Formation satellite attitude determination method based on infrared monocular vision recognition

Abstract

The application discloses a kind of formation satellite attitude determination methods based on infrared monocular vision identification, including have: step one, the heat spot setting of atmospheric sail satellite is carried out on the ground, the relative position between each heat spot is measured;Each heat spot is obtained using infrared camera, and heat spot is regarded as feature point in infrared image;Then the pixel position of each feature point is demarcated in infrared image;Step two, according to saturation and brightness, feature point cluster is obtained from infrared image, then image-total feature point is screened from feature point cluster using weighted algorithm;Step three, the coordinate of image-total feature point is mapped to the number of heat spot of atmospheric sail satellite, then the coordinate of image-total feature point is substituted into pose solution using Kalman filter and is solved, and the coordinate of image-total feature point in camera coordinate system is output;Step four, the attitude of atmospheric sail satellite is calculated by the coordinate in camera coordinate system.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of formation satellite attitude determination, more particularly, to a formation satellite attitude determination method based on infrared monocular vision recognition. BACKGROUND

[0002] Satellite formation flight is a form commonly used in the field of aerospace at present, which refers to a form that multiple spacecraft maintain certain relative positions and shapes in flight and communicate and work cooperatively with each other through inter-satellite links. Formation satellites can greatly improve the efficiency of space missions and can realize distributed InSAR formation satellites, three-dimensional imaging, electronic reconnaissance on the ground and other tasks, and are currently developing towards high precision and low cost.

[0003] At present, the relative position and posture measurement methods of formation multi-satellites mainly include relative and absolute state measurement methods. The former includes GPS measurement, GPS-like measurement, laser ranging, ground station measurement and control and other methods. GPS and GPS-like measurement need to carry signal transceivers on formation satellites, radar ranging also needs to carry large receiving antennas in terms of mass and volume, and ground station measurement and control needs to occupy more measurement and control resources. The latter is to obtain the relative position and posture by calculating the difference between the absolute positions of each satellite measured by star sensors, magnetometers, inertial attitude sensors and other means. Although the above methods can meet the actual application of satellites to some extent, they have great limitations for micro-satellite formation, constellation networking and low-cost formation satellites of private enterprises. The required sensor devices will increase the mass and power consumption budget of formation satellites, and devices with high positioning accuracy will also greatly increase the development cost of the formation system. SUMMARY

[0004] The present application provides a formation satellite attitude determination method based on infrared monocular vision recognition to solve the problems of large mass, large volume, large power consumption and high cost of traditional measurement and control methods.

[0005] The formation satellite attitude determination method based on infrared monocular vision recognition comprises the following steps:

[0006] Step 1: setting the heat points of the atmospheric sail satellite on the ground and mapping the heat points and image feature points;

[0007] Step 11: setting the heat source of the atmospheric sail satellite on the ground;

[0008] A plurality of heat points of the atmospheric sail satellite are set on the ground, which are denoted as a heat point set HOTS; wherein the position of any heat point hot a is denoted as (x_hot a , y_hot a );

[0009] Step 12, using ground infrared camera to acquire heat point image;

[0010] The atmospheric sail satellite is placed in the effective shooting distance of the infrared camera, and the feature points of each heat point on the atmospheric sail satellite in the image are acquired; the atmospheric sail satellite image captured by the infrared camera is referred to as a satellite infrared image;

[0011] The feature points in the atmospheric sail satellite image acquired by the infrared camera are denoted as an image feature point set PC_HOTS; wherein, the pixel position of any image feature point is denoted as

[0012] Step 13, using the satellite infrared image to reverse calibrate the scale factor of the infrared camera;

[0013] In order to realize the mapping of each heat point in the image physical coordinate system S p and each image feature point in the image pixel coordinate system S uv , the coordinate system conversion relationship is used to re-calibrate the internal and external parameters of the infrared camera; the calibration result can obtain the scale factors of the x C1 and y C1 axes of the camera coordinate system S C1 ; that is, the internal parameter pixel coordinate system scale factor in the x C1 axis direction is denoted as and the internal parameter pixel coordinate system scale factor in the y C1 axis direction is denoted as

[0014] Step 2, ground calibration of atmospheric sail satellite feature points;

[0015] Step 21, reading the satellite infrared image;

[0016] Using Python language programming on the on-board computer, using Opencv library function to read a single frame image as the satellite infrared image; that is, reading PC_HOTS;

[0017] Step 22, using HSV space to acquire feature point cluster;

