According to the invention, the unmanned aerial vehicle aerial photography technology is adopted to shoot the rail head joints in the railway section to be detected, and then the images are processed to obtain accurate pit degree characterization parameters and side abrasion degree characterization parameters of the rail head joints; and state evaluation work of all rail head joints in the railway section to be detected is realized based on the characterization parameters, so that the states of the rail head joints of the steel rail can be accurately and efficiently detected and evaluated, and meanwhile, the labor cost is greatly reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of transportation engineering facility status detection technology, and in particular to a UAV aerial photography system and method based on transportation engineering. Background Art

[0002] Transportation engineering focuses on the planning, design, construction, operation, and management of transportation infrastructure, including railways, highways, waterways, and aviation. Railway infrastructure primarily includes rails, sleepers, trackbeds, and turnouts. Rails are the primary components of railway tracks, guiding the train wheels and bearing the weight and impact of the train. Sleepers are laid beneath the rails to support them and evenly transfer the pressure from the rails to the trackbed. The trackbed, located beneath the sleepers, primarily absorbs and distributes the pressure from the sleepers to the roadbed, while also providing elasticity and drainage for the track. Turnouts are the connecting devices that allow trains to switch from one track to another. Common examples include single-opening turnouts, symmetrical turnouts, and triple-opening turnouts.

[0003] For railway facilities, timely condition monitoring is crucial for ensuring railway transportation safety and improving transportation efficiency. Rail condition monitoring, in particular, is crucial for ensuring safe train operation. Existing technologies for rail condition monitoring primarily rely on manual inspection and on-site instrument testing. Manual inspection involves visual inspection of rail condition using simple tools (such as rulers and feeler gauges). While this method is simple and intuitive, it suffers from strong subjectivity, low efficiency, and high labor costs. On-site instrument testing involves on-site inspection of rail condition using various specialized testing instruments. This method significantly improves accuracy, but still suffers from low efficiency and high labor costs.

[0004] How to accurately and efficiently detect and evaluate the condition of rail head joints using drone aerial photography technology is an urgent problem to be solved. To this end, a drone aerial photography system and method based on transportation engineering is proposed. Summary of the Invention

[0005] The technical problem to be solved by the present invention is: how to accurately and efficiently detect and evaluate the rail head joints based on UAV aerial photography technology while reducing labor costs, and provide a UAV aerial photography system based on transportation engineering.

[0006] The present invention solves the above technical problems through the following technical solutions, which include a flight parameter setting module, an image capture and preprocessing module, a pit detection module, a side wear detection module and a status assessment module; The flight parameter setting module is used to set the flight parameters of the UAV; The image capture and preprocessing module is used to use the industrial camera carried by the drone to capture each rail head joint in the railway section to be inspected at each set shooting position while the drone is flying according to the set flight parameters, thereby obtaining multiple pairs of overhead images of the rail head joints, and then preprocessing the pairs of overhead images of the rail head joints; wherein a single pair of overhead images of the rail head joints includes an overhead color image and an overhead depth image of the rail head joints, both of which have consistent scene information and specifications; The pit detection module is used to process the pre-processed overhead image pair of the rail head connection, and then obtain the pit degree characterization parameter of the current rail head connection; The side wear detection module is used to process the pre-processed overhead image pair of the rail head connection, and then obtain the side wear degree characterization parameter of the current rail head connection; The state evaluation module is used to obtain the state score of the current rail head connection based on the pit degree characterization parameters and the side wear degree characterization parameters of the current rail head connection, so as to realize the state evaluation of the current rail head connection.

[0007] Furthermore, the flight parameter setting module includes a start and end point setting unit, a flight path setting unit, a flight altitude setting unit, a flight speed setting unit and a shooting position setting unit; the start and end point setting unit is used to set the starting and end points of the UAV according to the actual situation on site, the flight path setting unit is used to collect the design data of the railway section to be inspected, and then set the flight path of the UAV, the flight altitude setting unit is used to set the flight altitude range of the UAV according to the shooting mission requirements, UAV performance, the height of obstacles along the railway and regulatory requirements; the flight speed setting unit is used to set the flight speed of the UAV according to the endurance of the UAV and the time requirements of the shooting mission; the shooting position setting unit is used to add multiple shooting positions to the set flight path according to the rail head connection position in the design data of the railway section to be inspected, so as to realize fixed-point shooting.

