A control point gross error elimination method for unmanned aerial vehicle visual positioning

Abstract

The application discloses a control point gross error elimination method for unmanned aerial vehicle visual positioning and belongs to the technical field of visual autonomous positioning. The method eliminates control point wild values by adopting point position distance and direction constraint conditions among control points. Firstly, the size of a current frame aerial photograph is calculated according to prior knowledge, and the distance threshold of the control points is determined. Then, the point position distance among the control points is calculated according to the longitude and latitude coordinates of the control points, and compared with the point position distance threshold, and the control points exceeding the threshold are eliminated. Then, the true value data of the point position direction and topological relationship among the control points are calculated, and the point position direction threshold among the control points is set. Next, the point position direction and topological relationship among the control points are calculated, and compared with the true value data of the point position direction, and the control points exceeding the point position direction threshold are eliminated. Finally, the current frame unmanned aerial vehicle position is solved according to the camera imaging geometric relationship. Compared with the traditional unmanned aerial vehicle visual positioning method, the application can effectively improve the reliability of the unmanned aerial vehicle visual positioning.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of visual autonomous positioning, and particularly relates to a UAV autonomous positioning method based on a control point gross error elimination method. BACKGROUND

[0002] The UAV has the characteristics of good economy and strong maneuverability, and is widely used in military operations and civilian tasks. The premise for the UAV to perform tasks is to have stable and reliable positioning information. When the flight control system cannot receive or receive unreliable position information, the UAV cannot work normally.

[0003] At present, the UAV positioning system widely uses the Global Navigation Satellite System (GNSS) and the Inertial Navigation System (INS). In urban environment, the GNSS signal is weak due to the shielding of buildings, and the UAV cannot receive continuous and reliable positioning data. Due to the consideration of load and economy, small UAVs are difficult to carry high-precision INS systems. In addition, the INS positioning error will accumulate with the increase of working time, and it is difficult to realize reliable UAV autonomous positioning.

[0004] The UAV visual autonomous positioning system has been a research hotspot for UAV autonomous positioning in recent years due to its good economy and strong autonomy. However, there are often outliers in the control points obtained by visual positioning, which have large deviations from the true geographical position, resulting in large instantaneous deviations of the visual positioning system, which also cannot meet the reliable UAV positioning requirements. SUMMARY

[0005] The purpose of the present application is to provide a control point gross error elimination method for UAV visual positioning, to solve the problem of large positioning error caused by large deviation of the control points obtained in the existing UAV visual positioning technology.

[0006] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0007] A control point gross error elimination method for UAV visual positioning, comprising the following steps:

[0008] Step 1: Calculate the aerial image map size according to the UAV height above ground, focal length, pixel number and pixel size data, and determine the point distance threshold between control points;

[0009] Step 2: Calculate the point distance between control points according to the control point latitude and longitude coordinates, and then compare it with the point distance threshold to eliminate the control points exceeding the point distance threshold;

[0010] Step 3, the inertial navigation measurement data and control point image coordinate data of the unmanned aerial vehicle flight control system are used to calculate the true value data of the point direction and topological relationship between control points, and the point direction threshold between control points is set;

[0011] Step 4, the point direction and topological relationship between control points are calculated according to the latitude and longitude coordinates of the control points, and are compared with the true value data of the point direction and topological relationship respectively, and the control points exceeding the point direction threshold and the control points not satisfying the geographical topological relationship are removed;

[0012] Step 5, the current frame unmanned aerial vehicle position is solved according to the camera imaging geometric relationship, so that the current time unmanned aerial vehicle autonomous positioning is realized.

[0013] Further, the calculation method of the point distance threshold between control points in step 1 is: using the unmanned aerial vehicle height from the ground H, the imaging camera focal length f, the image element size P s and the pixel number N, the point distance threshold between control points is calculated Wherein, rows is the number of image rows, cols is the number of image columns, and sigma is the compensation coefficient.

[0014] Further, step 4 is specifically:

[0015] The current frame aerial image is corrected using the camera interior orientation parameters and the unmanned aerial vehicle inertial navigation measurement attitude data, the corrected image control points are distributed according to the geographical direction, the point direction yaw between control points is calculated using the image coordinates of the control points, the control points not satisfying |yaw-θ|<T2 are removed, and the control points not satisfying the geographical topological relationship are removed according to the geographical distribution relationship of the control points; wherein, theta is the true value data of the point direction, and T2 is the point direction threshold.