[0018] The saturation in the HSV space is denoted as BHD, and the saturation threshold in the HSV space is denoted as BHD 阈值 ;

[0019] The brightness in the HSV space is denoted as LD, and the brightness threshold in the HSV space is denoted as LD 阈值 ;

[0020] The feature point cluster formation condition is (BHD>BHD 阈值 ) or (LD>LD 阈值 );

[0021] The satellite infrared image PC HOTS is defined in the HSV space, and then the pixel points with saturation and brightness greater than the set threshold in the HSV space are screened, the coordinates of the pixel points are output, a feature point cluster set is formed, and is recorded as HOTS 簇 ; wherein, if there are multiple pixel points in any one feature point cluster , then

[0022] the saturation of the feature point cluster in the HSV space is

[0023] the brightness of the feature point cluster in the HSV space is

[0024] the saturation of each feature point cluster in the classified feature point cluster set HOTS 簇 is recorded as a feature point-saturation set MBHD;

[0025] the brightness of each feature point cluster in the classified feature point cluster set HOTS 簇 is recorded as a feature point-brightness set MLD;

[0026] Step 23, feature point cluster-pixel point screening with saturation and brightness weight values;

[0027] the weight value calculation formula of the saturation and the brightness is

[0028] each feature point cluster in the feature point cluster set HOTS 簇 is calculated by using the weight value calculation formula of the saturation and the brightness, and a feature point cluster-pixel point set in the satellite infrared image is obtained, and is recorded as MHOT;

[0029] Step 24, cluster center and cluster radius are used to update the weight-value feature point cluster;

[0030] cluster condition update: the pixel point range of the feature point cluster is a cluster radius, recorded as is a cluster center, recorded as the feature point cluster is updated by using and ;

[0031] each weight-value feature point cluster in the feature point cluster-pixel point set MHOT is processed by using the cluster condition update, and an updated feature point cluster-pixel point set MHOT new is obtained;

[0032] Step 25, maximum value selection of the saturation and the brightness;

[0033] The pixel points corresponding to the maximum saturation and brightness are selected to update each feature point cluster in MHOT new , and an image-total feature point set is obtained, denoted as HHOT.

[0034] Step three, coordinate calculation of the image-total feature points in the camera coordinate system.

[0035] Step 31, mapping of the image-total feature points to the heat points.

[0036] Each image-total feature point in the image-total feature point set HHOT is numbered and matched with the actual heat point HOTS on the atmospheric sail satellite through the geometric relationship of coordinate conversion.

[0037] Step 32, Kalman filtering to ensure continuous movement of the total feature points.

[0038] The image-total feature point set HHOT is subjected to Kalman filtering calculation to exclude the noise generated by the sudden image-total feature points, so that the movement of the image-total feature points in the image is continuous.

[0039] Step 33, PNP pose solution.

[0040] The image-total feature point set HHOT is solved by using a pose solution matrix, and then the total feature point pose in the camera coordinate system is obtained by combining the PNP algorithm, denoted as HHOT C1 ; wherein the parameters in the pose solution matrix have been determined in the ground camera calibration process.

[0041] Step four, calculation of the atmospheric sail satellite attitude through the coordinates of the image-total feature points in the camera coordinate system.

[0042] Step 41, determination of the positions of the image-total feature points in the camera coordinate system.

[0043] The geometric parameters η of the total feature point pose HHOT C1 in the camera coordinate system are calculated.

[0044] Step 42, determination of the position and attitude of the atmospheric sail satellite relative to the main star.

[0045] The relative position and Euler angle parameters of the satellite pose in the camera coordinate system S C1 are extracted from the geometric parameters η, and further combined with the camera installation mode to determine the relative position and Euler angle parameters of the satellite pose in the main star coordinate system S C , and subsequent attitude control is performed.

[0046] Compared with the prior art, the formation satellite attitude determination method based on infrared monocular vision recognition has the following beneficial effects:

[0047] (1). Without installing the pose determination component on the target satellite, compared with the traditional pose determination mode relying on momentum wheel, laser ranging and the like, the mass, volume, power consumption and cost are greatly reduced, and the measurement and control of a larger number of formation flying satellites is adapted.

[0048] (2). The target satellite feature point is only a thermal control material, and the asymmetric identification point is adapted to the determination of the attitude of the satellite of any configuration.

[0049] (3). The space environment temperature is low, and the thermal interference is small, so that the attitude determination through infrared image is suitable for most orbital environments, and is still applicable in the earth shadow area or the area with strong electromagnetic interference. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 is the flowchart of the formation satellite attitude determination based on the infrared monocular vision recognition of the application.