[0008] Furthermore, the pit detection module includes a target detection unit, a contour detection unit, a first area division unit and a pit degree characterization parameter acquisition unit; the target detection unit is used to detect the rail head connection area in the overhead color image of the rail head connection using the trained detection model, and obtain the rail head connection area detection frame and its position information in the image; the contour detection unit is used to detect the contour lines inside the rail head connection area detection frame based on the rail head connection area detection frame and its position information in the image, delete the contour lines whose length values are less than the set length threshold, and obtain the retained contour lines; the first area division unit is used to divide the rail head connection area into the first area division unit and the second area division unit according to the retained contour lines. The head connection area detection frame is divided into three areas, namely area B, area L1 and area R1. The area of each area in the image is calculated, and the areas with an area smaller than the set area threshold are deleted. The remaining areas are area L1 and area R1; the pit degree characterization parameter acquisition unit is used to read the pixel value, that is, the depth value, of each pixel point in area L2 and area R2 in the corresponding overhead depth image of the rail head connection according to the positions of area L1 and area R1, and obtain the pit degree characterization parameter of the current rail head connection according to the depth value of each pixel point in area L2 and area R2, wherein the positions of area L1 correspond to area L2, and the positions of area R1 correspond to area R2.

[0009] Furthermore, the specific processing process of the contour detection unit is as follows: Step S311: obtaining the rail head connection area detection frame and its position information in the image, where the position information refers to the coordinates of the upper left corner and lower right corner of the rail head connection area detection frame in the image; Step S312: using the contour detection function in OpenCV to detect the contour lines inside the rail head connection area detection frame to obtain multiple contour lines; Step S313: Delete the data whose length is less than the set length threshold D. 设定 The contour line is obtained by retrieving the retained contour line.

[0010] Furthermore, the specific processing process of the pit degree characterization parameter acquisition unit is as follows: Step S321: Read the pixel value of each pixel in area L2 and area R2 in the corresponding rail head connection aerial depth image according to the position of area L1 and area R1, i.e., the depth value. The depth value of each pixel in area L2 is recorded as PL2. i , the depth value of each pixel in region R2 is recorded as PR2 j , i and j are the serial numbers of the pixels in the corresponding area; Step S322: In area L2, calculate the mean depth value of all pixels, denoted as PL2 avgeAt the same time, in region R2, calculate the mean depth value of all pixels, denoted as PR2 avge ; Step S323: Calculate the minimum depth value PL2 in area L2 min and mean PL2 avge The difference between them is recorded as PL2 c , and calculate the minimum depth value PR2 in region R2 min and mean PR2 avge The difference between them is recorded as PR2 c ; Step S324: Calculate the difference PL2 c With PR2 c The mean of avge , as a parameter characterizing the pit degree at the current rail head joint.

[0011] Furthermore, the side wear detection module includes a second area division unit and a side wear degree characterization parameter acquisition unit; the second area division unit is used to divide the area L2 and the area R2 in the overhead depth image of the rail head connection to obtain the area L 21 , Area L 22 , Region R 21 With area R 22 The side wear degree characterization parameter acquisition unit is used according to the area L 21 , Area L 22 , Region R 21 With area R 22 The area in the image is used to obtain the side wear degree characterization parameter of the current rail head connection.

[0012] Furthermore, the specific processing process of the second area division unit is as follows: Step S411: reading the pixel value of each pixel point in the area L2 and the area R2 in the overhead depth image of the rail head connection, that is, the depth value; Step S412: Sort the depth values of the pixels in region L2 from large to small, select the first g pixels, and record the coordinates of the g pixels in region L2 in the image; at the same time, sort the depth values of the pixels in region R2 from large to small, select the first h pixels, and record the coordinates of the h pixels in region R2 in the image, where g and h are both positive integers; Step S413: performing straight line fitting based on the coordinates of g pixels in region L2 in the image to obtain the center line of region L2; and performing straight line fitting based on the coordinates of h pixels in region R2 in the image to obtain the center line of region R2; Step S414: Divide both region L2 and region R2 into two regions through the center line, wherein region L2 is divided into region L21 With area L 22 , region R2 is divided into regions R 21 With area R 22 .

