A tunnel detection global calibration method

Abstract

The application discloses a tunnel detection global calibration method, and belongs to the technical field of tunnel calibration, and specifically comprises the following steps: firstly, a tunnel simulation environment is constructed; then, a calibration environment of the tunnel simulation environment is constructed; then, a coordinate mapping relationship of a calibration object is calculated in the calibration environment; and the calibration object is calibrated according to the coordinate mapping relationship; the calibration of the calibration object comprises full-section calibration and image splicing calibration; the application adopts a customized calibration plate to replace a traditional checkerboard calibration plate, removes a human subjective splicing process, avoids the introduction of artificial errors, and completes an automatic calibration splicing process through combination of different types of calibration plates, so that the calibration workload is small, and the precision is relatively high.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tunnel calibration, and particularly relates to a tunnel detection global calibration method. BACKGROUND

[0002] At present, the calibration of a tunnel detection camera often adopts a checkerboard calibration board. The checkerboard calibration board is very suitable for the calibration of a single camera, but has limitations for the calibration of adjacent cameras with overlapping areas. In addition, the checkerboard calibration board cannot conveniently find one-to-one corresponding points, and the checkerboard is mostly suitable for surface calibration and needs to be matched with a light source for lighting, which is inconvenient for a structured light imaging device. The splicing calibration method for a multi-view linear array camera group still needs manual assistance, and the splicing calibration is performed through manual observation and auxiliary rotation and translation, so that manual errors are introduced, resulting in unstable system errors. SUMMARY

[0003] The present application aims at the limitations and inconvenience of the current calibration of a tunnel detection camera using a checkerboard calibration board, and the problem that the current splicing calibration method for a multi-view linear array camera group introduces manual errors, resulting in unstable system errors, and provides a tunnel detection global calibration method, which uses multiple forms of calibration boards to assist in completing tunnel detection global calibration, thereby solving the above problems.

[0004] The technical scheme of the present application is as follows:

[0005] A tunnel detection global calibration method comprises the following steps:

[0006] Step S1: constructing a tunnel simulation environment;

[0007] Step S2: constructing a calibration environment of the tunnel simulation environment;

[0008] Step S3: calculating a coordinate mapping relationship of a calibration object in the calibration environment, and calibrating the calibration object according to the coordinate mapping relationship; the calibration of the calibration object comprises full-section calibration and image splicing calibration.

[0009] Further, the step S3 comprises:

[0010] calculating a coordinate mapping relationship of a calibration object;

[0011] calibrating the calibration object according to the coordinate mapping relationship by cascade calculation.

[0012] Further, the step S2 of constructing the calibration environment of the tunnel simulation environment comprises:

[0013] setting a needle-shaped target for single-camera calibration, and moving the needle-shaped target in a direction perpendicular to each camera.

[0014] setting a tooth-shaped target for two adjacent camera pose relationship calibration, and locating the tooth-shaped target in the overlapping area of the two adjacent cameras;

[0015] setting a simulation calibration target for camera group image calibration, and making the camera recognize the origin center of the simulation calibration target.

[0016] Further, the calibration of the calibration object comprises:

[0017] Each camera acquires an image and recognizes corresponding coordinates of the image, the corresponding coordinates comprising image coordinates and world coordinates of the needle-shaped target, image coordinates of the tooth-shaped target, and image coordinates of the simulation calibration target.

[0018] Further, the full-section calibration comprises:

[0019] Calibrating a single camera to obtain a conversion relationship from an image coordinate system of the single camera to a world coordinate system;

[0020] According to the conversion relationship, converting image coordinates of images taken by adjacent cameras into world coordinates;

[0021] According to the world coordinates of two groups of images taken by adjacent cameras, calculating a mapping relationship of world coordinate systems of the adjacent cameras to complete pose relationship calibration of the adjacent cameras;

[0022] Using cascade calculation, completing full-section calibration of all cameras.

[0023] Further, the image splicing calibration comprises:

[0024] According to image coordinates of images taken by each camera, calculating a mapping relationship of image coordinate systems of adjacent cameras;

[0025] According to the mapping relationship of the image coordinate systems of the adjacent cameras, using cascade calculation, completing image splicing calibration of all cameras.