[0051] Figure 2 is the picture of the atmospheric sail satellite.

[0052] Figure 3 is the distribution schematic diagram of the heat-emitting points on the atmospheric sail satellite.

[0053] Figure 3A is another distribution schematic diagram of the heat-emitting points on the atmospheric sail satellite.

[0054] Figure 4 is the schematic diagram of the atmospheric sail satellite in the camera coordinate system.

[0055] Figure 5 is the range schematic diagram of the saturation and brightness representation in different clusters of the application.

[0056] Figure 5A is the selection schematic diagram of the feature point cluster taking one cluster as the range and the pixel point as the center of the circle.

[0057] Figure 5B is the selection schematic diagram of the feature point cluster taking another cluster as the range and the pixel point as the center of the circle.

[0058] Figure 6 is the satellite attitude control schematic diagram implemented by the method of the application. DETAILED DESCRIPTION

[0059] The technical solutions in the patent embodiment of the application will be clearly and completely described below with reference to the drawings in the patent embodiment of the application. Obviously, the described embodiments are not a limitation on the patent of the application.

[0060] Setting the heat-emitting points of the atmospheric sail satellite on the ground

[0061] In the application, refer to Figure 3 ,Figure 3A As shown, the plurality of heat points provided on the atmosphere sail satellite are expressed in a set form, denoted as HOTS, and

[0062] hot1 represents the first heat point provided on the atmosphere sail satellite.

[0063] hot2 represents the second heat point provided on the atmosphere sail satellite.

[0064] hot3 represents the third heat point provided on the atmosphere sail satellite.

[0065] hot4 represents the fourth heat point provided on the atmosphere sail satellite.

[0066] hot5 represents the fifth heat point provided on the atmosphere sail satellite.

[0067] hot6 represents the sixth heat point provided on the atmosphere sail satellite.

[0068] hot a represents the ath heat point provided on the atmosphere sail satellite.

[0069] hot b represents the bth heat point provided on the atmosphere sail satellite.

[0070] hot A represents the last heat point provided on the atmosphere sail satellite. The subscript A is the total number of heat points provided on the atmosphere sail satellite.

[0071] For convenience of illustration, the heat point hot a is also referred to as any one heat point. The heat point hot b is also referred to as any other heat point. The heat point hot a is not the same heat point as the heat point hot b .

[0072] In the present application, the position of a heat point is marked by using the body coordinate system of the atmosphere sail satellite. The position of the heat point hot a is denoted as (x_hot a , y_hot a ), x_hot a is the position of the heat point hot a on the X axis, and y_hot a is the position of the heat point hot a on the Y axis.

[0073] Similarly, the position of the heat spot hot1 is recorded as (x_hot1, y_hot1). The position of the heat spot hot2 is recorded as (x_hot2, y_hot2). The position of the heat spot hot3 is recorded as (x_hot3, y_hot3). The position of the heat spot hot4 is recorded as (x_hot4, y_hot4). The position of the heat spot hot5 is recorded as (x_hot5, y_hot5). The position of the heat spot hot6 is recorded as (x_hot6, y_hot6). The position of the heat spot hot b is recorded as (x_hot b , y_hot b ). The position of the heat spot hot A is recorded as (x_hot A , y_hot A ).

[0074] In the present application, in order to solve the position and attitude of the atmosphere sail satellite relative to the main star, the coordinates of the heat spots in the atmosphere sail satellite body coordinate system need to be converted to the main star coordinate system S C .

[0075] Image position of heat spots in the infrared camera

[0076] In the present application, the heat spots are taken as image feature points in the image obtained by the infrared camera, and the image feature points are recorded as PC_HOTS, and

[0077] represents the feature point of the heat spot hot1 in the image. The pixel position of the image feature point is recorded as is the pixel position on the u-axis in the image pixel coordinate system S uv , is the pixel position on the v-axis in the image pixel coordinate system S uv .

[0078] represents the feature point of the heat spot hot2 in the image. The pixel position of the image feature point is recorded as is the pixel position on the u-axis in the image pixel coordinate system S uv , is the pixel position on the v-axis in the image pixel coordinate system S uv .

[0079] represents the feature point of the heat spot hot3 in the image. The pixel position of the image feature point is recorded as is the pixel position on the u-axis in the image pixel coordinate system S uv , In the image pixel coordinate system S uv The pixel position on the middle v-axis.