[0013] Furthermore, the specific processing process of the side wear degree characterization parameter acquisition unit is as follows: Step 421: Calculate the area L in the image respectively 21 With area L 22 The area is denoted as area S 21L With S 22L ; At the same time, calculate the area R in the image 21 With area R 22 The area is denoted as area S 21R With S 22R ; Step 422: Calculate area S 21L With S 22L The ratio of rL ; Calculate the area S at the same time 21R With S 22R The ratio of rR ; Step 423: Calculate the ratio S rL With S rR The mean of ravge , as a parameter representing the degree of side wear at the current rail head connection.

[0014] Furthermore, the specific processing process of the status assessment module is as follows: Step S51: Obtaining the mean PLR2 avge and mean S ravge , search and compare in the preset scoring database to obtain the mean PLR2 avge Corresponding pit degree score T1 and mean S ravge The corresponding side grinding degree score is T2; Step S52: Calculate the status score T of the current rail head connection total , the calculation formula is as follows: T total =W1*T1+W2*T2; Among them, W1 and W2 are the pit degree score T1 and the side wear degree score T2 in the condition score T total The weight of .

[0015] The present invention also provides a method for aerial photography using a drone based on transportation engineering, which is used to photograph the rail head connection of a rail using the above-mentioned aerial photography system, thereby realizing status detection and evaluation of the rail head connection, comprising the following steps: Step S1: Set the flight parameters of the drone, including the starting and ending points, flight path, flight altitude, flight speed, and shooting position; Step S2: The drone is flown according to the set flight parameters. During the flight, the drone uses an industrial camera mounted on the drone to photograph rail head joints at set shooting positions in the railway section to be inspected. Multiple pairs of overhead images of the rail head joints are obtained and pre-processed. Step S3: Processing the pre-processed overhead image pair of the rail head joint to obtain a characterization parameter of the pit degree and a characterization parameter of the side wear degree of the current rail head joint; Step S4: obtaining a status score of the current rail head connection according to the pit degree characterization parameter and the side wear degree characterization parameter of the current rail head connection, thereby implementing a status evaluation of the current rail head connection; Step S5: According to the processing from step S3 to step S4, the status evaluation of all rail head joints in the railway section to be inspected is completed.

[0016] Compared with the existing technology, the present invention has the following advantages: the drone aerial photography system based on transportation engineering uses drone aerial photography technology to shoot each rail head connection in the railway section to be inspected, and then obtains accurate pit degree characterization parameters and side wear degree characterization parameters of the rail head connection by processing the image, and then realizes the status evaluation of all rail head connections in the railway section to be inspected based on the above characterization parameters, which can accurately and efficiently detect and evaluate the status of the rail head connections, while greatly reducing labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a schematic block diagram of the structure of a UAV aerial photography system based on transportation engineering in an embodiment of the present invention; Figure 2 is a partial schematic diagram of a color image of a rail head connection taken from above in an embodiment of the present invention; Figure 3 Schematic diagram of the area position in the overhead color image of the rail head connection according to an embodiment of the present invention; Figure 4 It is a schematic diagram of the implementation process of the UAV aerial photography method based on transportation engineering in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following is a detailed description of an embodiment of the present invention. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process. However, the protection scope of the present invention is not limited to the following embodiment.

[0019] like Figure 1As shown, this embodiment provides a technical solution: a UAV aerial photography system based on transportation engineering, comprising the following modules: a flight parameter setting module, an image capture and preprocessing module, a pothole detection module, a side wear detection module, and a status assessment module; In this embodiment, the flight parameter setting module is used to set the flight parameters of the UAV.

