A method for inferring occluded line vectors based on aircraft images
By using the masked straight line vector inference method of aircraft images in optical attitude measurement, the state of the masked vector is restored by using 3D-2D projection and multi-vector combined object image mapping, and the straight line caused by viewing angle changes and image blur in optical attitude measurement is solved. The difficult problem of vector extraction improves the accuracy and reliability of posture processing.
Patent Information
- Application Number
- CN202111241972.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2021-09-17
- Filing Date
- 2021-10-25
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2041-10-25
AI Technical Summary
During the optical attitude measurement process, vector occlusion and image blur due to target motion, viewing angle changes and light rays reduce the accuracy of linear vector extraction and attitude processing accuracy.
The occlusion linear vector inference method based on aircraft images is adopted, and the state of the occlusion vector is restored through 3D-2D projection and multi-vector joint object image mapping method is used to perform inference analysis using the intersection mode and single-station measurement mode to provide posture processing and fault diagnosis. Data support.
It improves the accuracy and reliability of attitude processing, and solves the problem of difficulty in extracting straight vectors caused by viewing angle conversion and image blur, especially when the straight line features are not obvious in aircraft images.
Smart Images

Figure CN114022550B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of conventional weapon testing, and in particular to the field of shooting range optical posture measurement. Background Art
[0002] The interpretation of optical attitude feature quantities is a necessary prerequisite for range attitude measurement and is also the main cause of measurement errors during range attitude measurement. Therefore, it is crucial to improve the accuracy of feature quantities involved in attitude processing. During optical attitude measurement, multiple sets of substations are usually used for intersection optimization processing. During the measurement implementation process, the target movement usually causes changes in perspective, light, defocus, etc., which in turn causes some vectors in the measurement process to be blocked or difficult to accurately extract, which reduces the interpretation efficiency to a certain extent and reduces the accuracy of attitude processing. Summary of the invention
[0003] In order to overcome the shortcomings of the prior art, the present invention provides an obstruction line vector reasoning method based on aircraft images. From the intersection mode and the single-station measurement mode, 3D-2D projection and multi-vector joint object-image mapping are used to reason and analyze the obstruction vector, and the state of the obstruction vector in the sub-station image is restored, providing intuitive data support for the next step of attitude processing optimization and fault diagnosis.
[0004] The technical solution adopted by the present invention to solve the technical problem is: adopting intersection mode or single-station measurement mode for reasoning.
[0005] The rendezvous mode comprises the following steps:
[0006] Use two or more substation images to extract features for attitude intersection processing to obtain the three attitude angles of the aircraft, including the yaw angle ψ, the pitch angle and roll angle γ; using the optical axis pointing of substation i (A i E i ), the obscured vector (abc) in the aircraft coordinate system T Project to the imaging surface and obtain the image (abc) T The corresponding occlusion vector state (y i z i ),
[0007] in,
[0008]
[0009]
[0010]
[0011]
[0012]
[0013] The single-station measurement mode includes the following steps:
[0014] Extract more than three unrelated straight line vectors (y i , z i ), where the real straight line vector is the normalized straight line vector in the aircraft coordinate system (a i , b i , c i ) T ;
[0015] Establish the straight line vector (a i , b i , c i ) T Extract the vector (y from the corresponding image i , z i ),
[0016]
[0017]
[0018] Among them, fy i 、fz i is the 3 components in the 3*3 matrix with the optical axis pointing and the theoretical attitude angle as factors; calculate fy i 、fz i , and then calculate the occlusion vector (abc) T The vector state in the image is (yz),
[0019]
[0020] The beneficial effects of the present invention are: a certain shielding vector is recovered by reasoning from other irrelevant straight line vectors of a multi-station intersection system and a single-station image system, solving the problem of difficulty in extracting straight line vectors caused by key vector shielding and image blur due to perspective conversion, and laying a foundation for improving the accuracy and reliability of attitude processing in the future. The present invention can also be used to solve the problem that the straight line vector is not easy to extract as a whole due to unclear features of certain straight lines (such as a round nose) in an aircraft image. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION
[0022] The present invention is further described below in conjunction with the accompanying drawings and embodiments. The present invention includes but is not limited to the following embodiments.
[0023] The present invention is based on the inference and restoration of the shielded straight line vector of the aircraft image, starting from the single-station image or multi-station image mode, to solve the problem of the difficulty in extracting the straight line vector caused by the vector shielding of the perspective conversion and the image blur caused by light, movement and other reasons. Among them, the vector shielding inference and interpretation method based on the attitude intersection system includes the following steps:
[0024] Step 1: Use two or more substation images to extract features for attitude intersection processing to obtain the aircraft's three attitude angles (yaw angle ψ, pitch angle , roll angle γ);
[0025] Step 2: Based on the three attitude angles obtained in the first step, use the substation optical axis pointing (AE) to convert the obscured vector (abc) in the aircraft coordinate system T Project to the imaging surface and obtain the image (abc) T The corresponding occlusion vector state (yz).