[0026] Further, the calibration of the calibration object in the step S3 further comprises laser calibration.

[0027] The calibration environment of the tunnel simulation environment in the step S2 further comprises:

[0028] Setting a plane target for laser calibration calibration, and making a horizontal line of the plane target coincide with a light ray of the laser.

[0029] Further, the laser calibration comprises:

[0030] With the aid of software algorithm, the included angle of horizontal mark line and laser line in the same imaging camera is calculated, and the vertical width distribution under the horizontal equal interval of the line group is calculated.

[0031] When the included angle is 0 degree and the vertical width distribution is in the preset threshold range, the laser calibration is completed.

[0032] A tunnel detection global calibration device comprises:

[0033] The mobile device, the calibration device arranged on the mobile device for calibrating the calibration object, the tunnel full-section multi-vision sensor for acquiring calibration data of the calibration device, and the multi-view linear array camera group for acquiring tunnel images.

[0034] The calibration device comprises a target, wherein the target comprises a planar target for calibrating a laser, a needle-shaped target for calibrating a single camera, and a tooth-shaped target for calibrating the pose relationship of adjacent cameras.

[0035] Further, the horizontal mark line of the planar target coincides with the light line of the laser, so that the laser calibration is achieved.

[0036] The needle-shaped target moves in the direction of the vertical camera, so that each camera is calibrated to obtain the image coordinates and the world coordinates of each image.

[0037] The tooth-shaped target is located in the overlapping area of two adjacent cameras, so that the pose relationship of the two adjacent cameras is calibrated to obtain the world coordinates of the corresponding corner points in the overlapping area.

[0038] Compared with the existing technology, the present application has the following advantages:

[0039] A tunnel detection global calibration method comprises the following steps: step S1, constructing a tunnel simulation environment; step S2, constructing a calibration environment of the tunnel simulation environment; step S3, calculating the coordinate mapping relationship of a calibration object in the calibration environment; and calibrating the calibration object according to the coordinate mapping relationship; the calibration of the calibration object comprises full-section calibration and image stitching calibration; the present application uses a customized calibration board to replace a traditional checkerboard calibration board, removes the human subjective stitching process to avoid the introduction of artificial errors, and completes the automatic calibration stitching process through the combination of different types of calibration boards, so that the calibration workload is small and the precision is relatively high. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 It is a flowchart of a tunnel detection global calibration method.

[0041] Figure 2 It is a structural schematic diagram of a planar target.

[0042] Figure 3This is a schematic diagram of the structure of a needle-shaped target;

[0043] Figure 4 This is a schematic diagram of the toothed target structure;

[0044] Figure 5 A schematic diagram of the calibration environment for image stitching of a pre-built multi-view linear array camera group;

[0045] Figure 6 This is a detailed data flow diagram of step S22 in Example 2;

[0046] Figure 7 This is a schematic diagram of the camera assembly in Embodiment 3.

[0047] Attached reference numerals: 1-Camera 1, 2-Camera 2, 3-Camera 3, 4-Camera 4, 5-Camera 5, 6-Horizontal datum line, 7-Target pin, 8-Target tooth, 9-Corner point, 10-Slide table, 11-Original point calibration paper, 12-Multi-view linear array camera group, 13-Tunnel simulation environment. Detailed Implementation

[0048] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0049] The features and performance of the present invention will be further described in detail below with reference to embodiments.

[0050] Example 1

[0051] Currently, checkerboard calibration boards are commonly used for calibrating tunnel inspection cameras. While checkerboard calibration boards are very suitable for calibrating individual cameras, they have limitations in calibrating adjacent cameras with overlapping areas. In addition, checkerboard calibration boards cannot easily find one-to-one correspondence points, and checkerboards are mostly suitable for surface calibration and require a light source for illumination, which is inconvenient for structured light imaging devices. Furthermore, the splicing calibration method for multi-view linear array camera groups is still manually assisted, using visual observation to assist in rotation and translation for splicing calibration. This introduces human error, leading to unstable system errors.

