基于偏振光的光流导航定位方法-模式识别与智能系统专业论文.docx

华 华 中 科 技 大 学 硕 士 学 位 论 文 II II Abstract Among the existing methods of navigation, GPS navigation sometimes has trouble receiving signal from the satellites, Inertial Navigation and Celestial Navigation are too expensive and take up too much space, while Doppler Navigation relies on radar signal therefore can be easily discovered. This article proposed a novel navigation algorithm that combines polarized navigation and optical flow navigation and can be much cheaper, free from interference and takes less space, with good navigation and positioning accuracy. First, from the heading angle of the aircraft detected by the polarization sensor, the constraint relation between the ego-motion parameters could be obtained. Specifically, the heading angle tells us the ratio between the two horizontal components of the displacement of the aircraft, variations of the heading angle in two adjacent time gaps tell us the rotation angles of the aircraft around vertical direction. Subsequently, as the aircraft flies at high altitude, the ground scene can be approximated as a plane, so planar flow model can be used to calculate the six camera motion parameters. In accordance with the heading angle and the rotation angle about the vertical axis, this calculation process can be improved. According to the minimum mean square error method, estimate the ego-motion of the aircraft by optical flow and then the position of the aircraft can be tracked. The principle of the algorithm proposed in this paper is above. The 6 applicable conditions for using the proposed method are also analyzed in this paper. Finally, this paper implemented the simulation of pure optical flow navigation, and optical flow navigation based on skylight polarization. Numerical simulation proved that the integrated navigation system can get more accurate position than pure optical flow navigation. Since the method proposed in this paper requires smaller space for physical devices and lower cost, it makes the most