[0080] This represents the feature point of the heat source hot4 in the image. Image feature points. The pixel position is denoted as In the image pixel coordinate system S uv Pixel position on the U-axis In the image pixel coordinate system S uv The pixel position on the middle v-axis.

[0081] This represents the feature point of the heat source hot5 in the image. Image feature points. The pixel position is denoted as In the image pixel coordinate system S uv Pixel position on the U-axis In the image pixel coordinate system S uv The pixel position on the middle v-axis.

[0082] This represents the feature point of the heat source hot6 in the image. Image feature points. The pixel position is denoted as In the image pixel coordinate system S uv Pixel position on the U-axis In the image pixel coordinate system S uv The pixel position on the middle v-axis.

[0083] Indicates the heating point (hot) a Feature points in an image. Image feature points. The pixel position is denoted as In the image pixel coordinate system S uv Pixel position on the U-axis In the image pixel coordinate system S uv The pixel position on the middle v-axis.

[0084] Indicates the heating point (hot) b Feature points in an image. Image feature points. The pixel position is denoted as In the image pixel coordinate system S uv Pixel position on the U-axis In the image pixel coordinate system S uv The pixel position on the middle v-axis.

[0085] Indicates the heating point (hot) AFeature points in image. Image feature points The pixel position of the feature point in the image is denoted as The pixel position of the feature point in the image is denoted as uv The pixel position of the feature point in the image is denoted as The pixel position of the feature point in the image is denoted as uv The pixel position of the feature point in the image is denoted as

[0086] Coordinate system conversion

[0087] In the present application, referring to the three coordinate systems shown in FIG. 1, first, the master-slave formation satellite is defined as follows: Figure 4

[0088] Inertial coordinate system S i , the inertial system S i is the same as the satellite motion space inertial system.

[0089] Master satellite coordinate system S C , the origin O C is located at the satellite center of mass, the x C axis points to the satellite running direction, the y C axis points to the origin from the center of the earth, and the z C axis is determined by the right-hand rule.

[0090] Image pixel coordinate system S uv , the u-axis is parallel to the camera plane and points to the right, the v-axis is perpendicular to the u-axis and points downward, and the coordinates are represented by (u, v). In the present application, the image pixel coordinate system S uv represents the projection of a three-dimensional space object on the image plane, and the pixels are discretized, with the coordinate origin at the top left corner of the camera image plane.

[0091] Image physical coordinate system S p , the origin O p is located at the center of the camera image, the x p axis is parallel to the u-axis of the image pixel coordinate system S uv , the y p axis is parallel to the v-axis of the image pixel coordinate system S uv , and the camera focal length is f.

[0092] Camera coordinate system S C1 , the origin O C1 is located at the camera mounting point, the x C1 axis is connected to the z C1 axis at the camera lens plane, the y C1 axis is perpendicular to the lens plane, and the relative position is determined according to the right-hand rule.

[0093] Image physical coordinate system S P is the same as the image pixel coordinate system S uv ​Both are in the imaging plane, the former is physical coordinate system, unit is mm; the latter is pixel coordinate system, unit is pixel.

[0094] In the present application, the coordinate system conversion relationship is as follows:

[0095]

[0096]

[0097] K = K in · K out (3)

[0098]

[0099] R is a 3x3 transpose matrix.

[0100] t is a 3x1 displacement matrix.

[0101]

[0102]

[0103]

[0104] is the matrix of coordinate system S i transposed to coordinate system S C .

[0105] is the matrix of coordinate system S C transposed to coordinate system S C1 .

[0106] is the matrix of coordinate system S C1 transposed to coordinate system S p .

[0107] is the matrix of coordinate system S p transposed to coordinate system S uv .

[0108] F x is the scale factor of pixel coordinate system and the proportional coefficient of focal length f.

[0109] F y is the scale factor of pixel coordinate system and the proportional coefficient of focal length f.

[0110] is the matrix of camera coordinate system S C1 x C1Scale factor of intrinsic pixel coordinate system in axis direction.

[0111] S is a camera coordinate system C1 Upper y C1 Scale factor of intrinsic pixel coordinate system in axis direction.

[0112] f is a camera focal length.

[0113] (u0, v0) is a coordinate of an image pixel coordinate system S uv origin in the image pixel coordinate system.

[0114] S is an inertial coordinate system i of the upper atmosphere sail satellite.

[0115] K is a camera parameter.

[0116] K in is a camera intrinsic parameter.

[0117] K out is a camera extrinsic parameter.