[0052] A tunnel detection global calibration method, specifically comprising the following steps:

[0053] Step S1: constructing a tunnel simulation environment; that is, building a tunnel model, and setting corresponding simulation calibration targets in the tunnel simulation environment; preferably, the simulation calibration target is a point-shaped calibration paper, used for multi-view linear array camera group image stitching calibration;

[0054] Step S2: constructing a calibration environment of the tunnel simulation environment;

[0055] In this embodiment, specifically, the step S2 of constructing the calibration environment of the tunnel simulation environment comprises:

[0056] Setting a needle-shaped target for single-camera calibration, so that the needle-shaped target moves in a direction perpendicular to each camera;

[0057] Setting a tooth-shaped target for calibration of the pose relationship of two adjacent cameras, so that the tooth-shaped target is located in the overlapping area of the two adjacent cameras;

[0058] Setting a simulation calibration target for camera group image calibration, so that the camera recognizes the origin center of the simulation calibration target; preferably, the simulation calibration target is a calibration paper, which is attached in the tunnel model;

[0059] In this embodiment, specifically, the step S2 of constructing the calibration environment of the tunnel simulation environment further comprises:

[0060] A moving device for installing each target is provided to realize image acquisition and calibration at different positions; preferably, the moving device is a sliding table provided in the tunnel model and movable along the tunnel direction;

[0061] Step S3: calculating the coordinate mapping relationship of the calibration object in the calibration environment; and calibrating the calibration object according to the coordinate mapping relationship; the calibration of the calibration object comprises full-section calibration and image stitching calibration; wherein the full section refers to the tunnel section photographed by the camera, reflecting the shape characteristics of the tunnel, and the image stitching calibration mainly reflects the overall texture of the tunnel; the specific calibration object is a camera;

[0062] In this embodiment, the step S3 comprises:

[0063] Calculating the coordinate mapping relationship of the calibration object;

[0064] Calibrating the calibration object according to the coordinate mapping relationship; that is, converting all cameras into a set reference coordinate system until all cameras are calibrated.

[0065] In this embodiment, the calibration of the calibration object comprises:

[0066] Each camera acquires images and identifies corresponding coordinates of the images, including image coordinates and world coordinates of the pin-shaped target, image coordinates of the tooth-shaped target, and image coordinates of the simulated calibration target;

[0067] In the embodiment, the full-section calibration includes:

[0068] Calibrate the single camera to obtain a conversion relationship from an image coordinate system of the single camera to a world coordinate system;

[0069] According to the conversion relationship, convert image coordinates of images captured by adjacent cameras into world coordinates;

[0070] According to the world coordinates of two groups of images captured by adjacent cameras, calculate a mapping relationship of world coordinate systems of the adjacent cameras to complete calibration of a pose relationship of the adjacent cameras;

[0071] Using cascading calculation, complete full-section calibration of all cameras.

[0072] In the embodiment, the image splicing calibration includes:

[0073] According to image coordinates of images captured by each camera, calculate a mapping relationship of image coordinate systems of adjacent cameras;

[0074] According to the mapping relationship of the image coordinate systems of the adjacent cameras, using cascading calculation, complete image splicing calibration of all cameras.

[0075] The calibration of the calibration object in the step S3 further includes laser calibration.

[0076] The calibration environment of the tunnel simulation environment in the step S2 further includes:

[0077] Set a planar target for laser calibration and calibration, so that a horizontal line of the planar target coincides with a light line of the laser.

[0078] Further, the detailed steps of the laser calibration in the step S2 are as follows:

[0079] Using software algorithm assistance, calculate an included angle between a horizontal line and a laser line in the same imaging camera, and a vertical width distribution under horizontal equal intervals of a line group;

[0080] When the included angle is 0 degrees and the vertical width distribution is within a preset threshold range, the laser calibration and calibration are completed.

