A method for measuring the speed of a moving target based on a rotating mirror high-speed camera

Abstract

The application discloses a kind of motion target speed measurement methods based on rotating mirror type high-speed camera, first select two images, and the three-dimensional coordinates of motion target are calculated in different time through two images, then all motion targets are located in same plane, and then the camera projection point constraint of plane three-dimensional point under two visual angles is constructed, subsequently the pixel coordinates of 3D point on a camera on the same plane in space and homography matrix H are used to obtain the unique corresponding pixel coordinates of the 3D point in another image;The application calculates the virtual parallax diagram of motion target by using background information, and then realizes the three-dimensional reconstruction of motion target feature point, then the real three-dimensional coordinates of target in world coordinate system are calculated according to calibration data, so that the displacement of target can be calculated, and then the motion speed of target can be calculated by combining camera frame rate, realize the function of shooting the motion process of target with large field of view and high resolution, and be suitable for being widely promoted and used.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of motion target speed measurement, in particular to a motion target speed measurement method based on a rotating mirror type high-speed camera. BACKGROUND

[0002] The binocular camera system is used for shooting the same target by two cameras, obtaining the parallax map of the target and performing feature point matching pair, so that the extrinsic matrix of the two cameras can be obtained, and the feature points on the target are reconstructed in three dimensions by triangulation. For a rigid object, the displacement is the displacement of the object shape without change, which can be regarded as the displacement of a particle, and the displacement amount can be represented by the difference between the three-dimensional coordinates of the object at two different times in the world coordinate system.

[0003] At present, the speed of high-speed motion target is generally measured by visual method, which generally uses a binocular camera system for shooting and analysis. However, this method is limited by the field of view of the camera, so that the range that can be shot is small, and because the quality of the camera is large, it is not realistic to directly move the camera body to realize the tracking and shooting of the high-speed motion target. Therefore, a motion target speed measurement method based on a rotating mirror type high-speed camera is needed. SUMMARY

[0004] The present application aims to overcome the shortcomings of the prior art, and better solve the problem that the range that can be shot is small due to the limitation of the field of view of the camera, and it is not realistic to directly move the camera body to realize the tracking and shooting of the high-speed motion target because the quality of the camera is large. A motion target speed measurement method based on a rotating mirror type high-speed camera is provided, which calculates the virtual parallax map of the motion target by using background information, realizes the three-dimensional reconstruction of the feature points of the motion target, calculates the real three-dimensional coordinates of the target in the world coordinate system according to the calibration data, so as to calculate the displacement amount of the target, and then calculate the motion speed of the target combined with the frame rate of the camera, realizing the function of shooting the motion process of the target with large field of view and high resolution.

[0005] In order to achieve the above purpose, the technical scheme adopted by the present application is:

[0006] A motion target speed measurement method based on a rotating mirror type high-speed camera, comprising the following steps,

[0007] Step (A), selecting two images, and calculating the three-dimensional coordinates of the motion target at different times through the two images;

[0008] Step (B), locating the motion target on the same plane, and constructing the camera projection point constraint of the three-dimensional points on the plane at two viewing angles;

[0009] Step (C), using the pixel coordinates of the 3D point on the same plane in space on a camera and the homography matrix H to find the unique corresponding pixel coordinates of the 3D point in another image;

[0010] Step (D), based on the camera projection point constraint, calculating the corresponding homography matrix H of the projection of the target plane on the two virtual cameras T , and then according to the homography matrix H, finding the virtual corresponding points;

[0011] Step (E), according to the virtual corresponding points obtained, reconstructing the three dimensions of the target point;

[0012] Step (F), based on the three-dimensional coordinates of the target point, calculating the actual displacement of the target to complete the velocity measurement of the moving target.

[0013] Preferably, step (A), selecting two frames of images and calculating the three-dimensional coordinates of the moving target at different times through the two images, wherein the two frames of images are selected by the virtual cameras O c1 and O c2 formed by the rotating mirror, and the high-speed moving target moves from M1 to M2, and the feature points P on the moving target move from P1 to P2, and the projections of the moving target on the cameras O c1 and O c2 are p1 and p2 respectively.

[0014] Preferably, step (B), moving the target to the same plane, and then constructing the camera projection point constraint of the three-dimensional points on the plane under two perspectives, wherein the constraint relationship between the camera projection points p(u, v, 1) and p'(u', v, 1) of the three-dimensional points on the plane under two perspectives is shown in formula (1),

[0015] p=Hp'(1)

[0016] Wherein, H is a homography matrix.