[0118] In the ground calibration process of the application, the calibration image feature point position coordinates are measured, the pixel system coordinates are determined, and the camera intrinsic parameter K in is further back calculated.

[0119] Referring to Figure 1 , a formation satellite attitude determination method based on infrared monocular vision recognition of the application comprises the following steps:

[0120] Step one, setting a heat source on the ground for the upper atmosphere sail satellite, and mapping the heat source and the image feature point;

[0121] Step 11, setting a heat source on the ground for the upper atmosphere sail satellite;

[0122] In the application, referring to Figure 2 , the main heat source on the satellite is a small rudder based on the satellite structure, so the heat source is set based on the rudder, and in addition, the line and data transmission line can also be used as the total feature point in the form of heat source, and the object demand surface of the total feature point is coated with heat control paint to distinguish the background temperature.

[0123] In the application, the multiple heat sources set on the upper atmosphere sail satellite are expressed in a set form, denoted as heat source set The position of any heat source hot a is denoted as (x_hot a , y_hot a ).

[0124] Step 12, heat point image acquisition by ground infrared camera;

[0125] The atmospheric sail satellite is placed in the effective shooting distance of the infrared camera to obtain feature points of each heat point on the atmospheric sail satellite in the image. The atmospheric sail satellite image captured by the infrared camera is referred to as a satellite infrared image.

[0126] In the present application, the heat point set The satellite pose is solved, and at least 6 heat points are required, such as Figure 3 、 Figure 3A As shown, the four feature points hot1, hot2, hot3, and hot4 are located on the front of the atmospheric sail satellite, and the two feature points hot5 and hot6 are located on the side of the atmospheric sail satellite. Further, the four front feature points (hot1, hot2, hot3, and hot4) are arranged in an asymmetric T shape to break the symmetry of the front of the atmospheric sail satellite and facilitate the calculation of Euler angles.

[0127] In the present application, the feature points in the atmospheric sail satellite image obtained by the infrared camera are denoted as an image feature point set The pixel position of any image feature point is denoted as

[0128] Step 13, reverse calibration of the scale factor of the infrared camera using the satellite infrared image;

[0129] Generally, the intrinsic parameter K in and the extrinsic parameter K out of the infrared camera are set before factory delivery. In the present application, in order to realize the mapping of each heat point in the image physical coordinate system S p and each image feature point in the image pixel coordinate system S uv , the intrinsic and extrinsic parameters of the infrared camera are re-calibrated using the coordinate system conversion relationship. The calibration result can obtain the scale factors of the x C1 and y C1 axes in the camera coordinate system S C1 . That is, the intrinsic pixel coordinate system scale factor in the x C1 axis direction is denoted as and the intrinsic pixel coordinate system scale factor in the y C1 axis direction is denoted as

[0130] Step 2, ground calibration of atmospheric sail satellite feature points;

[0131] Step 21, reading the satellite infrared image;

[0132] In the application, Python language is used for programming on the satellite computer, and Opencv library function is used to read single frame image as satellite infrared image.

[0133] Step 22, the feature point cluster is obtained by using HSV space;

[0134] In the application, the saturation in HSV space is recorded as BHD, and the saturation threshold in HSV space is recorded as BHD 阈值 .

[0135] In the application, the brightness in HSV space is recorded as LD, and the brightness threshold in HSV space is recorded as LD 阈值 .

[0136] In the application, the feature point cluster formation condition (BHD>BHD 阈值 ) or (LD>LD 阈值 ).

[0137] In the application, the satellite infrared image is defined in HSV space, then the pixel points with saturation and brightness greater than the set threshold in HSV space are screened, the coordinates of the pixel points are output, a feature point cluster set is formed, and is recorded as

[0138] In the application, there are multiple pixel points in any one feature point cluster , so that The subscript i represents the pixel point identification number in the feature point cluster . The subscript n represents the total number of pixel points in the feature point cluster . Wherein:

[0139] represents the first pixel point (referred to as the first cluster-pixel point) belonging to the feature point cluster . represents the i-th pixel point (referred to as any cluster-pixel point) belonging to the feature point cluster .

[0140] represents the last pixel point (referred to as the last cluster-pixel point) belonging to the feature point cluster .

