A Vehicle Shape Measurement System and Method Based on Two Concentric Spherical Features

Abstract

This invention discloses a vehicle topography measurement system and method based on double concentric spherical features, aiming to solve the problem of vehicle topography measurement based on double concentric spherical features. The vehicle topography measurement system based on double concentric spherical features mainly consists of an external camera (1), a tripod (2), a double concentric calibration plate (3), a support base (4), an adjustable support platform (5), a calibration block (6), an internal camera (7), a laser line projector (8), and a two-dimensional target (9). The threaded through holes on the three sides of the calibration block (6) are respectively threaded and fixedly connected to three threaded rods. The other ends of the three threaded rods pass through three wing nuts, three through holes of the double concentric calibration plate (3), and three wing nuts, respectively. The distance between the three double concentric calibration plates (3) and the center of the calibration block (6) is consistent. This invention provides a vehicle topography measurement method and system based on double concentric spherical features that can be used for large-scale spatial detection and has stable performance.

Description

Technical Field

[0001] This invention relates to a measuring device in the field of automobile inspection, and more specifically, to an automobile morphology measurement system and method based on dual concentric spherical features. Background Technology

[0002] In recent years, with the increasing prevalence of vehicles in people's production and daily lives, the safe operation of vehicles has become increasingly important to society. Measuring the morphology of a vehicle is a crucial step in vehicle inspection and diagnosis, and the accuracy of the measurement significantly impacts the vehicle's operational status. Commonly used morphology measurement methods are mainly divided into contact and non-contact methods. Contact measurement methods are prone to damaging the vehicle surface, while non-contact methods such as eddy current measurement and ultrasonic measurement have high requirements for the smoothness of the surface of the measured object. Therefore, machine vision methods have gradually become the mainstream method for morphology measurement. To achieve feature point reconstruction of vehicle morphology without motion constraints, a vehicle morphology measurement system and method based on biconcentric spherical features were designed. Summary of the Invention

[0003] This invention addresses the issues of accuracy and efficiency in automotive 3D information measurement, as well as the equipment requirements for using visual measurement methods. It proposes a vehicle shape measurement system and method based on biconcentric spherical features. The method mainly consists of two industrial cameras, a laser line projector whose relative position to the cameras remains constant, three calibration plates with double-layered concentric circle textures, a support base, an adjustable support platform, and a cubic support block. Based on biconcentric spherical features and structured light, it employs an active vision method to measure vehicle contour information.

[0004] Referring to the accompanying drawings, the present invention is implemented using the following technical solution:

[0005] The vehicle topography measurement system based on the dual concentric spherical features includes an external camera, a tripod, a dual concentric calibration plate, a support base, an adjustable support platform, a calibration block, an internal camera, a laser line projector, and a two-dimensional target.

[0006] A circular base is placed on the ground. The threaded hole in the middle of the adjustable support platform is threadedly fixed to the threaded rod of the support base. The threaded through hole of the calibration block is threadedly fixed to the threaded rod on the top of the support base and contacts the upper surface of the adjustable support platform. The threaded through holes on the three sides of the calibration block are threadedly fixed to the three threaded rods respectively. The other ends of the three threaded rods pass through the three wing nuts, the through holes of the three double concentric calibration plates, and the three wing nuts respectively. The end face of the wing nuts contacts the surface of the double concentric calibration plates. The distance between the three double concentric calibration plates and the center of the calibration block is consistent. The laser projector is welded and fixedly connected to the screw. The internal camera is fixedly connected to the screw through U-bolts. The tripod is placed on a horizontal ground. The threaded hole at the bottom of the external camera is threadedly fixed to the stud at the top of the tripod. The two-dimensional target is placed on the ground.

[0007] The support base described in the technical solution is welded together from a circular base and a threaded rod;

[0008] The adjustable support platform described in the technical solution is a circular part made of steel plate with threaded holes.

[0009] The calibration block described in the technical solution is a cubic part made of steel, with mutually perpendicular intersecting threaded through holes machined at the center of each face of the cube;

[0010] The double concentric calibration plate described in the technical solution is a 5mm thick plate made of acrylic material. A circular through hole is machined in the center of the plate, and double concentric circular target paper is attached to the outer surface of the double concentric calibration plate.

