A monocular vision system based on cooperative targets detection for aircraft pose measurement

Abstract

Abstract. In this paper, a monocular vision measurement system based on cooperative targets detection is proposed, which can capture the three-dimensional information of objects by recognizing the checkerboard target and calculating of the feature points. The aircraft pose measurement is an important problem for aircraft's monitoring and control. Monocular vision system has a good performance in the range of meter. This paper proposes an algorithm based on coplanar rectangular feature to determine the unique solution of distance and angle. A continuous frame detection method is presented to solve the problem of corners' transition caused by symmetry of the targets. Besides, a displacement table test system based on three-dimensional precision and measurement system human-computer interaction software has been built. Experiment result shows that it has a precision of 2mm in the range of 300mm to 1000mm, which can meet the requirement of the position measurement in the aircraft cabin. IntroductionThe aircraft pose measurement system is an important subject of aircraft monitoring and control. Rich information can be obtained by visual measurement methods, which has a good performance in short distance. Targets detection in visual measurement can be divided into cooperative targets and non-cooperative targets. When measuring based on a cooperative target, a target made with a specific size and shape is used, together with a single camera to require pictures of it. The key problem of monocular vision measurement is target detection and pose solution [5,7,8]. Target detection is to detect the feature points for the later calculation. In the aspect of target detection, this paper uses the checkerboard as the target to detect the four corners of the boundary, which transforms the problem into similar P4P problem with four points as the characteristic point [2]. In terms of pose solution, a four-point fixed method based on spatial fixation is adopted.In order to verify the measurement range and precision of the whole measurement system, a stepper motor is used to test the distance and angle of X, Y and Z in three directions. The reliability and measurement of the algorithm accuracy is verified as well at the same time. The human-computer interaction interface based on MFC (Microsoft Foundation Classes) is developed, and the location information can be observed in real time and the intuitive real-time dynamic curve is displayed. Experiment shows that the chessboard target has more features and is easy to extract the feature reliability, and the extracted feature pixel is sub-pixel level, which can meet the requirement of high precision calculation within 300mm ~ 1000mm.

Aircraft Pose Estimation Based on Geometry Structure Features and Line Correspondences

“…For an aircraft with an on-board monocular camera, its pose can be determined by sensing the 3D environments. An on-board monocular vision system based on cooperative target detection was established in [42] for aircraft pose estimation. Coplanar rectangular feature on a known checkerboard target is extracted to determine the pose parameters and a continuous frame detection technique is presented to avoid corners’ confusion.…”