[0020] To be more specific, the flight parameter setting module includes a start and end point setting unit, a flight path setting unit, a flight altitude setting unit, a flight speed setting unit and a shooting position setting unit; the start and end point setting unit is used to set the starting and end points of the UAV according to the actual situation on site, the flight path setting unit is used to collect the design data of the railway section to be inspected, and then set the flight path of the UAV, the flight altitude setting unit is used to set the flight altitude range of the UAV according to the shooting mission requirements, UAV performance, the height of obstacles along the railway and regulatory requirements; the flight speed setting unit is used to set the flight speed of the UAV according to the UAV's endurance and the shooting mission time requirements; the shooting position setting unit is used to add multiple shooting positions to the set flight path according to the rail head connection position in the design data of the railway section to be inspected, so as to realize fixed-point shooting.

[0021] More specifically, the design data of the railway section to be inspected includes line parameters, curve radius, slope, rail head connection position, etc., to clarify the exact direction of the railway and the location of key nodes.

[0022] It should be noted that at the rail head connection position, due to insufficient material hardness and further expansion of fish-scale cracks, pits are likely to appear on the rail top surface. Large pits on the rail top will affect driving safety, so timely detection of rail top pits is very necessary. At the rail head connection position in the curved section, due to excessive lateral force between the wheel flange and the rail, asymmetric wear of the rail head side is likely to occur, that is, rail head side wear. Rail head side wear will shorten the rail life to a certain extent and increase wheel-rail noise. Therefore, timely detection of rail head side wear is also very necessary.

[0023] In this embodiment, the image capture and preprocessing module is used to use the industrial camera carried below the drone to capture each rail head connection in the railway section to be inspected below at each set shooting position while the drone is flying according to the set flight parameters (set by the flight parameter setting module), to obtain multiple pairs of overhead images of the rail head connection, and then preprocess the pairs of overhead images of the rail head connection. A single pair of overhead images of the rail head connection includes a color image and a depth image with consistent scene information and specifications, namely, an overhead color image of the rail head connection and an overhead depth image of the rail head connection.

[0024] It should be noted that in this embodiment, during the drone flight, only the rail head connection of the rails on one side of the railway section to be inspected is photographed during the outbound flight, and only the rail head connection of the rails on the other side of the railway section to be inspected is photographed during the return flight, and the image includes the complete rail head connection area.

[0025] More specifically, in the image capturing and preprocessing module, the industrial camera is an RGBD camera, which is a camera that can simultaneously acquire color information (RGB) and depth information (D) of a scene.

[0026] To be more specific, in the image capture and preprocessing module, preprocessing includes noise reduction processing and enhancement processing, wherein the noise reduction processing includes but is not limited to the use of mean filtering processing, and the enhancement processing includes but is not limited to the histogram equalization processing. After noise reduction processing and enhancement processing, the quality of the image can be greatly improved, thereby ensuring the accuracy of subsequent recognition and calculation processing.

[0027] In this embodiment, the pit detection module is used to process the pre-processed overhead image pair of the rail head connection, and then obtain the pit degree representation parameter of the current rail head connection.

[0028] More specifically, the pit detection module includes a target detection unit, a contour detection unit, a first area division unit and a pit degree characterization parameter acquisition unit; the target detection unit is used to detect the rail head connection area in the overhead color image of the rail head connection using the trained detection model, and obtain the rail head connection area detection frame and its position information in the image; the contour detection unit is used to detect the contour lines inside the rail head connection area detection frame based on the rail head connection area detection frame and its position information in the image, delete the contour lines whose length values are less than the set length threshold, and obtain the retained contour lines; the first area division unit is used to divide the rail head connection area detection frame into three areas through the retained contour lines, namely area B, area L1 and area R1, see Figure 2 , calculate the area of each region in the image, delete the region with an area smaller than the set area threshold, and the remaining regions are region L1 and region R1; the pit degree characterization parameter acquisition unit is used to read the pixel value, that is, the depth value, of each pixel point in region L2 and region R2 in the corresponding overhead depth image of the rail head connection according to the positions of region L1 and region R1, and obtain the pit degree characterization parameter of the current rail head connection according to the depth value of each pixel point in region L2 and region R2, wherein the positions of region L1 correspond to region L2, and the positions of region R1 correspond to region R2.