[0011] The laser line projector described in the technical solution is a cylindrical part capable of emitting a laser beam from a single surface;

[0012] The internal camera described in the technical solution is an industrial camera with a filter, and the bandpass of the filter is consistent with the wavelength of the laser line projector.

[0013] The external camera described in the technical solution is an industrial camera with a filter, and the bandpass of the filter is consistent with the wavelength of the laser line projector.

[0014] The two-dimensional target described in the technical solution is a flat plate made of acrylic material, with a checkerboard target paper attached to its outer surface.

[0015] The specific steps of the vehicle topography measurement method based on double concentric spherical features are as follows:

[0016] Step 1: Image acquisition based on a vehicle topography measurement method using concentric spherical features:

[0017] The tripod is placed on the ground, the external camera is fixed on the top of the tripod, the support base is placed on the ground, and three double concentric calibration plates are fixed on the support base by screws, calibration blocks, and adjustable support platforms. The laser projector is fixed on the screws. When the laser projector is turned on, the laser plane emitted by the laser projector intersects with the vehicle to be inspected by a laser intersection line. An image is captured using the internal camera, which contains the laser projection lines of the laser projector and the vehicle to be inspected. At the same time, an image is captured using the external camera, which contains at least two double concentric calibration plates.

[0018] Step 2: Extraction of concentric spherical features in the vehicle topography measurement method:

[0019] First, a dataset of concentric spherical target images is created using images of concentric spherical features captured by an external camera. The SSD algorithm is then used to detect the concentric spherical feature regions. The loss function of the algorithm is...

[0020]

[0021] Where N is the number of positive samples of the biconcentric spherical features in the prior bounding box, and b i，j ∈{0,1} represents the matching degree between the i-th prior box and the j-th dataset ground truth box. It is the softmax activation result of the predicted class values ​​of the two concentric spherical features, l i g i These are the shape and position parameters of the prior bounding box and the true bounding box of the dataset, respectively. smooth() is the L1 loss function. The double concentric spherical feature region can be obtained using the SSD algorithm.

[0022] Based on the detected biconcentric spherical feature regions, all concentric circular points within the region are extracted. An adaptive clustering algorithm based on DBSCAN is used to automatically classify the circular points of different concentric circles. Then, the following bimodal Gaussian constraint function is applied for further selection of the circular points.

[0023]

[0024] Where A1 and A2 are the mixing coefficients of the two concentric circle distributions, m1 and m2 are the means of the two distributions, and σ1 and σ2 are the standard deviations of the two distributions. The selected circumferential points are obtained using the bimodal Gaussian constraint function.

[0025] The Legendre moment method is used to extract sub-pixel-level filtered circumferential point coordinates from the projected image of the double concentric calibration plate, and then rotates. The (p+q) order Legendre moment after the angle is

[0026]

[0027] Where, p p(x) is a Legendre polynomial, and f(x, y) represents the pixel grayscale on the projected image of the double concentric calibration plate. The sub-pixel coordinates of the edge points of the concentric circles on the double concentric calibration plate can be obtained as follows:

[0028]

[0029]

[0030] Where, l=(8L′) 20 +5L′ 00 ) / (15L′ 10 The sub-pixel coordinates of the edge points of the concentric circle can be used to obtain the coordinates z of the double spherical feature points on the double concentric calibration plate;

[0031] Step 3: Calibration of the pose relationship between the internal camera and the dual-concentric calibration plate in the vehicle topography measurement method based on dual-concentric spherical features:

[0032] First, the dual-concentric calibration plate is placed within the field of view of the external camera, and the two-dimensional target is placed within the common field of view of both the external and internal cameras. Image information of the two-dimensional target is simultaneously acquired by both cameras. Based on the correspondence between the two-dimensional coordinates of the upper corner of the target and the image coordinates, and using the DLT algorithm, the intrinsic parameter matrix K of the external camera can be calculated. E The intrinsic parameter matrix K of the internal camera I The Lie algebra ψ representing the pose relationship between the external camera and the two-dimensional target E，T The Lie algebra ψ of the pose relationship between the internal camera and the two-dimensional target I，T Based on the correspondence between the 3D coordinates of feature points on the dual concentric calibration board and the image coordinates, the Lie algebra ψ of the pose relationship between the external camera and the dual concentric calibration board can be solved using the DLT calibration method. E，S Design a two-concentric spherical optimization objective function