[0141] The saturation in HSV space of the feature point cluster is and

[0142] The brightness in HSV space of the feature point cluster is and

[0143] Saturation of HSV space for pixel point

[0144] Saturation of HSV space for pixel point

[0145] Saturation of HSV space for pixel point

[0146] Brightness of HSV space for pixel point

[0147] Brightness of HSV space for pixel point Brightness of HSV space for pixel point

[0148] Brightness of HSV space for pixel point

[0149] Selected first feature point cluster is denoted by Saturation of HSV space for pixel point Brightness of HSV space for pixel point

[0150] Selected second feature point cluster is denoted by Saturation of HSV space for pixel point Brightness of HSV space for pixel point

[0151] Selected third feature point cluster is denoted by Saturation of HSV space for pixel point Brightness of HSV space for pixel point

[0152] Selected fourth feature point cluster is denoted by Saturation of HSV space for pixel point Brightness of HSV space for pixel point

[0153] Selected fifth feature point cluster is denoted by Saturation of HSV space for pixel point Brightness of HSV space for pixel point

[0154] ​​​​​represents the selected bth feature point cluster. the saturation in the HSV space of the brightness in the HSV space of

[0155] represents the selected bth feature point cluster. the saturation in the HSV space of the brightness in the HSV space of the feature point cluster has multiple pixel points, then the subscript j represents the pixel point identification number in the feature point cluster the subscript m represents the total number of pixel points in the feature point cluster represents the first pixel point belonging to the feature point cluster represents the jth pixel point belonging to the feature point cluster represents the last pixel point belonging to the feature point cluster represents the selected last feature point cluster. the saturation in the HSV space of the brightness in the HSV space of

[0156] the saturation of each feature point cluster in the classified feature point cluster set HOTS 簇 is recorded as feature point-saturation

[0157] the brightness of each feature point cluster in the classified feature point cluster set HOTS 簇 is recorded as the feature point-brightness set

[0158] Step 23, feature point cluster-pixel point screening with saturation and brightness weight values;

[0159] In the present application, the weight value calculation formula of the saturation and the brightness is as follows:

[0160]

[0161] i represents the pixel point identification number in the feature point cluster.

[0162] n represents the total number of pixel points in the feature point cluster.

[0163] q i represents the saturation weight value coefficient.

[0164] p i ​​​​This represents the brightness weighting coefficient.

[0165] Formula (5) is used to analyze the feature point cluster set. The calculations are performed on each feature point cluster to obtain the set of feature point clusters – pixels – in the satellite infrared image, denoted as .

[0166]

[0167] In this invention, formula (5) is used for feature point clusters. Calculations were performed to obtain feature point clusters in the satellite infrared image. The weighted feature point cluster, denoted as

[0168]

[0169] Represents weight-feature point clusters The first weight – the pixel.

[0170] Represents weight-feature point clusters The i-th weight pixel.

[0171] Represents weight-feature point clusters The last weight – the pixel.

[0172] In this invention, formula (5) is used for feature point clusters. Calculations were performed to obtain feature point clusters in the satellite infrared image. The weighted feature point cluster, denoted as

[0173]

[0174] Represents weight-feature point clusters The first weight – the pixel.

[0175] Represents weight-feature point clusters The j-th weight pixel.

[0176] Represents weight-feature point clusters The last weight – the pixel.

[0177] This represents a cluster of feature points assigned weights for saturation and brightness.

[0178] This represents a cluster of feature points assigned weights for saturation and brightness.

[0179] representing a feature point cluster to which a weight of saturation and brightness is given

[0180] representing a feature point cluster to which a weight of saturation and brightness is given

[0181] representing a feature point cluster to which a weight of saturation and brightness is given

[0182] representing a feature point cluster to which a weight of saturation and brightness is given

[0183] representing a feature point cluster to which a weight of saturation and brightness is given

[0184] representing a feature point cluster to which a weight of saturation and brightness is given

[0185] representing a feature point cluster to which a weight of saturation and brightness is given

[0186] Step 24, updating the weight-feature point cluster with the cluster center and the cluster radius;

[0187] In the present application, the cluster condition is updated as follows: the pixel point range of the feature point cluster is the cluster radius, denoted as is the cluster center, denoted as The feature point cluster is updated with and .