[0029] More specifically, in the target detection unit, the detection model is trained based on the yolo v3s target detection network. The parameter amount of the yolo v3s target detection network is relatively small and is more suitable for deployment in edge devices.

[0030] More specifically, the specific processing process of the contour detection unit is as follows: Step 1: Obtain the rail head connection area detection frame and its position information in the image. The position information refers to the coordinates of the upper left corner and lower right corner of the rail head connection area detection frame in the image. Step 2: Use the contour detection function in OpenCV to detect the contour lines inside the rail head connection area detection frame to obtain m contour lines. In this embodiment, m=8 (excluding the boundary line of the rail head connection area detection frame); Step 3: Delete the length value less than the set length threshold D 设定 The contour lines are obtained to obtain the retained contour lines, and the retained contour lines are n. In this embodiment, n=6; the two shorter contour lines between the two rail heads are deleted, and the remaining contour lines are retained.

[0031] More specifically, the specific processing process of the pit degree characterization parameter acquisition unit is as follows: Step 1: Read the pixel value of each pixel in area L2 and area R2 in the corresponding rail head connection aerial depth image according to the position of area L1 and area R1, that is, the depth value. The depth value of each pixel in area L2 is recorded as PL2. i , the depth value of each pixel in region R2 is recorded as PR2 j , i and j are the serial numbers of the pixels in the corresponding area; Step 2: In area L2, calculate the mean depth value of all pixels, recorded as PL2 avge At the same time, in region R2, calculate the mean depth value of all pixels, denoted as PR2 avge ; Step 3: Calculate the minimum depth value PL2 in area L2 min and mean PL2 avge The difference between them is recorded as PL2 c , and calculate the minimum depth value PR2 in region R2 min and mean PR2 avge The difference between them is recorded as PR2 c ; Step 4: Calculate the difference PL2 c With PR2 c The mean of avge , as a parameter characterizing the pit degree at the current rail head joint.

[0032] In the present invention, the mean PLR2 is used avge As a parameter representing the pit degree at the current rail head connection, the pit degree at the rail head connection can be more accurately represented.

[0033] It should be noted that if Figure 2 As shown, the rail head portion of the rail 1 in this embodiment is connected by fasteners, which include positioning plates 21 and fastening bolts 22. The fastening bolts 22 pass through the positioning plates 21 on both sides of the rail waist and the rail 1 body to achieve a fastening connection.

[0034] In this embodiment, the side wear detection module is used to process the pre-processed overhead images of the rail head connection, and then obtain the characterization parameters of the side wear degree of the current rail head connection.

[0035] As more specific, the side wear detection module includes a second area division unit and a side wear degree characterization parameter acquisition unit; the second area division unit is used to divide the area L2 and the area R2 in the overhead depth image of the rail head connection to obtain the area L 21 , Area L 22 , Region R 21 With area R 22 ,See Figure 3 The side wear degree characterization parameter acquisition unit is used according to the area L 21 , Area L 22 , Region R 21 With area R 22 The area in the image is used to obtain the side wear degree characterization parameter of the current rail head connection.

[0036] More specifically, the specific processing process of the second area division unit is as follows: Step 1: Read the pixel value of each pixel in area L2 and area R2 in the overhead depth image of the rail head connection, that is, the depth value; Step 2: Sort the depth values of each pixel in region L2 from large to small, select the first g pixels, and record the coordinates of g pixels in region L2 in the image; at the same time, sort the depth values of each pixel in region R2 from large to small, select the first h pixels, and record the coordinates of h pixels in region R2 in the image, where g and h are both positive integers; Step 3: Perform straight line fitting based on the coordinates of g pixels in region L2 in the image to obtain the center line of region L2; at the same time, perform straight line fitting based on the coordinates of h pixels in region R2 in the image to obtain the center line of region R2; the fitting method is the least squares method; Step 4: Divide both region L2 and region R2 into two regions through the center line, where region L2 is divided into region L 21With area L 22 , region R2 is divided into regions R 21 With area R 22 .