[0033]

[0034] The Lie algebra ψ, representing the optimized transformation relationship between the external camera and the dual concentric calibration board, is obtained. E，S , where z i z j These are the coordinates of the feature points on the double-sphere surface on the double-concentric calibration plate, where r1 and r2 are the radii of the small sphere and the large sphere, respectively, and K... E It is the intrinsic parameter matrix of the external camera, π T These are the planar coordinates of the two concentric calibration plates;

[0035] After applying the coordinate system transformation relationships described above, the pose relationship between the internal camera and the biconcentric calibration plate is obtained as follows:

[0036]

[0037] Step 4: Calibration of the pose relationship between the internal camera and the laser plane projected by the laser line projector in the vehicle topography measurement method based on dual concentric spherical features:

[0038] The laser projection device is activated, and the internal camera acquires the laser intersection information between the laser plane and the two-dimensional target at different positions. The laser plane is then fitted using the following formula.

[0039]

[0040] Among them, y i K is the image coordinate of the intersection point of the laser plane and the two-dimensional target, projected by the internal camera. I This is the intrinsic parameter matrix of the internal camera obtained in the third step; using the above formula and the SVD method, the coordinates π of the laser plane in the internal camera coordinate system can be obtained;

[0041] Step 5: Reconstruction of laser intersection points based on a vehicle topography measurement method using concentric spherical features:

[0042] The vehicle to be inspected is placed within the field of view of the internal camera. The internal camera captures an image of the intersection line between the laser plane projected by the laser line projector and the vehicle body. Based on the internal parameter matrix KI of the internal camera obtained in step three and the coordinates π of the laser plane in the internal camera coordinate system obtained in step four, the coordinates of the vehicle body intersection point in the internal camera coordinate system can be obtained using the following formula.

[0043]

[0044] Based on the calibration results of step three, ψ E，S , ψ I，S The coordinates of the intersection point of the vehicle body in the internal camera coordinate system obtained by the above formula. The coordinates of the laser intersection point of the vehicle body in the external camera coordinate system can be obtained as follows:

[0045]

[0046] In the process of vehicle topography reconstruction, the fifth step is a cyclical process;

[0047] Based on the coordinates of the laser intersection points of the vehicle body in the external camera coordinate system, the world coordinates of each feature point of the vehicle on the laser plane when the dual concentric calibration plate is at the j-th position under the external camera can be calculated. By moving the dual concentric calibration plate and the laser projection device fixedly connected to it within the field of view of the external camera, the laser intersection points of other laser planes and the vehicle body can be obtained, thereby determining the world coordinates of the intersection points of the vehicle and the laser plane, and completing the vehicle shape reconstruction based on the dual concentric spherical features.

[0048] The beneficial effects of this invention are:

[0049] 1. This invention establishes a three-dimensional reconstruction model based on biconcentric spherical features in the form of Lie algebra. The Lie algebra form can avoid redundancy of degrees of freedom and ensure the orthogonality of rotation matrices during pose solving, which has a good effect on improving the accuracy and efficiency of the measurement system.

[0050] 2. This invention utilizes SSD to achieve multi-scale, automatic extraction of the target region of the dual concentric calibration plate 3; adaptively classifies candidate points of concentric circle features; and uses a bimodal Gaussian distribution to constrain and achieve rapid feature extraction based on the distribution pattern of concentric circle points, thereby improving the measurement efficiency of the vehicle shape measurement system based on dual concentric spherical features in many ways.