[0081] A tunnel detection global calibration device includes:

[0082] Mobile device, calibration device arranged on the mobile device for calibrating a calibration object, tunnel full-section multi-vision sensor for acquiring calibration data of the calibration device, multi-view linear array camera group for acquiring tunnel images;

[0083] The calibration device comprises a target, wherein the target comprises a planar target for calibrating a laser, a needle-shaped target for calibrating a single camera, and a tooth-shaped target for calibrating a pose relationship of adjacent cameras.

[0084] Further, the horizontal line of the planar target coincides with the light line of the laser, so as to realize calibration and adjustment of the laser.

[0085] Further, the needle-shaped target moves in the direction of the vertical camera, so as to realize calibration of each camera and obtain image coordinates and world coordinates of each image.

[0086] Further, the tooth-shaped target is located in the overlapping area of two adjacent cameras, so as to realize calibration of the pose relationship of the two adjacent cameras and obtain the world coordinates of the corresponding corner points in the overlapping area.

[0087] Embodiment two

[0088] Embodiment two is a further description of embodiment one, please refer to Figures 1-3 .

[0089] The specific steps of full-section calibration for the tunnel full-section multi-vision sensor are as follows:

[0090] 1. A calibration environment for the laser is built in the tunnel simulation environment; specifically, a planar target with horizontal lines is placed on a sliding table, and the specific structure of the planar target is as shown in Figure 2 .

[0091] 2. The laser is calibrated in the calibration environment; specifically, the position of the laser is adjusted so that the light line emitted by the laser coincides with the horizontal line, and the calibration of the laser is completed; that is, the module with the laser and the camera is placed at the corresponding detection position in the tunnel simulation environment, and then the position of the laser is continuously adjusted so that the light line emitted by the laser coincides with the horizontal line, and the calibration of the laser is completed.

[0092] The basis for judging whether the two lines coincide is that: the software algorithm is used for assistance, the included angle of the horizontal line and the laser line in the same camera is calculated, and the vertical width distribution of the region formed by the two lines at equal intervals is used to assist the collinear calibration; preferably, the included angle of the two lines is 0 degrees in the ideal case, the width distribution is concentrated and meets the predetermined threshold, and the calibration of the laser is considered to be completed; the specific data flow is as shown in Figure 6 .

[0093] 3. Build a single camera calibration environment in the tunnel simulation environment; specifically, install the needle-shaped target on the sliding table, that is, complete the construction of the single camera calibration environment; as shown in FIG. 2, the needle-shaped target is provided with target needles arranged at intervals; Figure 3

[0094] 4. Calibrate the single camera in the calibration environment to obtain the conversion relationship of the image coordinate system of the single camera to the world coordinate system, specifically, including the following steps:

[0095] (1) Control the needle-shaped target to move with the sliding table, and record the world coordinates of each target needle in the needle-shaped target;

[0096] (2) The camera collects images of the needle-shaped target at different positions, and identifies the image coordinates of each target needle in the image;

[0097] (3) Solve the conversion relationship of the image coordinates to the world coordinates, complete the single camera calibration, and obtain the conversion relationship of the image coordinate system of the single camera to the world coordinate system; the specific solving method of the image coordinates to the world coordinates can be known by those skilled in the art without creative labor, which will not be repeated here; then calibrate each camera in the module according to the above steps, that is, complete the calibration of all single cameras in the tunnel full-face multi-vision sensor;

[0098] 5. Build a calibration environment for the pose relationship of adjacent cameras in the tunnel simulation environment; specifically, set the tooth-shaped target on the sliding table, and make the tooth-shaped target located in the overlapping area of the visual angle of the adjacent two cameras, that is, complete the construction of the calibration environment; Figure 4 The structure diagram of the tooth-shaped target is given, preferably, the tooth-shaped target is provided with continuous target teeth, the top end of the target teeth is an angle point, the target teeth are in the shape of an isosceles right triangle, and the length of the right angle side is 80 mm;

[0099] 6. Adjacent two cameras shoot the tooth-shaped target, identify the image coordinates of the same angle point on the tooth-shaped target at different positions, and convert the two groups of image coordinates into world coordinates according to the conversion relationship of the above image coordinate system to the world coordinate system; the calculation method can be known by those skilled in the art without creative labor, which will not be repeated here;