[0037] More specifically, the specific processing process of the side wear degree characterization parameter acquisition unit is as follows: Step 1: Calculate the area L in the image separately 21 With area L 22 The area is denoted as area S 21L With S 22L ; At the same time, calculate the area R in the image 21 With area R 22 The area is denoted as area S 21R With S 22R ; Step 2: Calculate the area S 21L With S 22L The ratio of rL ; Calculate the area S at the same time 21R With S 22R The ratio of rR ; Step 3: Calculate the ratio S rL With S rR The mean of ravge , as a parameter representing the degree of side wear at the current rail head connection.

[0038] In the present invention, the mean S is used ravge As a parameter representing the degree of side wear at the current rail head connection, the degree of side wear at the rail head connection can be represented more accurately.

[0039] In this embodiment, the state evaluation module is used to obtain the state score of the current rail head connection based on the pit degree characterization parameters and the side wear degree characterization parameters of the current rail head connection, thereby realizing the state evaluation of the current rail head connection.

[0040] More specifically, the specific processing process of the status assessment module is as follows: Step 1: Get the mean PLR2 avge and mean S ravge , search and compare in the preset scoring database to obtain the mean PLR2 avge Corresponding pit degree score T1 and mean S ravge The corresponding side grinding degree score is T2; Step 2: Calculate the status score T of the current rail head joint total , the calculation formula is as follows: T total =W1*T1+W2*T2; Among them, W1 and W2 are the pit degree score T1 and the side wear degree score T2 in the condition score T total The weight of can be set and improved according to actual application.

[0041] It should be noted that, in this embodiment, the scoring database has a preset mean PLR2 avge The corresponding relationship between the pit degree score and the mean S ravge The correspondence between the score of side wear and the degree of side wear.

[0042] like Figure 4 As shown, this embodiment also provides a UAV aerial photography method based on transportation engineering, which is used to use the above-mentioned aerial photography system to photograph the rail head connection of the rail, thereby realizing the state detection and evaluation of the rail head connection, including the following steps: Step 1: Set the flight parameters of the drone, including the starting and ending points, flight path, flight altitude, flight speed, and shooting position; Step 2: The drone is flown according to the set flight parameters. During the flight, the drone uses the industrial camera mounted on the underside of the drone to photograph each rail head joint in the railway section to be inspected at each set shooting position. Multiple pairs of overhead images of the rail head joints are obtained and pre-processed. Step 3: Process the pre-processed overhead image pair of the rail head joint to obtain the characterization parameters of the pit degree and the side wear degree of the current rail head joint; Step 4: Obtain a status score of the current rail head joint based on the pit degree characterization parameters and the side wear degree characterization parameters of the current rail head joint, thereby achieving a status assessment of the current rail head joint; Step 5: Following the process from Step 3 to Step 4 above, complete the status assessment of all rail head joints in the railway section to be inspected.

[0043] In this embodiment, the detailed processing procedures of each step in the aerial photography method can be found in the corresponding description of the above-mentioned aerial photography system, which will not be repeated here.

[0044] In summary, the UAV aerial photography system and method based on transportation engineering in the above-mentioned embodiment adopts UAV aerial photography technology to photograph each rail head connection in the railway section to be inspected, and then obtains accurate pit degree characterization parameters and side wear degree characterization parameters of the rail head connection by processing the image, and then realizes the status evaluation of all rail head connections in the railway section to be inspected based on the above-mentioned characterization parameters, which can accurately and efficiently detect and evaluate the status of the rail head connections, while greatly reducing labor costs.

[0045] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A UAV aerial photography system based on transportation engineering, characterized in that: include: Flight parameter setting module, image capture and pre-processing module, pit detection module, side wear detection module and status assessment module; The flight parameter setting module is used to set the flight parameters of the UAV; The image capture and preprocessing module is used to use the industrial camera carried by the drone to capture each rail head joint in the railway section to be inspected at each set shooting position while the drone is flying according to the set flight parameters, thereby obtaining multiple pairs of overhead images of the rail head joints, and then preprocessing the pairs of overhead images of the rail head joints; wherein a single pair of overhead images of the rail head joints includes an overhead color image and an overhead depth image of the rail head joints, both of which have consistent scene information and specifications; The pit detection module is used to process the pre-processed overhead image pair of the rail head connection, and then obtain the pit degree characterization parameter of the current rail head connection; The side wear detection module is used to process the pre-processed overhead image pair of the rail head connection, and then obtain the side wear degree characterization parameter of the current rail head connection; The state evaluation module is used to obtain the state score of the current rail head connection based on the pit degree characterization parameters and the side wear degree characterization parameters of the current rail head connection, so as to realize the state evaluation of the current rail head connection.