[0051] 3. The proposed measurement system can obtain images of the vehicle from any orientation by transforming the relative positional relationship between the system and the vehicle under test, thereby reconstructing the target and realizing the free reconstruction of the global shape of the vehicle. Attached Figure Description

[0052] Figure 1 It is an isometric drawing of the calibration system for an automotive topography measurement system based on the features of two concentric spherical surfaces;

[0053] Figure 2 It is a reconstructed isometric view of an automobile topography measurement system based on the features of two concentric spherical surfaces;

[0054] Figure 3 This is an isometric view of the double concentric calibration plate 3, support base 4, adjustable support platform 5, calibration block 6, internal camera 7, and laser line projector 8 in an automotive topography measurement system based on double concentric spherical features.

[0055] Figure 4 This is an isometric view of external camera 1 in an automotive topography measurement system based on the features of two concentric spherical surfaces;

[0056] Figure 5 This is an axonometric view of tripod 2 in an automotive topography measurement system based on the features of two concentric spherical surfaces;

[0057] Figure 6 This is an axonometric view of the double concentric calibration plate 3 in the automotive topography measurement system based on the double concentric spherical features;

[0058] Figure 7 This is an axonometric view of support 4 in an automotive morphology measurement system based on the features of two concentric spherical surfaces;

[0059] Figure 8 This is an isometric view of the adjustable support platform 5 in an automotive topography measurement system based on the features of two concentric spherical surfaces;

[0060] Figure 9This is an axonometric view of calibration block 6 in an automotive topography measurement system based on the features of two concentric spherical surfaces;

[0061] Figure 10 This is an isometric view of the internal camera 7 in an automotive topography measurement system based on the features of two concentric spherical surfaces;

[0062] Figure 11 This is an isometric view of the laser line projector 8 in an automotive topography measurement system based on the features of two concentric spherical surfaces;

[0063] Figure 12 This is an axonometric view of a two-dimensional target 9 in an automotive topography measurement system based on the features of two concentric spherical surfaces;

[0064] Figure 13 This is a flowchart of the calibration process in a vehicle topography measurement method based on the features of two concentric spherical surfaces;

[0065] Figure 14 This is a flowchart of the pose relationship calibration between the external camera 1 and the dual-concentric calibration plate 3 in the automobile topography measurement method based on the features of a dual-concentric spherical surface;

[0066] In the diagram: 1. External camera, 2. Tripod, 3. Double concentric calibration plate, 4. Support base, 5. Adjustable support platform, 6. Calibration block, 7. Internal camera, 8. Laser projector, 9. Two-dimensional target. Detailed Implementation

[0067] The present invention will now be described in further detail with reference to the accompanying drawings:

[0068] See Figures 1 to 12 The vehicle topography measurement system based on the dual concentric spherical features includes an external camera 1, a tripod 2, a dual concentric calibration plate 3, a support base 4, an adjustable support platform 5, a calibration block 6, an internal camera 7, a laser line projector 8, and a two-dimensional target 9.

[0069] The support base 4 is welded from a circular base and a threaded rod. The circular base is placed on the ground. The adjustable support platform 5 is a circular part made of steel plate with threaded holes. The threaded hole in the middle of the adjustable support platform 5 is threadedly fixed to the threaded rod of the support base 4. The calibration block 6 is a cubic part made of steel. Each face of the cube has a threaded through hole that is perpendicular to each other. The threaded through hole of the calibration block 6 is threadedly fixed to the threaded rod on the top of the support base 4 and contacts the upper surface of the adjustable support platform 5. The threaded through holes on the three sides of the calibration block 6 are threadedly fixed to the three threaded rods respectively. The double concentric calibration plate 3 is a 5mm thick plate made of acrylic material. The plate has a circular through hole in its center. The outer surface of the double concentric calibration plate 3 is covered with double concentric circular target paper. The other ends of the three threaded rods pass through three wing nuts and three... The double concentric calibration plate 3 has through holes and three wing nuts. The end face of the wing nuts contacts the surface of the double concentric calibration plate 3. The distance between the three double concentric calibration plates 3 and the center of the calibration block 6 is consistent. The laser projector 8 is a cylindrical part that can emit a laser beam in one area. The laser projector 8 is welded and fixedly connected to the screw. The internal camera 7 is an industrial camera with a filter. The bandpass of the filter is consistent with the wavelength of the laser projector 8. The internal camera 7 is fixedly connected to the screw by U-bolts. The tripod 2 is placed on a horizontal ground. The threaded hole at the bottom of the external camera 1 is fixedly connected to the stud at the top of the tripod 2. The external camera 1 is an industrial camera with a filter. The bandpass of the filter is consistent with the wavelength of the laser projector 8. The two-dimensional target 9 is a flat plate made of acrylic material with checkerboard target paper attached to its outer surface. The two-dimensional target 9 is placed on the ground.