[0100] 7. According to the two groups of world coordinates, the mapping relationship of the two groups of world coordinate systems is calculated; that is, the positional relationship of the two world coordinate systems is obtained; the calculation method can be known by those skilled in the art without creative labor, which will not be repeated here;

[0101] 8. Convert the world coordinate system of one of the cameras to the world coordinate system of the other camera through the mapping relationship; that is, the pose relationship calibration of the two adjacent cameras is completed; ​

[0102] 9. Set a camera's world coordinate system as a target coordinate system;

[0103] 10. Convert all cameras' world coordinate systems into the target coordinate system through cascading calculation, and complete the full-section calibration of the tunnel full-section multi-vision sensor.

[0104] The specific steps for image stitching calibration of the multi-view linear array camera group are as follows:

[0105] 1. Install the multi-view linear array camera group on a sliding table; the multi-view linear array camera group can move left and right with the sliding table;

[0106] 2. Set a point-shaped calibration paper on the inner wall of the tunnel simulation environment; preferably, the point-shaped calibration paper can be pasted on the inner wall of the tunnel simulation environment;

[0107] 3. Control the multi-view linear array camera group to move with the sliding table, and at the same time, the multi-view linear array camera group shoots the point-shaped calibration paper image;

[0108] 4. Adjust the positions of the multi-view linear array camera group and the point-shaped calibration paper to ensure that the image origin center can be recognized from the shot point-shaped calibration paper image; that is, the preparation work before calibration is done well; Figure 5 to build the calibration environment for the image stitching of the multi-view linear array camera group.

[0109] 5. Control the multi-view linear array camera group to move with the sliding table, and at the same time, the multi-view linear array camera group shoots the point-shaped calibration paper image again;

[0110] 6. Recognize the image coordinates of the image origin center according to the shot point-shaped calibration paper image;

[0111] 7. Calculate the mapping relationship of the image coordinate systems of adjacent cameras in the multi-view linear array camera group according to the image coordinates of the image origin center;

[0112] 8. Convert the image coordinate system of one camera into the image coordinate system of another camera through the mapping relationship; that is, the image stitching calibration of adjacent cameras in the multi-view linear array camera group is completed;

[0113] 9. Set a camera's image coordinate system as a target coordinate system;

[0114] 10. Convert the image coordinate systems of all cameras in the multi-view linear array camera group into the set target coordinate system through cascading calculation, and complete the image stitching calibration of the multi-view linear array camera.

[0115] Embodiment Three

[0116] Embodiment Three is a specific calculation scheme for the cascading calculation in Embodiment One, like Figure 7As shown, the camera group includes camera one, camera two, camera three, camera four and camera five arranged in sequence, the world coordinate system (or image coordinate system) of camera one is C1, the world coordinate system (or image coordinate system) of camera two is C2, the world coordinate system (or image coordinate system) of camera three is C3, the world coordinate system (or image coordinate system) of camera four is C4, and the world coordinate system (or image coordinate system) of camera five is C5; and camera one is set as the target coordinate system.

[0117] M21 is a mapping relationship matrix of camera two to camera one, M32 is a mapping relationship matrix of camera three to camera two, M43 is a mapping relationship matrix of camera four to camera three, and M54 is a mapping relationship matrix of camera five to camera four.

[0118] Then the conversion of camera two to camera one is C21=C2*M21.

[0119] The conversion of camera three to camera one is C31=C3*M32*M21.

[0120] The conversion of camera four to camera one is C41=C4*M43*M32*M21.

[0121] The conversion of camera five to camera one is C51=C5*M54*M43*M32*M21.

[0122] The above-described embodiments only express the specific implementation of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the protection scope of the present application. It should be noted that for ordinary skilled persons in the art, without departing from the technical concept of the present application, a number of modifications and improvements can be made, which are all within the protection scope of the present application.