2. The UAV aerial photography system based on transportation engineering according to claim 1, characterized in that: The flight parameter setting module includes a start and end point setting unit, a flight path setting unit, a flight altitude setting unit, a flight speed setting unit and a shooting position setting unit; the start and end point setting unit is used to set the starting and end points of the UAV according to the actual situation on site; the flight path setting unit is used to collect the design data of the railway section to be inspected, and then set the flight path of the UAV; the flight altitude setting unit is used to set the flight altitude range of the UAV according to the shooting mission requirements, UAV performance, the height of obstacles along the railway and regulatory requirements; the flight speed setting unit is used to set the flight speed of the UAV according to the endurance of the UAV and the time requirements of the shooting mission; the shooting position setting unit is used to add multiple shooting positions to the set flight path according to the rail head connection position in the design data of the railway section to be inspected, so as to realize fixed-point shooting.

3. The UAV aerial photography system based on transportation engineering according to claim 1, characterized in that: The pit detection module includes a target detection unit, a contour detection unit, a first region division unit, and a pit degree characterization parameter acquisition unit; the target detection unit is used to detect the rail head connection area in the overhead color image of the rail head connection using a trained detection model, and obtain the rail head connection area detection frame and its position information in the image; The contour detection unit is used to detect the contour lines inside the rail head connection area detection frame according to the rail head connection area detection frame and its position information in the image, delete the contour lines whose length values are less than the set length threshold, and obtain the retained contour lines; the first area division unit is used to divide the rail head connection area detection frame into three areas, namely area B, area L1 and area R1, through the retained contour lines, calculate the area of each area in the image, delete the area whose area is less than the set area threshold, and the retained areas are area L1 and area R1; the pit degree characterization parameter acquisition unit is used to read the pixel value, that is, the depth value, of each pixel point in area L2 and area R2 in the corresponding overhead depth image of the rail head connection according to the position of area L1 and area R1, and obtain the pit degree characterization parameter of the current rail head connection according to the depth value of each pixel point in area L2 and area R2, wherein the position of area L1 corresponds to that of area L2, and the position of area R1 corresponds to that of area R2.

4. The UAV aerial photography system based on transportation engineering according to claim 3, characterized in that: The specific processing process of the contour detection unit is as follows: Step S311: obtaining the rail head connection area detection frame and its position information in the image, where the position information refers to the coordinates of the upper left corner and lower right corner of the rail head connection area detection frame in the image; Step S312: using the contour detection function in OpenCV to detect the contour lines inside the rail head connection area detection frame to obtain multiple contour lines; Step S313: Delete the data whose length is less than the set length threshold D. 设定 The contour line is obtained by retrieving the retained contour line.

5. The UAV aerial photography system based on transportation engineering according to claim 4, characterized in that: The specific processing process of the pit degree characterization parameter acquisition unit is as follows: Step S321: Read the pixel value of each pixel in area L2 and area R2 in the corresponding rail head connection aerial depth image according to the position of area L1 and area R1, i.e., the depth value. The depth value of each pixel in area L2 is recorded as PL2. i , the depth value of each pixel in region R2 is recorded as PR2 j , i and j are the serial numbers of the pixels in the corresponding area; Step S322: In area L2, calculate the mean depth value of all pixels, denoted as PL2 avge At the same time, in region R2, calculate the mean depth value of all pixels, denoted as PR2 avge ; Step S323: Calculate the minimum depth value PL2 in area L2 min and mean PL2 avge The difference between them is recorded as PL2 c , and calculate the minimum depth value PR2 in region R2 min and mean PR2 avge The difference between them is recorded as PR2 c ; Step S324: Calculate the difference PL2 c With PR2 c The mean of avge , as a parameter characterizing the pit degree at the current rail head joint.