[0188] Referring to FIGS. 1-3, the feature point cluster Figure 5 , Figure 5A , Figure 5B has The feature point cluster has

[0189] The feature point cluster is acquired with as the cluster center and as the radius, obtaining the region-pixel point belonging to the pixel point , which has

[0190] and as the cluster center​ With radius to get the feature point cluster, get the region belonging to pixel point The pixel point has

[0191] Then, re-count the pixel points in the feature point cluster The pixel point belongs to the feature point cluster That is, the feature point cluster There are pixel points of the feature point cluster Pixel points

[0192] Then, re-count the pixel points in the feature point cluster The pixel point belongs to the feature point cluster That is, the feature point cluster There are pixel points of the feature point cluster Pixel points Pixel points

[0193] After traversing each pixel point in , the updated-weight-feature point cluster

[0194] Similarly, for each weight-feature point cluster in the feature point cluster-pixel point set , the cluster condition update processing is performed to obtain the updated-feature point cluster-pixel point set

[0195] The feature point cluster after updating the weight-feature point cluster is recorded as The feature point cluster after updating the weight-feature point cluster is recorded as The feature point cluster after updating the weight-feature point cluster is recorded as The feature point cluster after updating the weight-feature point cluster is recorded as The feature point cluster after updating the weight-feature point cluster is recorded as The feature point cluster after updating the weight-feature point cluster is recorded as The feature point cluster after updating the weight-feature point cluster is recorded as The feature point cluster after updating the weight-feature point cluster is recorded as The feature point cluster after updating the weight-feature point cluster is recorded as​​

[0196] Step 25: Select the maximum values ​​for saturation and brightness;

[0197] from Select the pixel corresponding to the maximum saturation and brightness values, and denote it as . And pixel coordinates Assigned to feature point cluster Denote as image - total feature points Then there is The pixel coordinates are

[0198] from Select the pixel corresponding to the maximum saturation and brightness values, and denote it as . And pixel coordinates Assigned to feature point cluster Denote as image - total feature points Then there is pixel coordinates

[0199] Therefore, we can update the pixel by selecting the pixel corresponding to the maximum value of saturation and brightness. Each feature point cluster in the image is used to obtain the total feature point set, denoted as .

[0200] Step 3: Calculate the coordinates of the total feature points in the camera coordinate system.

[0201] Step 31: Map the total feature points of the image to the hot spots;

[0202] Geometric relationships of coordinate transformation for image-total feature point set The images in the image—total feature points and actual heat points on the atmospheric sail satellite. Perform number matching.

[0203] In this invention, the total feature points in the image are numbered and mapped through the geometric relationship of coordinate transformation. Using the asymmetrical "T" geometric relationship, the numbers of four of the six total feature points, hot1, hot2, hot3, and hot4, are determined. Then, the numbers of the remaining side total feature points, hot5 and hot6, are determined through the "T" pointing. Of course, the side total feature points, hot5 and hot6, may not exist in the image.

[0204] Step 32: Kalman filtering ensures continuous movement of the total feature points;

[0205] For image – total feature point set Kallman filter is used to eliminate the noise of the image-total feature points, and make the image-total feature points move continuously in the image.

[0206] In the present application, Kallman filter is used in the image input process. In the initial stage of image input, the Kallman filter result is not input, only the Kallman filter calculation is performed, and the Kallman filter result is combined with the actual image input result for judgment and output after the Kallman filter result converges. For example, Kallman filter uses the content of pages 40-49 in the doctoral dissertation "Monocular target recognition and pose measurement algorithm research and application" by Lu Xiang, 2015.

[0207] Step 33, PNP pose solution;

[0208] The image-total feature point set HHOT is solved by using the pose solution matrix, and then the total feature point pose in the camera coordinate system is obtained by combining the PNP algorithm, which is denoted as HHOT C1 . The parameters in the pose solution matrix have been determined by the ground camera calibration process.

[0209] Step four, calculate the atmospheric sail satellite attitude through the coordinates of the image-total feature points in the camera coordinate system;

[0210] Step 41, determine the position of the image-total feature points in the camera coordinate system;

[0211] The geometric parameters η, such as Euler angles, of the total feature point pose HHOT C1 in the camera coordinate system are calculated. The content of pages 24-42 in "Formation microsatellite relative pose measurement and control based on dual quaternion" by Li Jing, 2013 can be used.

[0212] Step 42, determine the position and attitude of the atmospheric sail satellite relative to the main star;

[0213] The relative position and Euler angle parameters of the satellite pose in the camera coordinate system S C1 are extracted from the geometric parameters η, and further combined with the camera installation mode to determine the relative position and Euler angle parameters of the satellite pose in the main star coordinate system S C , and then the attitude control of the atmospheric sail satellite is performed as shown in Figure 6 . Figure 6 There are pitch, yaw, and roll attitudes of the satellite in the camera coordinate system S .