[0017] Preferably, step (C), using the pixel coordinates of the 3D point on the same plane in space on a camera and the homography matrix H to find the unique corresponding pixel coordinates of the 3D point in another image, wherein the homography matrix H is shown in formula (2),

[0018]

[0019] Wherein, K is the intrinsic matrix of the camera, R and t are the rotation and translation matrices between the virtual cameras respectively, n T is the unit normal vector of the target plane, and d is the distance from the plane to the origin of the world coordinate system.

[0020] Preferably, step (D), based on the camera projection point constraint, calculating the corresponding homography matrix H of the projection of the target plane on the two virtual camerasT According to the homography H, virtual corresponding points p1' and p2' are obtained, and the specific steps are as follows,

[0021] Step (D1), a rotation matrix R and a translation matrix t between virtual cameras are calculated, wherein the rotation matrix R between virtual cameras is calculated according to formula (1), B and the translation matrix t is calculated according to formula (2), B The specific calculation steps are as follows,

[0022] Step (D11), feature points on the static background of the image are selected for feature matching, if the feature points on the background are not from a plane, the projection points of the feature points on the camera satisfy the relationship shown in formula (3) and formula (4),

[0023] x' T Fx=0(3)

[0024] F=K -T t × RK -1 (4)

[0025] wherein x and x' are projection points, and F is a 3*3 matrix with a rank of 2;

[0026] Step (D12), the matrix expression of F is shown in formula (5) and formula (6), then corresponding points are searched on the backgrounds of the images taken by the two virtual cameras, and F is obtained according to formula (6) by using eight groups of corresponding points, and the rotation matrix R and the translation matrix t between virtual cameras are obtained by decomposing F, B , B

[0027]

[0028] uu'f 11 +uv'f 12 +uf 13 +vu'f 21 +vv'f 22 +vf 23 +u'f 31 +v'f 32 +f 33 =0(6);

[0029] Step (D13), if the feature points on the background are from the same plane, the projection of the three-dimensional points on the camera satisfies formula (2), and the homography H is shown in formula (7) and formula (8),

[0030]

[0031]

[0032] ​Step (D14), the homography H is a 3x3 matrix, and is decomposed according to the homography decomposition method of Faugeras to obtain the rotation matrix R between the virtual cameras B and the translation matrix t B ;

[0033] Step (D2), the unit normal vector of the target motion plane and the distance from the plane to the origin of the world coordinate system are calculated, and the homography between the target projection planes is solved. Specifically, the target region is selected in the image for feature point matching, and the homography H is solved by formula (7) and formula (8), and the parameter unit normal vector n of the target plane is obtained by decomposing the homography T T and the distance d from the plane to the origin of the world coordinate system T , and then the corresponding homography H of the target plane between the projection planes of the two cameras is obtained by formula (3) T , as shown in formula (9),

[0034]

[0035] Step (D3), the virtual corresponding points p1' and p2' are obtained, and the specific steps are as follows,

[0036] Step (D31), the projection point p1 of the feature point P in the virtual camera O c1 and the virtual corresponding point p1' in the virtual camera O c2 are obtained by formula (2), as shown in formula (10),

[0037] p1' = H T p1 (10);

[0038] Step (D32), the projection point p2 of the feature point P in the virtual camera O c2 and the virtual corresponding point p2' in the virtual camera O c1 are obtained by formula (2), as shown in formula (11),

[0039] p2' = H T p2 (11).

[0040] Preferably, step (E), according to the obtained virtual corresponding points, the three-dimensional of the target point is reconstructed, and the specific steps are as follows,

[0041] Step (E1), assuming that the world coordinate system coincides with the camera coordinate system with O c1 as the origin of the coordinate system, the mapping relationship of the three-dimensional point P to the pixel points p1(u, v, 1) and p2(u', v', 1) is obtained by the pinhole imaging model, as shown in formula (12),

[0042]

[0043] wherein K is the intrinsic matrix of the camera;

[0044] Step (E2), assuming K[I 0] = [m1 m2 m3] and K[R t] = [m1' m2' m3'], then formula (12) can be transformed into formula (13) as shown,

[0045]

[0046] Preferably, step (F), based on the three-dimensional coordinates of the target points, the actual displacement of the target is calculated, and the velocity measurement of the moving target is completed, and the specific steps are as follows,

