基于不规则三角网的遥感影像自动配准系统设计

摘要:

在遥感影像配准过程中,正射投影图像会在矩形框顶点展开排序。但由于地物场景形式的多样性与影像采集传感器平移、旋转的多变性,影像特征区域之间呈现重叠覆盖程度高、可区分性差的特点,致使相同地物在不同影像中的位置存在偏移现象,这种几何畸变导致遥感影像平均正确配准点率较低。为此,利用不规则三角网技术,设计遥感影像自动配准系统。改装遥感影像采集器、角度传感器、数据处理器,调整系统电源电路的连接方式,实现硬件系统的优化设计,以保障数据库数据的安全性和可靠性。利用遥感影像采集器光学原理,生成遥感影像,通过大气校正、影像增强等步骤,保证生成遥感影像的生成质量。采用Harris算子,提取遥感影像特征点,以特征点为顶点,利用Delaunay准则构建不规则三角网,构建不规则三角网,获取遥感影像的分割线,实现遥感影像的自动分割,以更精确地纠正影像间的偏移问题,解决几何畸变对影像配准的正确配准率的影响。通过粗配准、精配准和误配准剔除三个步骤,实现系统的遥感影像自动配准功能。通过系统测试实验得出结论:与传统配准系统相比,优化设计系统对低、高分辨率遥感影像的正确配准点率分别提高3.5%和1.9%,同时系统运行加速比明显提升。

Abstract:

In the process of remote sensing image registration, the orthographic projection image will be expanded and sorted at the vertex of the rectangular box. However, due to the diversity of scene forms of ground objects and the variability of translation and rotation of image acquisition sensors, image feature regions present high overlapping coverage and poor differentiation, resulting in the position deviation of the same ground objects in different images, and this geometric distortion leads to a low average correct on-time allocation rate of remote sensing images. Therefore, an automatic remote sensing image registration system is designed by using triangulation irregular network technology. The remote sensing image collector, Angle sensor and data processor were modified, and the connection mode of the system power circuit was adjusted to realize the optimal design of the hardware system, so as to ensure the security and reliability of the database data. Using the optical principle of remote sensing image collector, the remote sensing image is generated, and the quality of the generated remote sensing image is guaranteed by the steps of atmospheric correction and image enhancement. The Harris operator is used to extract the feature points of remote sensing images, and the feature points are taken as the vertices. The triangulation irregularity network is constructed by the Delaunay criterion, and the division lines of remote sensing images are obtained to realize automatic segmentation of remote sensing images, so as to correct the deviation problem between images more accurately and solve the influence of geometric distortion on the correct registration rate of images. Through three steps of rough registration, fine registration and misregistration elimination, the automatic registration function of remote sensing image is realized. The experimental results show that compared with the traditional registration system, the correct registration rate of low and high resolution remote sensing images is improved by 3.5% and 1.9% respectively, and the acceleration ratio of the system is obviously improved.