[0016] Compared with the background art, the present application has the following advantages:

[0017] 1. The control point gross error elimination method for unmanned aerial vehicle visual positioning provided by the present application uses the map size estimated according to the current frame image imaging resolution as the threshold of the control point point distance, removes the control points with large deviation from the current scene, and ensures that the control points participating in the unmanned aerial vehicle position solution do not have wild values.

[0018] 2. The control point gross error elimination method for unmanned aerial vehicle visual positioning provided by the present application uses the control point image coordinates to calculate the point direction and topological information between control points as the true value data, which can completely eliminate the situation that the geographical coordinates do not match the actual topological relationship in the control points, and realizes a high-reliability unmanned aerial vehicle visual positioning scheme. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 is a flow chart of a UAV visual positioning method based on a control point gross error elimination method provided by an embodiment of the application.

[0020] Figure 2 is a flow chart of a control point distance gross error elimination method.

[0021] Figure 3 is a flow chart of a control point direction and topological relationship gross error elimination method. DETAILED DESCRIPTION

[0022] The concept, technical solution advantages and technical effects of the application will be described clearly and completely in the following embodiments, so as to fully understand the purpose, features and effects of the application. It should be noted that the specific embodiments described herein are only used to explain the application and do not limit the application.

[0023] In this embodiment, as shown in a control point gross error elimination method for UAV visual positioning, the following steps are included: Figure 1

[0024] S1: Calculate the current frame aerial image map size according to the UAV height above ground, focal length, pixel number and pixel size data, and determine the point distance threshold between control points; the specific process is as follows:

[0025] (1) Calculate the current frame aerial image ground resolution β according to the following formula.

[0026]

[0027] Where H is the UAV height above ground, f is the aerial camera focal length, P s is the image pixel size, N is the maximum pixel number, rows is the image row number, and cols is the image column number.

[0028] (2) Given the current frame aerial image ground resolution and the aerial camera pixel size, then calculate the current frame aerial image map T = βP s .

[0029] (3) Given the current frame aerial image map, set the compensation coefficient of the map distance constraint, and get the point distance threshold T1 between control points. In this embodiment, it is set to 50 meters.

[0030] S2: Calculate the point distance D between control points according to the control point latitude and longitude coordinates (lat, lon), and then compare it with the point distance threshold, and eliminate the control points exceeding the threshold; as shown in Figure 2

[0031] ​​S3: using the camera interior orientation parameters and the UAV inertial navigation measurement attitude data to correct the current frame aerial image, using the inertial navigation measurement data of the UAV flight control system and the control point image coordinate data to calculate the true value data of the point direction and the topological relationship between the control points, and setting the point direction threshold T2 between the control points;

[0032] Further, (1) obtaining the inertial navigation attitude data (including the pitch angle , the yaw angle ω and the roll angle κ) and the interior orientation elements (the focal length f x , f y and the image principal point coordinates P x , P y ) corresponding to the current time of the aerial camera, then converting the three attitude angles from the degree system to the radian system; then, calculating the intermediate value of the attitude transformation matrix,

[0033] calculating the image attitude transformation matrix.

[0034] If the rotation order is , the attitude transformation matrix is calculated as shown in the following formula.

[0035]

[0036] If the rotation order is , the attitude transformation matrix is calculated as shown in the following formula.

[0037]

[0038] If the rotation order is , the attitude transformation matrix is calculated as shown in the following formula.

[0039]

[0040] If the rotation order is , the attitude transformation matrix is calculated as shown in the following formula.

[0041]

[0042] If the rotation order is , the attitude transformation matrix is calculated as shown in the following formula.

[0043]

[0044] If the rotation order is , the attitude transformation matrix is calculated as shown in the following formula.

[0045]

[0046] After obtaining the image pose matrix HR, the current frame aerial image is corrected to the geographic normal direction.

[0047] (2) Assuming that the image coordinates of two control points in the image are (x1, y1) and (x2, y2) respectively, the true value data θ of the point direction between the control points is calculated according to the following formula.