[0047] Step (F1), the three-dimensional coordinates p1(x1, y1, z1) and p2(x2, y2, z2) of the target at different times are obtained, and the distance d between the two points is calculated, as shown in formula (14),

[0048]

[0049] Step (F2), according to the calculated distance d1 between the target feature points and the measured actual distance d1' between the target feature points, the scale factor s is calculated, wherein d1' is calculated by using the average displacement of the K feature points as shown in formula (15), and the scale factor s is as shown in formula (16),

[0050]

[0051]

[0052] The present application has the following advantages:

[0053] (1), first, the high-speed rotating mirror is placed in front of the high-speed camera, and according to the motion law of the target, the mirror is controlled to make the mirror real-time align with the target, then the view angle of the camera is changed by the rotation of the mirror to realize the tracking and shooting of the target, which is equivalent to a series of virtual cameras for multi-angle shooting of the moving target, effectively realizing the function of shooting the target motion process with large field of view and high resolution, which provides conditions for image analysis.

[0054] (2), by using the background information to calculate the virtual parallax map of the moving target, the three-dimensional reconstruction of the moving target feature points is realized, then the real three-dimensional coordinates of the target in the world coordinate system are calculated according to the calibration data, so that the displacement of the target can be calculated, and then the motion speed of the target can be calculated combined with the camera frame rate, effectively realizing the function of quickly measuring the speed of the moving target, and the measurement speed is faster and the effect is better.

[0055] (3), the application has the advantages of low site requirement, simple experimental equipment construction and high robustness, and can still measure the target speed with high precision under relatively simple conditions, which lays a foundation for measuring the target speed in the scene of the weapon range, the workshop and the ball game in the harsh environment, the average relative error of the displacement measurement experiment of the method is 2.03%, the feasibility and accuracy of the method are effectively realized, and the method has the advantages of scientific and reasonable method, strong applicability and good effect. BRIEF DESCRIPTION OF DRAWINGS

[0056] Figure 1 is a rotating mirror type high-speed camera model of the application;

[0057] Figure 2 is a moving target projection model of the application;

[0058] Figure 3 is a virtual corresponding point pixel coordinate calculation flowchart of the application;

[0059] Figure 4 is a static experimental platform experiment diagram of the application;

[0060] Figure 5 is a rotating mirror type high-speed camera platform experiment diagram of the application;

[0061] Figure 6 is a high-speed moving target image experiment result diagram of the application;

[0062] Figure 7 is a target speed curve experiment result diagram of the application. DETAILED DESCRIPTION

[0063] The application will be further described below in conjunction with the drawings of the specification.

[0064] A moving target speed measurement method based on a rotating mirror type high-speed camera of the application comprises the following steps,

[0065] As shown in Figure 2 , step (A), two images are selected, and the three-dimensional coordinates of the moving target at different times are calculated through the two images, wherein the two images are selected by the virtual cameras O c1 and O c2 formed by the rotating mirror, and the high-speed moving target moves from M1 to M2, and the feature point P on the moving target moves from P1 to P2, and the projections of the moving target on the cameras O c1 and O c2 are p1 and p2 respectively.

[0066] As shown in Figure 1 , the rotating mirror type camera model of the application comprises a high-speed vibration mirror, a vibration mirror control system and a high-speed camera.

[0067] Step (B), if all the moving targets are located in the same plane, then the camera projection point constraints of the planar 3D points under two views are constructed, where the constraints between the camera projection points p(u, v, 1) and p'(u', v, 1) of the planar 3D points under two views are shown in equation (1),

[0068] p = H p'(1)

[0069] where H is a homography matrix, when the surface depth of the target is much smaller than the distance between the camera and the target, that is, when the relative scene depth is much smaller than the distance between the camera and the target during shooting, the field of view is small; the homography matrix H constrains the 2D coordinates of the 3D space points in the same plane on two pixel planes.

[0070] Step (C), using the pixel coordinates of the 3D points on the same plane in space on one camera and the homography matrix, the unique corresponding pixel coordinates of the 3D points in another image are obtained, where the homography matrix H is shown in equation (2),

[0071]

[0072] where K is the intrinsic matrix of the camera, R and t are the rotation and translation matrices between the virtual cameras, respectively, n T is the unit normal vector of the target plane, and d is the distance from the plane to the origin of the world coordinate system.