[0070] See Figures 13 to 14 The vehicle topography measurement method based on dual concentric spherical features provided by this invention can be divided into the following five steps:

[0071] Step 1: Image acquisition based on a vehicle topography measurement method using concentric spherical features:

[0072] Tripod 2 is placed on the ground, external camera 1 is fixed on the top of tripod 2, support base 4 is placed on the ground, three double concentric calibration plates 3 are fixed on support base 4 by screws, calibration blocks 6, and adjustable support platform 5, laser line projector 8 is fixed on screws, laser line projector 8 is turned on, the laser plane emitted by laser line projector 8 intersects the vehicle under inspection with a laser intersection line, internal camera 7 is used to capture an image, the image contains the laser projection line of laser line projector 8 and the vehicle under inspection, and external camera 1 is used to capture an image, the image contains at least two double concentric calibration plates 3;

[0073] Step 2: Extraction of concentric spherical features in the vehicle topography measurement method:

[0074] First, a dataset of concentric spherical target images is created using images of concentric spherical features captured by external camera 1. The SSD algorithm is then used to detect the concentric spherical feature regions. The loss function of the algorithm is...

[0075]

[0076] Where N is the number of positive samples of the biconcentric spherical features in the prior bounding box, and b i，j ∈{0,1} represents the matching degree between the i-th prior box and the j-th dataset ground truth box. It is the softmax activation result of the predicted class values ​​of the two concentric spherical features, l i g i These are the shape and position parameters of the prior bounding box and the true bounding box of the dataset, respectively. smooth() is the L1 loss function. The double concentric spherical feature region can be obtained using the SSD algorithm.

[0077] Based on the detected biconcentric spherical feature regions, all concentric circular points within the region are extracted. An adaptive clustering algorithm based on DBSCAN is used to automatically classify the circular points of different concentric circles. Then, the following bimodal Gaussian constraint function is applied for further selection of the circular points.

[0078]

[0079] Where A1 and A2 are the mixing coefficients of the two concentric circle distributions, m1 and m2 are the means of the two distributions, and σ1 and σ2 are the standard deviations of the two distributions. The selected circumferential points are obtained using the bimodal Gaussian constraint function.

[0080] The Legendre moment method is used to extract sub-pixel-level filtered circumferential point coordinates from the projected image of the double concentric calibration plate 3, and then rotates. The (p+q) order Legendre moment after the angle is

[0081]

[0082] Where, p p (x) is a Legendre polynomial, and f(x, y) represents the pixel grayscale on the projected image of the double concentric calibration plate 3. The sub-pixel coordinates of the edge points of the concentric circles on the double concentric calibration plate 3 can be obtained as follows:

[0083]

[0084]

[0085] Where, l=(8L′) 20 +5L′ 00 ) / (15L′ 10The sub-pixel coordinates of the edge points of the concentric circle can be used to obtain the coordinates z of the double spherical feature points on the double concentric calibration plate 3;

[0086] Step 3: Calibration of the pose relationship between the internal camera 7 and the dual-concentric calibration plate 3 in the vehicle topography measurement method based on dual-concentric spherical features:

[0087] First, the dual concentric calibration plate 3 is placed within the field of view of the external camera 1, and the two-dimensional target 9 is placed within the common field of view of the external camera 1 and the internal camera 7. Image information of the two-dimensional target 9 is simultaneously acquired by both cameras. Based on the correspondence between the two-dimensional coordinates of the upper corner of the two-dimensional target 9 and the image coordinates, and using the DLT algorithm, the intrinsic parameter matrix K of the external camera 1 can be solved. E The intrinsic parameter matrix KI of the internal camera 7, and the Lie algebra ψ representing the pose relationship between the external camera 1 and the two-dimensional target 9. E，T The Lie algebra ψ of the pose relationship between the internal camera 7 and the two-dimensional target 9 I，T Based on the correspondence between the three-dimensional coordinates of feature points on the dual concentric calibration plate 3 and the image coordinates, the Lie algebra ψ of the pose relationship between the external camera 1 and the dual concentric calibration plate 3 can be solved using the DLT calibration method. E，S Design a two-concentric spherical optimization objective function

[0088]

[0089] Find the Lie algebra ψ, representing the optimized transformation relationship between external camera 1 and the dual concentric calibration board 3. E，S , where z i z j These are the coordinates of the double-sphere feature points on the double-concentric calibration plate 3, where r1 and r2 are the radii of the small sphere and the large sphere, respectively, and K... E It is the intrinsic parameter matrix of external camera 1, π T These are the planar coordinates of the double concentric calibration plate 3;

[0090] After applying the coordinate system transformation relationships described above, the pose relationship between the internal camera 7 and the biconcentric calibration plate 3 is obtained as follows:

[0091]

[0092] Step 4: Calibration of the pose relationship between the internal camera 7 and the laser plane projected by the laser projector 8 in the vehicle topography measurement method based on dual concentric spherical features:

[0093] The laser projection device 8 is activated, and the internal camera 7 is used to acquire the laser intersection information between the laser plane and the two-dimensional target 9 at different positions. The laser plane is then fitted using the following formula.

[0094]

[0095] Among them, y i K is the image coordinate of the intersection point of the laser plane and the two-dimensional target 9 projected by the internal camera 7. I This is the intrinsic parameter matrix of the internal camera 7 obtained in the third step; using the above formula and the SVD method, the coordinates π of the laser plane in the coordinate system of the internal camera 7 can be obtained;

[0096] Step 5: Reconstruction of laser intersection points based on a vehicle topography measurement method using concentric spherical features:

[0097] The vehicle to be inspected is placed within the field of view of the internal camera 7. The internal camera 7 is used to acquire an image of the intersection line between the laser plane projected by the laser line projector 8 and the vehicle body. Based on the internal parameter matrix K of the internal camera 7 obtained in the third step... I And given the coordinates π of the laser plane in the internal camera 7 coordinate system obtained in step 4, the coordinates of the vehicle body intersection point in the internal camera 7 coordinate system can be obtained using the following formula.

[0098]

[0099] Based on the calibration results of step three, ψ E，S , ψ I，S The coordinates of the intersection point of the vehicle body obtained by the above formula in the coordinate system of the internal camera 7. The coordinates of the laser intersection point of the vehicle body in the coordinate system of external camera 1 can be obtained as follows:

[0100]

[0101] In the process of vehicle topography reconstruction, the fifth step is a cyclical process;

[0102] Based on the coordinates of the laser intersection points of the vehicle body in the coordinate system of external camera 1, the world coordinates of each feature point of the vehicle on the laser plane when the double concentric calibration plate 3 is at the j-th position under external camera 1 can be calculated. By moving the double concentric calibration plate 3 and the laser projection device 8 fixedly connected to it within the field of view of external camera 1, the laser intersection points of other laser planes and the vehicle body can be obtained, thereby determining the world coordinates of the intersection points of the vehicle and the laser plane, and completing the reconstruction of the vehicle shape based on the double concentric spherical features.