[0048]

[0049] The point direction threshold is set, and the point direction threshold in the embodiment is set to 5°.

[0050] S4: The point direction and topological relationship between the control points are calculated according to the longitude and latitude coordinates of the control points, and the true value data corresponding to the point direction and topological relationship are compared respectively, and the control points exceeding the point direction threshold and the control points not satisfying the geographic topological relationship are removed.

[0051] The current frame aerial image is corrected using the camera interior orientation parameters and the inertial navigation measurement pose data of the unmanned aerial vehicle, and the correction process is consistent with step S3. The corrected image control points are distributed according to the geographic direction, and the point direction between the control points is calculated using the image coordinates of the control points. If the geographic coordinates of two control points are (lon1, lat1) and (lon2, lat2) respectively, the point direction yaw between the control points is calculated according to the following formula.

[0052] yaw=atan2(yy,xx) / dpi

[0053] Wherein, dpi=0.017453292519943295, yy=sin((lon1-lon2)*dpi)*cos(lat1*dpi), xx=cos(lat2*dpi)*sin(lat1*dpi)-sin(lat2*dpi)*cos(lat1*dpi)*cos((lon1-lon2)*dpi). Then, the point direction yaw calculated is subtracted from the true value data of the point direction θ, and the control points exceeding the point direction threshold are removed.

[0054] S5: The current frame unmanned aerial vehicle position is solved according to the camera imaging geometric relationship, so as to realize the unmanned aerial vehicle visual positioning at the current time.

[0055] Although the above describes the specific embodiments of the present application in order to facilitate the understanding of the present application by those skilled in the art, it should be clear that the present application is not limited to the scope of the specific embodiments, and for those skilled in the art, it is obvious that various changes are within the spirit and scope of the present application defined and limited by the appended claims, and all the inventions utilizing the concept of the present application are within the scope of protection.

Claims

1. A method for eliminating gross errors in control points for visual positioning of unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: Step 1: Calculate the size of the current frame aerial image based on the drone's altitude, focal length, number of pixels, and pixel size data, and determine the threshold for the distance between control points; Step 2: Calculate the distance between control points based on their latitude and longitude coordinates, then compare it with the distance threshold and remove control points that exceed the distance threshold. Step 3: Using the inertial navigation measurement data and control point image coordinate data of the UAV flight control system, calculate the true data of the positional orientation and topological relationship between control points, and set the positional orientation threshold between control points; Step 4: Calculate the positional orientation and topological relationship between control points based on their latitude and longitude coordinates, and compare them with the true data of positional orientation and topological relationship respectively. Remove control points that exceed the positional orientation threshold and control points that do not meet the geographical topological relationship. The positional direction yaw between control points is calculated according to the following formula: yaw = atan2(yy,xx) / dpi Where dpi = 0.017453292519943295, yy = sin((lon1-lon2)*dpi)*cos(lat1*dpi), xx = cos(lat2*dpi)*sin(lat1*dpi)-sin(lat2*dpi)*cos(lat1*dpi)*cos((lon1-lon2)*dpi), and (lon1,lat1) and (lon2,lat2) are the geographic coordinates of two control points; Step 5: Calculate the drone's position in the current frame based on the camera's imaging geometry, thereby achieving autonomous drone localization at the current moment.

2. The method for eliminating gross errors in control points for UAV visual positioning according to claim 1, characterized in that, The method for calculating the point distance threshold between control points in step 1 is as follows: using the drone's altitude H, the aerial camera's focal length f, and the image pixel size P. s Given the number of pixels N, calculate the point distance threshold between control points. in, rows is the number of rows in the image, cols is the number of columns in the image, and σ is the compensation coefficient.

3. The method for eliminating gross errors in control points for UAV visual positioning according to claim 1, characterized in that, Step 4 specifically involves: The aerial image of the current frame is corrected using the camera's in-camera orientation parameters and the attitude measurement data from the UAV's inertial navigation system. The corrected image control points are distributed according to geographical direction, and the positional direction yaw between control points is calculated using the image coordinates of the control points. Control points that do not satisfy |yaw-θ|<T2 are removed, and control points that do not satisfy the geographical topological relationship are also removed based on the geographical distribution relationship of the control points. Here, θ is the true value data of the positional direction, and T2 is the positional direction threshold.

Images