[0073] As Figure 3 shown, step (D), based on the camera projection point constraints, the projection of the target plane on the two virtual cameras corresponds to the homography matrix H T , and then the virtual corresponding points p1' and p2' are obtained according to the homography matrix H, where the specific steps of obtaining the virtual corresponding points p1' and p2' are as follows,

[0074] Step (D1), the rotation and translation matrices between the virtual cameras are calculated, where the rotation matrix R B and the translation matrix t B between the virtual cameras are calculated, and the specific calculation steps are as follows,

[0075] Step (D11), feature points on the static background of the image are selected for feature matching, if the feature points on the background are not from a plane, their projection points on the camera satisfy the relationship shown in equation (3) and equation (4),

[0076] x' T Fx = 0 (3)

[0077] F = K -T t × R K -1 (4)

[0078] Wherein, x and x' are projection points, F is a 3*3 matrix with rank 2; since SIFT feature points have invariance to rotation and scale, and have good robustness to noise, view transformation and illumination change, the application preferentially selects SIFT feature points as matching points.

[0079] Step (D12), the matrix expression of F is shown in formula (5) and formula (6), then corresponding points are searched on the backgrounds of the images taken by the two virtual cameras, and F is obtained according to formula (6) by using eight groups of corresponding points, and then the rotation matrix R B and the translation matrix t B between the virtual cameras are obtained by decomposing F

[0080]

[0081] uu'f 11 +uv'f 12 +uf 13 +vu'f 21 +vv'f 22 +vf 23 +u'f 31 +v'f 32 +f 33 =0(6);

[0082] Wherein, the solution of F matrix can be obtained by a direct linear solution method;

[0083] Step (D13), if the feature points on the background come from the same plane, the projection of three-dimensional points on the camera satisfies formula (2), and the homography matrix H is shown in formula (7) and formula (8),

[0084]

[0085]

[0086] Wherein, h9 is a scale factor, which can be recorded as 1, then H is a matrix with 8 degrees of freedom, and each group of corresponding points has two equations, and it is known that at least four groups of corresponding points are needed to solve the homography matrix H;

[0087] Step (D14), the homography matrix H is a 3*3 matrix, and the rotation matrix R B and the translation matrix t B between the virtual cameras are obtained by decomposing the homography matrix H according to the Faugeras homography matrix decomposition method

[0088] Step (D2), calculating the unit normal vector of the target motion plane and the distance from the plane to the origin of the world coordinate system, and solving the homography matrix between the target projection planes, specifically, selecting a target region in the image for feature point matching, and solving the homography matrix H from formula (7) and formula (8), and then decomposing the homography matrix to obtain the parameter unit normal vector n of the target plane T T and the distance d from the plane to the origin of the world coordinate system T , and then the corresponding homography matrix H of the target plane between the two camera projection planes can be obtained from formula (3) T , as shown in formula (9),

[0089]

[0090] Step (D3), obtaining the virtual corresponding points p1' and p2', the specific steps are as follows,

[0091] Step (D31), the projection point p1 of the feature point P in the virtual camera O c1 and the virtual corresponding point p1' in the virtual camera O c2 pixel coordinates can be obtained from formula (2), as shown in formula (10),

[0092] p1' = H T p1 (10);

[0093] Step (D32), the projection point p2 of the feature point P in the virtual camera O c2 and the virtual corresponding point p2' in the virtual camera O c1 pixel coordinates can be obtained from formula (2), as shown in formula (11),

[0094] p2' = H T p2 (11).

[0095] Step (E), according to the obtained virtual corresponding points, the three-dimensional of the target point is reconstructed, the specific steps are as follows,

[0096] Step (E1), assuming that the world coordinate system coincides with the camera coordinate system with O c1 as the origin of the coordinate system, then the mapping relationship of the three-dimensional point P to the pixel points p1(u, v, 1), p2(u', v', 1) is shown in formula (12) according to the pinhole imaging model,

[0097]

[0098] wherein K is the intrinsic matrix of the camera;

[0099] Step (E2), let K[I 0] = [m1m2m3], K[R t] = [m1'm2'm3'], then formula (12) can be transformed into formula (13) as shown,

[0100]

[0101] Wherein, the corresponding projection points of the spatial three-dimensional points on two images and the camera internal and external parameter matrix can be obtained from formula (13), so that the three-dimensional coordinates of the pixel points in the world coordinate system can be obtained, and therefore the three-dimensional coordinates of the characteristic points P1 and P2 in the middle can be obtained. Figure 2