Claims

1. A measurement method for an automotive topography measurement system based on dual concentric spherical features, characterized in that, The vehicle topography measurement system based on the dual concentric spherical features includes an external camera (1), a tripod (2), a dual concentric calibration plate (3), a support base (4), an adjustable support platform (5), a calibration block (6), an internal camera (7), a laser line projector (8), and a two-dimensional target (9). The circular base is placed on the ground. The threaded hole in the middle of the adjustable support platform (5) is threadedly fixed to the threaded rod of the support base (4). The threaded through hole of the calibration block (6) is threadedly fixed to the threaded rod at the top of the support base (4) and contacts the upper surface of the adjustable support platform (5). The threaded through holes on the three sides of the calibration block (6) are threadedly fixed to the three threaded rods respectively. The other end of the three threaded rods passes through the three wing nuts, the through holes of the three double concentric calibration plates (3) and the three wing nuts respectively. The end face of the wing nut contacts the surface of the double concentric calibration plate (3). The distance between the three double concentric calibration plates (3) and the center of the calibration block (6) is consistent. The laser projector (8) is welded and fixedly connected to the screw. The internal camera (7) is fixedly connected to the screw by saddle bolts. The tripod (2) is placed on the horizontal ground. The threaded hole at the bottom of the external camera (1) is threadedly fixed to the stud at the top of the tripod (2). The two-dimensional target (9) is placed on the ground. The specific steps of the measurement method are as follows: Step 1: Image acquisition based on a vehicle topography measurement method using concentric spherical features: Tripod (2) is placed on the ground, external camera (1) is fixed on the top of tripod (2), support base (4) is placed on the ground, three double concentric calibration plates (3) are fixed on support base (4) by screws, calibration blocks (6) and adjustable support platform (5), laser line projector (8) is fixed on screws, laser line projector (8) is turned on, the laser plane emitted by laser line projector (8) intersects the vehicle under inspection with a laser intersection line, use internal camera (7) to collect an image, the image contains the laser projection line of laser line projector (8) and the vehicle under inspection, at the same time use external camera (1) to collect an image, the image contains at least two double concentric calibration plates (3); Step 2: Extraction of concentric spherical features in the vehicle topography measurement method: First, a dataset of concentric spherical target images is created by selecting images of concentric spherical features acquired by an external camera (1). The SSD algorithm is then used to detect the concentric spherical feature regions. The loss function of the algorithm is... Where N is the number of positive samples of the biconcentric spherical features in the prior bounding box, and b i,j ∈{0,1} represents the matching degree between the i-th prior box and the j-th dataset ground truth box. It is the softmax activation result of the predicted class values ​​of the two concentric spherical features, l i g i These are the shape and position parameters of the prior bounding box and the true bounding box of the dataset, respectively. smooth() is the L1 loss function. The double concentric spherical feature region can be obtained using the SSD algorithm. Based on the detected biconcentric spherical feature regions, all concentric circular points within the region are extracted. An adaptive clustering algorithm based on DBSCAN is used to automatically classify the circular points of different concentric circles. Then, the following bimodal Gaussian constraint function is applied for further selection of the circular points. Where A1 and A2 are the mixing coefficients of the two concentric circle distributions, m1 and m2 are the means of the two distributions, and σ1 and σ2 are the standard deviations of the two distributions. The selected circumferential points are obtained using the bimodal Gaussian constraint function. The Legendre moment method is used to extract sub-pixel-level filtered circular point coordinates from the projection image of the double concentric calibration plate (3). The (p+q) order Legendre moment after rotation by φ is: Where, p p (x) is the Legendre polynomial, and f(x, y) represents the pixel grayscale on the projected image of the double concentric calibration plate (3). The sub-pixel coordinates of the edge points of the concentric circles on the double concentric calibration plate (3) can be obtained as follows: in, The sub-pixel coordinates of the edge points of the concentric circle can be used to obtain the coordinates z of the double spherical feature points on the double concentric calibration plate (3); Step 3: Calibration of the pose relationship between the internal camera (7) and the dual-concentric calibration plate (3) in the vehicle topography measurement method based on dual-concentric spherical features: First, the dual concentric calibration plate (3) is placed within the field of view of the external camera (1), and the two-dimensional target (9) is placed within the common field of view of the external camera (1) and the internal camera (7). The image information of the two-dimensional target (9) is acquired simultaneously by the two cameras. Based on the correspondence between the two-dimensional