Research On Aircraft Pose Estimation And Ground Target Motion Parameter Measurement

Aircraft pose estimation and ground target motion parameter measurement is one of the key technologies in a variety of aerospace engineering applications.The attitude estimation technique of the aircraft can provide basic information about the flight state of the aircraft,which can be used for aircraft system design,autonomous flight and accident analysis.Ground target positioning and velocity measurement technology has important military application value and can provide valuable information for battle field situational awareness,target reconnaissance and strike,and the improvement of the combat effectiveness of weapons.This paper focuses on aircraft pose estimation and ground target positioning and velocity measurement.The main technical problems are: aircraft pose estimation based on external vision sensor,ground target motion estimation based on spaceborne optical sensor and target positioning based on airborne SAR sensor.Starting from the above technical problems,this paper designs and proposes the aircraft monocular pose estimation algorithm based on structural features and line correspondence,the aircraft binocular pose estimation algorithm based on structural features and surface intersection,Fourier transform based motion detection algorithm using remote sensing images and the airborne SAR ground target positioning algorithm based on ground elevation information.These algorithms are evaluated through experimental research and practical engineering applications.The main results achieved in this paper are as follows:A monocular aircraft pose estimation algorithm based on structural features and line correspondence is proposed.In order to directly obtain the absolute pose of the aircraft in the world coordinate system from a single image,the pose estimation method based on the external monocular camera often needs priori information such as the three-dimensional model or feature datasets.However,the creation and storage of 3D models or feature databases are time-ans storage-consuming,and the efficiency of feature or template matching is also low.These factors will reduce the flexibility and efficiency of pose estimation algorithms.The aircraft is a special kind of target,and its inherent geometric features can be used as priori information for aircraft structure extraction and recognition in 2D images.In this paper,the aircraft monocular pose estimation algorithm uses the line features to describe the aircraft structure in the image.Based on the analysis of the structural characteristics of the aircraft,the geometrical constraints between the line features are used to extract the fuselage reference line and the wing leading and trailing edge line.After extracting and identifying the corresponding line features of the aircraft in the image,the algorithm establishes the correspondence between the two-dimensional line features in the image and the 3D model line features in the body frame,and a Pn L algorithm is selected to solve the aircraft pose.In this paper,the monocular visual pose estimation algorithm does not need a complete three-dimensional model,and only needs part of the geometric description of the aircraft wing in the body coordinate system to accurately estimate the pose parameters of the aircraft,which is more efficient than the method based on model or feature matching.A binocular aircraft pose estimation algorithm based on structural features and planes intersection is proposed.The binocular visual pose estimation algorithm in this paper improves the structure extraction algorithm in the above-mentioned monocular vision method.According to the minimum configuration required for binocular visual pose estimation,only the leading edge line features of the aircraft wing are extracted to perform the pose measurement.Since the wing leading edge line features are more significant and stable than the trailing edge line feature during aircraft attitude changes,the improved structure extraction algorithm can accommodate more types of aircraft and a wider range of attitude changes.In addition,the algorithm directly establishes the correspondence between the features of wide-baseline images through the extraction and recognition of the aircraft structure,avoiding feature matching of wide-baseline images,so that the binocular vision measurement system of this paper can configure a wider baseline and obtain a larger measuring range and higher accuracy.An efficient and universal motion estimation algorithm using remote sensing images is proposed.Motion estimation using remote sensing images is generally divided into two steps: firstly,the target is detected and positioned,and then the image matching method is used to determine the displacement between the remote sensing images captured at different time.Target detection and positioning based on supervised learning or manual intervention will reduce the versatility and timeliness of the algorithm;the target detection method based on background difference is difficult to adapt to remote sensing images with large illumination differences,and its background modeling method will also affect the accuracy.To solve above problems,the proposed algorithm combines the target positioning and displacement estimation,and introduces the gradient correlation and iterative multi-grid image deformation framework to improve the accuracy and robustness of motion detection algorithm.Firstly,the algorithm calculates the displacement field between remote sensing images based on Fourier Merlin transform,and then uses density-based clustering to detect and locate the moving targets in an unsupervised manner,and at the same time determine the displacement of the target.A universal and efficient ground target positioning and velocity measurement framework is established.4.Improve the accuracy of airborne SAR target positioning by introducing ground elevation informationIn this paper,the terrain elevation of the airborne SAR imaging area is fitted by the ground elevation information,which is used to correct the positioning error caused by the terrain fluctuation.Firstly,the algorithm uses the ground elevation information to establish an elevation constraint equation,and by considering the average elevation in the earth ellipsoid equation as a variable,the relationship between the earth ellipsoid equation and the elevation constraint equation is established;then the method combines the Range equation,the Doppler equation,the earth ellipsoid equation and the local elevation constraint equation to form a nonlinear equation to estimate the position of the ground target.In this paper,the effective solution of the nonlinear equations is introduced in detail,and the influence of initial value selection is discussed.In this paper,aircraft pose estimation and ground target motion parameter measurement are studied in detail,and the corresponding algorithms are designed.The effectiveness and practicability of the proposed algorithms are verified by laboratory experiments and project experiments.