6. The UAV aerial photography system based on transportation engineering according to claim 5, characterized in that: The side wear detection module includes a second area division unit and a side wear degree characterization parameter acquisition unit; the second area division unit is used to divide the area L2 and the area R2 in the overhead depth image of the rail head connection to obtain the area L 21 , Area L 22 , Region R 21 With area R 22 The side wear degree characterization parameter acquisition unit is used according to the area L 21 , Area L 22 , Region R 21 With area R 22 The area in the image is used to obtain the side wear degree characterization parameter of the current rail head connection.

7. The UAV aerial photography system based on transportation engineering according to claim 6, characterized in that: The specific processing process of the second area division unit is as follows: Step S411: reading the pixel value of each pixel point in the area L2 and the area R2 in the overhead depth image of the rail head connection, that is, the depth value; Step S412: Sort the depth values of the pixels in region L2 from large to small, select the first g pixels, and record the coordinates of the g pixels in region L2 in the image; at the same time, sort the depth values of the pixels in region R2 from large to small, select the first h pixels, and record the coordinates of the h pixels in region R2 in the image, where g and h are both positive integers; Step S413: performing straight line fitting based on the coordinates of g pixels in region L2 in the image to obtain the center line of region L2; and performing straight line fitting based on the coordinates of h pixels in region R2 in the image to obtain the center line of region R2; Step S414: Divide both region L2 and region R2 into two regions through the center line, wherein region L2 is divided into region L 21 With area L 22 , region R2 is divided into regions R 21 With area R 22 .

8. The UAV aerial photography system based on transportation engineering according to claim 7, characterized in that: The specific processing process of the side wear degree characterization parameter acquisition unit is as follows: Step 421: Calculate the area L in the image respectively 21 With area L 22 The area is denoted as area S 21L With S 22L ; At the same time, calculate the area R in the image 21 With area R 22 The area is denoted as area S 21R With S 22R ; Step 422: Calculate area S 21L With S 22L The ratio of rL ; Calculate the area S at the same time 21R With S 22R The ratio of rR ; Step 423: Calculate the ratio S rL With S rR The mean of ravge , as a parameter representing the degree of side wear at the current rail head connection.

9. The UAV aerial photography system based on transportation engineering according to claim 8, characterized in that: The specific processing process of the status assessment module is as follows: Step S51: Obtaining the mean PLR2 avge and mean S ravge , search and compare in the preset scoring database to obtain the mean PLR2 avge Corresponding pit degree score T1 and mean S ravge The corresponding side grinding degree score is T2; Step S52: Calculate the status score T of the current rail head connection total , the calculation formula is as follows: T total =W1*T1+W2*T2; Among them, W1 and W2 are the pit degree score T1 and the side wear degree score T2 in the condition score T total The weight of .

10. A method for aerial photography using a drone for transportation engineering, comprising: photographing a rail head joint of a rail using the aerial photography system according to any one of claims 1 to 9, thereby detecting and evaluating the condition of the rail head joint; the method comprising the following steps: Step S1: Set the flight parameters of the drone, including the starting and ending points, flight path, flight altitude, flight speed, and shooting position; Step S2: The drone is flown according to the set flight parameters. During the flight, the drone uses an industrial camera mounted on the drone to photograph rail head joints at set shooting positions in the railway section to be inspected. Multiple pairs of overhead images of the rail head joints are obtained and pre-processed. Step S3: Processing the pre-processed overhead image pair of the rail head joint to obtain a characterization parameter of the pit degree and a characterization parameter of the side wear degree of the current rail head joint; Step S4: obtaining a status score of the current rail head connection according to the pit degree characterization parameter and the side wear degree characterization parameter of the current rail head connection, thereby implementing a status evaluation of the current rail head connection; Step S5: According to the processing from step S3 to step S4, the status evaluation of all rail head joints in the railway section to be inspected is completed.

Images