[0102] Step (F), based on the three-dimensional coordinates of the target points, the actual displacement of the target is calculated, and the velocity measurement of the moving target is completed, and the specific steps are as follows,

[0103] Step (F1), the distance d between two points is obtained from the three-dimensional coordinates p1(x1, y1, z1) and p2(x2, y2, z2) of the target at different time, as shown in formula (14),

[0104]

[0105] Step (F2), according to the calculated distance d1 between the target characteristic points and the measured actual distance d1' between the target characteristic points, the scale factor s is calculated, wherein the average displacement of the K characteristic points is used to calculate d1' as shown in formula (15), and the scale factor s is as shown in formula (16),

[0106]

[0107]

[0108] In order to better illustrate the use effect of the present application, a specific embodiment of the present application is described below;

[0109] As Figure 4 shown, the calibration plate is installed on the high-precision mechanical arm to translate, and a picture is taken by changing the camera angle every time it moves, simulating the shooting of the target by the rotating mirror camera, and the displacement of the target is measured by the method proposed in the present application. The average relative error is 2.03%, and the experimental results are shown in Table 1,

[0110] Table 1

[0111]

[0112]

[0113] As Figure 5 ​As shown in the figure, a rotating mirror type high-speed camera platform is built in the laboratory, and the platform is used to shoot simulated high-speed moving targets. Figure 6 As shown in the figure, Figure 6 The black and white points represent the real projection of the feature points on the camera and the calculated virtual projection of the target, respectively.

[0114] When measuring the speed, two adjacent frames are selected as a group for analysis, 12 groups of 24 adjacent frames are selected for analysis, the displacement of the target is calculated, the motion speed of the target is calculated in combination with the frame rate of the camera between each group of images, and the speed transformation curve of the target in the tracking range is drawn, as shown in the figure. Figure 7 As shown in the figure.

[0115] In summary, the motion target speed measurement method based on the rotating mirror type high-speed camera of the application can calculate the virtual parallax map of the moving target by using background information, realize three-dimensional reconstruction of the feature points of the moving target, then calculate the real three-dimensional coordinates of the target in the world coordinate system according to the calibration data, so as to calculate the displacement of the target, and then calculate the motion speed of the target in combination with the camera frame rate, effectively realizing the function of quickly measuring the motion speed of the target, and the measurement speed is faster and the effect is better. The application has the advantages of low site requirement, simple experimental equipment construction and high robustness, and can still measure the target speed with high precision under relatively simple conditions, which lays a foundation for measuring the target speed in the scene of the weapon range, the workshop and the ball game under the relatively harsh environment. The average relative error of the displacement measurement experiment of the application method is 2.03%, effectively realizing the feasibility and accuracy of the method.

[0116] The basic principles, main features and advantages of the application are shown and described above. Those skilled in the art should understand that the application is not limited by the above examples, and the above examples and descriptions in the specification are only to illustrate the principles of the application. Without departing from the spirit and scope of the application, various changes and improvements can be made to the application, and these changes and improvements all fall within the scope of the claimed application. The scope of protection of the application is defined by the appended claims and their equivalents.

Claims

1. A method for measuring the speed of a moving target based on a rotating mirror high-speed camera, characterized in that: The method comprises the following steps, Step (A), select two images, and calculate the three-dimensional coordinates of the moving object at different time through the two images, wherein the two images are selected from the virtual cameras O c1 and O c2 formed by the rotating mirror, and the high-speed moving object moves from M1 to M2, and the feature point P on the moving object moves from P1 to P2, and the projections of the moving object on the cameras O c1 and O c2 are p1 and p2 respectively; Step (B), when the moving targets are all located in the same plane, a camera projection point constraint of the planar three-dimensional points under two view angles is constructed, wherein the constraint relationship between the camera projection points p(u, v, 1) and p'(u', v', 1) of the planar three-dimensional points under two view angles is shown in formula (1), p = Hp'(1) Wherein, H is a homography matrix; Step (C), the pixel coordinates of the 3D point on one camera and the homography matrix H are used to obtain the unique corresponding pixel coordinates of the 3D point in another image, wherein the homography matrix H is shown in formula (2), where K is the intrinsic matrix of the camera, R and t are the rotation and translation matrices between the virtual cameras, n T is the unit normal vector of the plane where the target is located, and d is the distance from the plane to the origin of the world coordinate system. Step (D), based on the camera projection point constraint, calculating the projection of the target plane on the two virtual cameras corresponding to the homography matrix H T and the virtual corresponding points are obtained according to the homography matrix H. Step (E), the three-dimensional coordinates of the target point are calculated based on the virtual corresponding points, and the three-dimensional reconstruction of the target point is completed; Step (F), the actual displacement of the target is calculated based on the three-dimensional coordinates of the target point, and the velocity measurement of the moving target is completed.

