Method for autonomous measurement of pose of tripod structure of solar panel on non-cooperative spacecraft

[0019] A method for autonomously measuring the position and attitude of a tripod structure of solar panels on a non-cooperative spacecraft. The image processing steps are as follows:

[0020] (1) Use Canny operator to extract the edge of the image: smooth the image with a Gaussian filter, then use the finite difference of the first-order partial derivative to calculate the magnitude and direction of the gradient, then perform non-maximum suppression on the gradient magnitude, and finally use Double threshold algorithm to detect and connect edges;

[0021] (2) Perform average filtering to reduce pixel noise;

[0022] (3) Image binarization: that is, gray scale division, which refers to setting a gray value. If the gray level of the image itself is greater than it, it will be a bright spot, and if the gray value is lower than the set value, it will be Dark spots, so a binary image is obtained.

[0023] A method for autonomously measuring the position and attitude of a tripod structure of solar panels on a non-cooperative spacecraft. The specific implementation of the triangle detection based on the sliding Hough change is:

[0024] Using the point-line duality, that is, using the parameter equation ρ=xcos(θ)+ysin(θ) to correspond a point on the image plane to a curve on the parameter plane, where ρ is drawn from the origin to a straight line The length of the perpendicular, θ is the angle between the perpendicular and the positive x-axis. In the original coordinate system, all points of a straight line have the same slope and intercept. This feature is manifested in the parametric coordinate system, that is, all curves are gathered at the same point (θ 0 ,ρ 0 ), see figure 1. The characteristic of the inscribed circle from the triangle: the distance between the center of the inscribed circle and the three sides is equal, select a small window to slide on the image, specify the center of the window as the origin of the coordinate, and perform a Hough change on the image in the window. When the window slides to the origin of the coordinate system, When the centers of the triangle inscribed circles coincide, the triangle will have three peak points in the corresponding Hough parameter space, and the absolute values ​​of the ρ coordinates of the three peak points are equal (see figure 2 ), from which the image coordinates of the vertex and the center of the triangle can be determined. The specific implementation steps are as follows:

[0025] (1) Preprocess the image, extract features and calculate its gradient direction;

[0026] (2) Quantize the (θ, ρ) parameter plane and set the two-dimensional accumulation matrix H(θ i ,ρ j );

[0027] (3) Edge refinement, that is, keep extreme points in the gradient direction of the edge points, and eliminate those non-extreme points;

[0028] (4) For each edge point, with its gradient direction ψ as the center, set a small area [ψ-θ 0 ,ψ+θ 0 ], where θ 0 For experience, generally 5°~10° can be taken. In this area, Δθ is used as the step size, the corresponding ρ value is calculated for the quantized value of θ in each interval, and a unit value is added to the corresponding cumulative matrix element;

[0029] (5) Perform threshold detection on the cumulative matrix, and use points greater than the threshold as candidate points;

[0030] (6) Take the maximum points in the candidate points in the cumulative matrix (namely the parameter space) as the required peak points, and the coordinates of the parameter space corresponding to these points are the parameters of the detected straight line.

[0031] (7) Determine whether the length and angle of three line segments in the set of straight line segments detected in the previous step meet the following conditions (because the straight line segments in the digital image cannot accurately meet its mathematical definition, thresholds need to be set, in the following conditional expression T a That is the set threshold):

[0032] A |ρ a -ρ b | a ,|ρ a -ρ c | a ,|ρ c -ρ b | a (The distance from the center of the inscribed circle to the three sides is equal);

[0033] B AB×AO, AO×AC, OB×OC are all parallel to AB×AC in the same direction (the center of the window is inside the polygon surrounded by three line segments);

[0034] C a+b> c,a+c> b,b+c> a (the sum of any two sides of the triangle is greater than the third side);

[0035] If the detected straight line segment meets the above conditions, the three line segments form a triangle, and the three sides and three corners of the triangle can be obtained. The triangle can also be determined, and the center of the triangle inscribed circle can also be obtained. Coordinates (x 0 ,y 0 ), the distance ρ from the center of the circle to the three sides and the angle θ between the line where ρ is located and the x-axis, based on which the coordinates of the three vertices of the triangle can be determined, and the positioning of the characteristic points of the tripod can be completed.