Claims

1. A global calibration method for tunnel detection, characterized in that, include: Step S1: Construct a tunnel simulation environment; Step S2: Construct the calibration environment for the tunnel simulation environment; Step S3: Within the calibration environment, calculate the coordinate mapping relationship of the calibration object; The calibration object is then calibrated based on the coordinate mapping relationship. The calibration of the calibration object includes: full-section calibration and image stitching calibration; The calibration environment for constructing the tunnel simulation environment in step S2 includes: A needle-shaped target is set for single-camera calibration, and the needle-shaped target moves in a direction perpendicular to each camera. A toothed target is set for calibrating the pose relationship between two adjacent cameras, such that the toothed target is located in the overlapping area of ​​the two adjacent cameras; A simulated calibration target is set for camera group image calibration, so that the camera can recognize the origin center of the simulated calibration target; the simulated calibration target is an origin-shaped calibration paper, which is a calibration paper printed with an array of circular marks; The calibration of the calibration object includes: Each camera acquires images and identifies the corresponding coordinates of the images. The corresponding coordinates include the image coordinates and world coordinates of the needle-shaped target, the image coordinates of the serrated target, and the image coordinates of the simulated calibration target. The full-section calibration includes: The single camera is calibrated to obtain the transformation relationship from the single camera's image coordinate system to the world coordinate system; Based on the transformation relationship, the image coordinates of images captured by adjacent cameras are transformed into world coordinates; Based on the world coordinates of two sets of images taken by adjacent cameras, calculate the world coordinate system mapping relationship of the adjacent cameras and complete the pose relationship calibration of the adjacent cameras. Cascaded computation was used to complete the full-section calibration of all cameras; The image stitching calibration includes: Image stitching calibration is performed on a multi-view linear array camera group. Origin-shaped calibration paper is set on the inner wall of the tunnel simulation environment. The image coordinates of the origin center are identified based on the captured image of the origin-shaped calibration paper. Based on the image coordinates of the origin center, the mapping relationship of the image coordinate systems of adjacent cameras in the multi-view linear array camera group is calculated. The image coordinate system of one camera is transformed to the image coordinate system of another camera through the mapping relationship. The image coordinate system of one camera is set as the target coordinate system. Through cascade calculation, the image coordinate systems of all cameras in the multi-view linear array camera group are transformed to the set target coordinate system, thus completing the image stitching calibration of the multi-view linear array camera.

2. The tunnel detection global calibration method according to claim 1, characterized in that, Step S3 includes: Calculate the coordinate mapping relationship of the calibrated object; Based on the coordinate mapping relationship, cascaded calculations are used to calibrate the calibration object.

3. The tunnel detection global calibration method according to claim 1, characterized in that, The calibration of the calibration object in step S3 further includes: laser calibration; The calibration environment for constructing the tunnel simulation environment in step S2 further includes: A planar target is set for laser calibration, such that the horizontal line of the planar target coincides with the light beam of the laser.

4. The tunnel detection global calibration method according to claim 3, characterized in that, The laser calibration includes: With the aid of software algorithms, the angle between the horizontal marking lines and the laser lines in the same imaging camera, as well as the vertical width distribution of the line composition area under horizontally equal intervals, are calculated. When the included angle is 0 degrees and the vertical width distribution is within a preset threshold range, the laser calibration is completed.

5. A tunnel detection global calibration device, characterized in that, A method for global calibration of tunnel detection as described in any one of claims 1-4, comprising: Mobile device, calibration device mounted on the mobile device for calibrating the calibration object, tunnel full-section multi-vision sensor for acquiring calibration data from the calibration device, and multi-view linear array camera group for acquiring tunnel images; The calibration device includes targets, wherein the targets include a planar target for calibrating a laser, a needle-shaped target for calibrating a single camera, and a toothed target for calibrating the pose relationship between adjacent cameras.

6. A tunnel detection global calibration device according to claim 5, characterized in that, The horizontal markings on the planar target coincide with the light rays from the laser, thereby achieving laser calibration. The needle-shaped target moves in the direction perpendicular to the camera to calibrate each camera and obtain the image coordinates and world coordinates of each image; The toothed target is located in the overlapping area of ​​two adjacent cameras, which enables the calibration of the pose relationship between the two adjacent cameras and obtains the world coordinates of the corresponding corner points in the overlapping area.

Images