coordinates of the upper corner of the two-dimensional target (9) and the image coordinates and the DLT algorithm, the intrinsic parameter matrix K of the external camera (1) can be solved. E The intrinsic parameter matrix K of the internal camera (7) I The Lie algebra ψ of the pose relationship between the external camera (1) and the two-dimensional target (9) E,T The Lie algebra ψ of the pose relationship between the internal camera (7) and the two-dimensional target (9) I,T Based on the correspondence between the three-dimensional coordinates of the feature points on the bi-concentric calibration plate (3) and the image coordinates, the Lie algebra ψ of the pose relationship between the external camera (1) and the bi-concentric calibration plate (3) can be solved using the DLT calibration method. E,S Design a two-concentric spherical optimization objective function The Lie algebra ψ is obtained as the transformation relationship between the optimized external camera (1) and the bi-concentric calibration plate (3). E,S , where z i z j The coordinates of the double-sphere feature points on the double-concentric calibration plate (3) are given, where r1 and r2 are the radii of the small sphere and the large sphere, respectively, and K is the coordinate of the double-sphere feature points on the double-concentric calibration plate (3). E It is the intrinsic parameter matrix of the external camera (1), π T These are the plane coordinates of the double concentric calibration plate (3); After the coordinate system transformation relationship described above, the pose relationship between the internal camera (7) and the dual concentric calibration plate (3) is obtained as follows: Step 4: Calibration of the pose relationship between the internal camera (7) and the laser plane projected by the laser projector (8) in the vehicle topography measurement method based on dual concentric spherical features: Turn on the laser projection device (8), use the internal camera (7) to obtain the laser intersection information of the laser plane and the two-dimensional target (9) at different positions, and use the following formula to fit the laser plane. Among them, y i K is the image coordinate of the intersection point of the laser plane and the two-dimensional target (9) projected by the internal camera (7). I It is the intrinsic parameter matrix of the internal camera (7) obtained in the third step; using the above formula and the SVD method, the coordinates π of the laser plane in the coordinate system of the internal camera (7) can be obtained; Step 5: Reconstruction of laser intersection points based on a vehicle topography measurement method using concentric spherical features: The vehicle to be inspected is placed within the field of view of the internal camera (7). The internal camera (7) is used to acquire the image of the intersection line between the laser plane projected by the laser line projector (8) and the vehicle body. Based on the internal parameter matrix K of the internal camera (7) obtained in the third step... I And the coordinates π of the laser plane in the coordinate system of the internal camera (7) obtained in step four can be used to obtain the coordinates of the intersection point of the vehicle body in the coordinate system of the internal camera (7) according to the following formula. Based on the calibration results of step three, ψ E,S , ψ I,S The coordinates of the intersection point of the vehicle body obtained by the above formula in the coordinate system of the internal camera (7) The coordinates of the laser intersection point of the vehicle body in the coordinate system of the external camera (1) can be obtained as follows: In the process of vehicle topography reconstruction, the fifth step is a cyclical process; Based on the coordinates of the laser intersection point of the vehicle body in the coordinate system of the external camera (1), the world coordinates of each feature point of the car on the laser plane when the double concentric calibration plate (3) is at the j-th position under the external camera (1) can be calculated. By moving the double concentric calibration plate (3) and the laser projection device (8) fixedly connected to it within the field of view of the external camera (1), the laser intersection points of other laser planes and the car body can be obtained, thereby determining the world coordinates of the intersection point of the car and the laser surface, and completing the reconstruction of the car shape based on the double concentric spherical features.

2. The vehicle topography measurement method based on double concentric spherical features according to claim 1, characterized in that, The support base (4) is welded together from a circular base and a threaded rod.

3. The vehicle topography measurement method based on double concentric spherical features according to claim 1, characterized in that, The adjustable support platform (5) is a circular part made of steel plate with threaded holes.

4. The vehicle topography measurement method based on double concentric spherical features according to claim 1, characterized in that, The calibration block (6) is a cubic part made of steel, with intersecting threaded holes machined at the center of each face of the cube.

5. The vehicle topography measurement method based on double concentric spherical features according to claim 1, characterized in that, The double concentric calibration plate (3) is a 5mm thick plate made of acrylic material. A circular through hole is machined in the center of the plate. Double concentric target paper is attached to the outer surface of the double concentric calibration plate (3).

Images