[0036] A method for autonomously measuring the pose of a tripod structure of solar panels on a non-cooperative spacecraft. The specific implementation of the relative pose calculation based on the P4P algorithm is:

[0037] Imaging model from the center perspective (see image 3 ) Describe the coordinate Q of the feature point on the spacecraft in the camera coordinate system c (x c ,y c ,z c ) And its image coordinates are U o The mapping relationship between (u,v):

[0038] z c u v 1 = f u 0 u o 0 0 f v v o 0 0 0 1 0 x c y c z c 1 - - - ( 1 )

[0039] Where f u , F v Is the effective focal length, u o , V o The main point of the camera. The coordinate Q of the feature point in the target coordinate system is also known o (x o ,y o ,z o ) And its coordinate Q in the camera coordinate system c (x c ,y c ,z c The rigid body relationship between) can be described as:

[0040] x c y c z c 1 = R T 0 T 1 x o y o z o 1 - - - ( 2 )

[0041] among them R = r 0 r 1 r 2 r 3 r 4 r 5 r 6 r 7 r 8 By Q o (x o ,y o ,z o ) To Q c (x c ,y c ,z c ) Rotation transformation matrix of coordinates, T = t 1 t 2 t 4 Is Q o (x o ,y o ,z o ) Coordinate origin to Q c (x c ,y c ,z c ) The translation vector of the coordinate origin.

[0042] Bring equation (2) into equation (1), and set λ=z c Get:

[0043] λ u v 1 = f u 0 u o 0 0 f v v o 0 0 0 1 0 R T 0 T 1 x o y o z o 1 - - - ( 3 )

[0044] When the image coordinates (u, v) and space coordinates (x o ,y o ,z o ), then the equation (3) can be purely quantified to get the 1 ~λ 4 , R 0 ~r 8 , T 1 ~t 3 , F u , F v , U o , V o A total of 12 linear equations with 20 unknown variables, and the rotation matrix R is known to be orthogonal, and 15 unknown variables are left after removing the constraint relationship. The actual meaning of the scale factor λ is the projection of the distance from the object point to the optical center in the direction of the optical axis. When the four feature points are coplanar, λ 1 =λ 2 =λ 3 =λ 4 , At this time, there are only 12 unknown variables left in the equation system, which has a unique linear solution. That is, the internal parameters of the camera, the target rotation matrix R and the translation matrix T can all be solved.

[0045] Specific experimental case:

[0046] The present invention will now be further described in conjunction with implementation cases and drawings.

[0047] The invention builds a ground experimental platform for autonomous measurement of non-cooperative spacecraft's pose, and the structure block diagram is as follows Figure 4 Shown. The experimental platform mainly includes space background, on-board computer, CCD camera, target spacecraft, etc. The technical parameters of related modules are as follows:

[0048] 1) The space background is constructed by a multi-projection immersive virtual environment application platform for virtual manufacturing-the "five-channel 220-degree large-scale ring screen stereo display system", which has a wide field of view, high resolution, and overall image color consistency Good, good immersion, etc., which can simulate a more realistic space environment;

[0049] 2) The on-board computer is simulated by an ordinary personal computer, which is mainly responsible for image processing and calculation;

[0050] 3) The camera uses a Kinect color camera with a resolution of 1280×960, an effective viewing distance of 1.2~3.5m, and its internal parameters: equivalent focal length f u = F v =1600, the principal point coordinate u c = U c =(400,300), the image size is 1280×960;

[0051] 4) The target spacecraft is the model of the Chang'e-2 lunar exploration satellite (scale 1:30), the model is made of zinc-aluminum alloy, the weight is 1.3kg, and the size parameters are: 680mm (length) × 220mm (height) × 125mm.

[0052] 5) The main structure is a cube of 56mm×66mm×70mm. Since the present invention uses a satellite model, it is converted according to the scale, and the close distance is about 0 to 1 m. The size of the solar panel tripod is 52mm×52mm×55mm, which is in line with reality.

[0053] Based on the relative pose measurement method of the present invention and the above-mentioned related technical parameters, the internal and external parameters of the camera should be calibrated in advance during the experiment, and the camera should be used to take pictures of the target spacecraft (complete imaging of any solar sailboard tripod), and use The method of the present invention automatically completes the extraction of feature points, and calculates the pose parameters between the target spacecraft coordinate system and the camera coordinate system.