Claims

1.A method for determining the attitude of a formation satellite based on infrared monocular vision recognition, characterized in that It includes the following steps: Step 1: Set up the hot spots of the atmospheric sail satellite on the ground and map the hot spots to image feature points; Step 11: The atmospheric sail satellite sets up its heat source on the ground; Multiple hotspots were set up on the ground for the atmospheric sail satellite, and this set is denoted as the hotspot set. Any heat source The position is ( ); ; Step 12: Obtain images of the heat source using a ground-based infrared camera; Atmospheric sail satellite images captured by infrared cameras are called satellite infrared images; The feature points in the atmospheric sail satellite images acquired by the infrared camera constitute the image feature point set. Any image feature point The pixel position is ( ); Step 13: Use satellite infrared images to reverse-calibrate the scale factor of the infrared camera; To realize the physical coordinate system of the image Each heat source and the image pixel coordinate system The mapping of each image feature point is performed, and the intrinsic and extrinsic parameters of the infrared camera are recalibrated using coordinate system transformation relationships. The calibration results yield the camera coordinate system. superior Intrinsic pixel coordinate system scale factor along the axis , Intrinsic pixel coordinate system scale factor along the axis ; Step 2: Ground-based calibration of feature points for atmospheric sail satellites; Step 21: Read satellite infrared images; Using Python on the onboard computer and OpenCV library functions, single-frame images are read as satellite infrared images. ; Step 22: Obtain feature point clusters using the HSV space; The saturation in HSV space is The saturation threshold is ; The brightness in the HSV space is The brightness threshold is ; Will Defined in the HSV color space, pixels with both saturation and brightness greater than a set threshold are selected, and the coordinates of these pixels are output to form a set of feature point clusters. Any cluster of feature points There are multiple pixels in it. ; Representing feature point clusters The first in 1 pixel; ; This represents the total number of pixels in a feature point cluster. Feature point clusters The saturation in the HSV space is ; For pixels The saturation of the HSV space; Feature point clusters The brightness in the HSV space is ; For pixels The brightness of the HSV space; Classification The saturation of each feature point cluster in the data is denoted as the feature point-saturation set. ; Classification The brightness of each feature point cluster in the data is denoted as the feature point-brightness set. ; Step 23: Feature point clusters – pixel selection – assigning saturation and brightness weights; The formula for calculating the weighting of saturation and brightness is: ； Indicates the pixel identifier in the feature point cluster; This represents the saturation weighting coefficient; Indicates the brightness weighting coefficient; Using formula (5) Calculations were performed to obtain the satellite infrared image. Weights – Feature point clusters ; Represents weight-feature point clusters The Each weight value is a pixel; Using formula (5) The feature point clusters are calculated in the data to obtain the feature point cluster-pixel set in the satellite infrared image. ; Step 24: Update the weight-feature point clusters using the cluster center and cluster radius; Cluster Condition Update: Weight-Feature Point Cluster The range of pixels is the cluster radius. ; The cluster center is ;by and Update weights – feature point clusters; right Each weight-feature point cluster in the dataset undergoes cluster conditional update to obtain the updated-feature point cluster-pixel set. ; Update weights – feature point clusters The feature point clusters after that are denoted as ; Step 25: Select the maximum values ​​for saturation and brightness; from Select the pixel corresponding to the maximum saturation and brightness values. And assign to denoted as the total feature points of the image. ; Therefore, we can update the pixel by selecting the pixel corresponding to the maximum value of saturation and brightness. The image-total feature point set is obtained by combining the feature point clusters in the image. ; Step 3: Calculate the coordinates of the total feature points in the camera coordinate system. Step 31: Map the total feature points of the image to the hot spots; Geometric relationships through coordinate transformation The images in the data - total feature points and Perform number matching; Step 32: Kalman filtering ensures continuous movement of the total feature points; right Kalman filtering is performed to eliminate noise generated by bursts of image-total feature points, ensuring that the movement of image-total feature points in the image is continuous; Step 33, PNP pose calculation; Solving using the pose calculation matrix determined during the ground camera calibration process The total feature point pose in the camera coordinate system is obtained by combining the PNP algorithm. ; Step 4: Calculate the attitude of the atmospheric sail satellite using the coordinates of the total feature points in the image in the camera coordinate system; Step 41: Determine the position of the total feature points in the camera coordinate system; Calculation in camera coordinate system geometric parameters ; Step 42: Determine the position and attitude of the atmospheric sail satellite relative to the host star; from Extracting the camera coordinate system The relative position and Euler angle parameters of the satellite are used to determine the coordinate system of the primary star. The attitude control of the satellite is achieved by using its relative position and Euler angle parameters.

Images