[0026]
[0027] in:
[0028]
[0029]
[0030]
[0031]
[0032]
[0033] All the coordinate systems involved are based on the coordinate systems involved in the conventional posture processing algorithm. The target coordinate system takes the target center axis as the X T Axis, head is positive; the reference coordinate system takes north as X axis; the camera coordinate system takes the optical axis as X axis C Axis, parallel to the image plane and upward is Y C Axis; the image plane coordinate system y and Z axes are parallel to the camera coordinate system Y C Axis, Z C Axis. The initial zero pose means that the target coordinate system is parallel to the reference coordinate system.
[0034] The vector occlusion inference and interpretation method based on a single image posture processing system includes the following steps:
[0035] Step 1: Extract three or more unrelated straight line vectors (y i ,z i), where the real straight line vector is the normalized straight line vector in the aircraft coordinate system (a i ,b i ,c i ) T ;
[0036] Step 2: Establish the straight line vector (a i ,b i ,c i ) T (3D) and the corresponding image to extract the vector (y i ,z i )(2D) correspondence;
[0037]
[0038]
[0039] Among them, fy i ,fz i It is the 3 components in a 3*3 matrix with the optical axis pointing and the theoretical attitude angle as factors.
[0040] Step 3: Calculate fy from step 2 i ,fz i ;
[0041] Step 4: The shielding vector (abc) can be calculated from the calculation results of step 3 T The vector state in the image is (yz).
[0042]
[0043] Take the case where the central axis vector of substation 1 is blocked by the viewing angle in the attitude intersection mode as an example for explanation.
[0044] At a certain moment, due to the change of aircraft attitude, the key part of the axis vector of substation 1 is blocked. The vector shielding inference and interpretation method based on the attitude intersection system includes the following steps:
[0045] Step 1: Use two or more sub-station image intersection processing to obtain the three attitude angles (yaw angle ψ, pitch angle , roll angle γ);
[0046] It is known that the tail image extraction vector of substation 1 is (-90,459), and the optical axis is pointing to (30.5° 24.67°); the central axis image extraction vector of substation 2 is (-26,162), the tail image extraction vector is (279,252), and the optical axis is pointing to (299.327°29.728°). After intersection processing, the attitude angle of attitude 3 is (5.114° 3.7556° 0.1099°);
[0047] Step 2: Based on the three attitude angles obtained in the first step, use the optical axis pointing (AE) of substation 1 to convert the obscured vector (1 0 0) in the aircraft coordinate system to T Project to the imaging plane and obtain the image (1 0 0) T The corresponding occlusion vector state is (-0.317 -0.428).
[0048] Taking the aircraft center axis vector being shielded under the single image attitude processing system as an example, the attitude interpretation method based on vector shielding inference under the single image system includes the following steps:
[0049] Step 1: Extract three or more unrelated straight line vectors (y i ,z i ), where the real straight line vector is the normalized straight line vector in the aircraft coordinate system (a i ,b i ,c i ) T ; Based on the image, three straight line vectors (-198.77 -115.07), (-158.21 -7.88), (-51.62 103.79) are extracted. The normalized coordinates of these three straight line vectors in the aircraft coordinate system are (0.82 0.03 0.58) T 、(0 0 1) T 、(0 0.99 0.02) T ;
[0050] Step 2: Establish the straight line vector (a i ,b i ,c i ) T (3D) and the corresponding image to extract the vector (y i ,z i )(2D) correspondence;
[0051]
[0052]
[0053] Among them, fy i 、fz i It is the 3 components in a 3*3 matrix with the optical axis pointing and the theoretical attitude angle as factors.
[0054] Step 3: Calculate fy from the corresponding relationship in step 2 i 、fz i ;
[0055] Note: The number of pixels extracted from the actual image should also be normalized accordingly.
[0056] After calculation, fy1=-44.3637, fy2=-39.9293, fy3=-71.9136; fz1=-56.4883, fz2=83.2774, fz3=-3.5818.
[0057] Step 4: The shielding vector (abc) can be calculated from the calculation results of step 3 T The vector state in the image is (yz).
[0058]
[0059] After calculation, the obscured median axis vector is (-44.3637 -56.4883).
Claims
1. A method for inferring obscured line vectors based on aircraft images, characterized in that: Use intersection mode or single station measurement mode for inference; The rendezvous mode comprises the following steps: Use more than two substation images to extract features for attitude intersection processing to obtain the three attitude angles of the aircraft, including the yaw angle ψ, the pitch angle and roll angle γ; using the optical axis pointing of substation i (A i E i ), the obscured vector (abc) in the aircraft coordinate system T Project to the imaging surface and obtain the image (abc) T The corresponding occlusion vector state (y i z i ), in, The single-station measurement mode includes the following steps: Extract more than three unrelated straight line vectors (y i ,z i ), where the real straight line vector is the normalized straight line vector in the aircraft coordinate system (a i ,b i ,c i ) T ; Establish the straight line vector (a i ,b i ,c i ) T Extract the vector (y from the corresponding image i ,z i ), Among them, fy i 、fz i is the 3 components in the 3*3 matrix with the optical axis pointing and the theoretical attitude angle as factors; calculate fy i 、fz i , and then calculate the occlusion vector (abc) T The vector state in the image is (yz),