2. The method of claim 1, wherein: Step (D), based on the camera projection point constraint, calculate the homography matrix H corresponding to the projection of the target plane on the two virtual cameras T According to the homography matrix H, the virtual corresponding points p1' and p2' are calculated. Step (D1), calculating the rotation and translation matrix between virtual cameras, where the rotation matrix R B and the translation matrix t B The specific calculation steps are as follows, Step (D11), the feature points on the static background of the image are selected for feature matching, if the feature points on the background are not from a plane, the projection points thereof on the camera satisfy the relationship shown in formula (3) and formula (4), x' T Fx = 0 (3) F = K -T t × RK -1 (4) Wherein, x and x' are projection points, and F is a 3*3 matrix with a rank of 2; Step (D12), the matrix expression of F is shown in equation (5) and equation (6), and then the corresponding points are searched on the backgrounds of the images taken by the two virtual cameras, and F is solved according to equation (6) by using eight groups of the corresponding points, and then F is decomposed to obtain the rotation matrix R and the translation matrix t between the virtual cameras B and B ​ uu'f 11 +uv'f 12 +uf 13 +vu'f 21 +vv'f 22 +vf 23 +u'f 31 +v'f 32 +f 33 = 0 (6); Step (D13), if the feature points on the background are from the same plane, the projection of the three-dimensional point on the camera satisfies formula (2), and the homography matrix H is shown in formula (7) and formula (8), Step (D14), the homography H is a 3x3 matrix, and decomposition according to the homography decomposition method of Faugeras can obtain the rotation matrix R between virtual cameras B and the translation matrix t B ; Step (D2), calculating the unit normal vector of the target motion plane and the distance from the plane to the origin of the world coordinate system, and solving the homography matrix between the target projection planes, which is specifically selecting a target region in the image for feature point matching, and solving the homography matrix H from formula (7) and formula (8), and then decomposing the homography matrix to obtain the parameter unit normal vector n of the target plane T T and the distance d from the plane to the origin of the world coordinate system T , and then the corresponding homography matrix H of the target plane between the projection planes of the two cameras can be obtained from formula (3) T , as shown in formula (9), Step (D3), the virtual corresponding points p1' and p2' are obtained, and the specific steps are as follows, Step (D31), the projection point p1 of the feature point P in the virtual camera O c1 and the virtual corresponding point p1' in the virtual camera O c2 pixel coordinates can be obtained from equation (2), as shown in equation (10), p1' = H T p1(10); Step (D32), the projection point p2 of the feature point P in the virtual camera O c2 and the virtual corresponding point p2' in the virtual camera O c1 pixel coordinates can be obtained from equation (2), as shown in equation (11), p2' = H T p2(11).

3. The method of claim 2, wherein: Step (E), the three-dimensional coordinates of the target point are calculated based on the virtual corresponding points, and the three-dimensional reconstruction of the target point is completed; Step (E1), assuming that the world coordinate system coincides with the camera coordinate system with O c1 as the origin, the mapping relationship of the three-dimensional point P to the pixel points p1(u, v, 1) and p2(u', v', 1) is shown in equation (12) according to the pinhole imaging model. Wherein, K is an internal parameter matrix of the camera; Step (E2), if K[I 0] = [m1m2m3] and K[R t] = [m1'm2'm3'], formula (12) can be transformed into formula (13), 4. The method of claim 3, wherein: Step (F), the actual displacement of the target is calculated based on the three-dimensional coordinates of the target point, and the velocity measurement of the moving target is completed, and the specific steps are as follows, Step (F1), the distance d1 between two points is obtained from the three-dimensional coordinates p1(x1, y1, z1) and p2(x2, y2, z2) of the target at different time instants, and the distance d1 is shown in formula (14), Step (F2), the scale factor s is calculated according to the calculated distance d1 between the target feature points and the measured actual distance d1' between the target feature points, wherein the average displacement of the K feature points is used to calculate d1' as shown in formula (15), and the scale factor s is shown in formula (16),

Images