无人机影像正射纠正与拼接技术的研究与应用

发布时间：2018-01-29 23:27

本文关键词： 无人机影像正射纠正有理函数模型 Dijkstra算法影像拼接　出处：《吉林大学》2017年硕士论文　论文类型：学位论文

【摘要】：近年来,由于人们对快速测绘与高精度测绘的研究与开发越来越关注,相比较于传统航天遥感平台所展现出来的容易受外界环境的强烈影响,作业耗时较长,机动性差等不可避免的因素影响,无人机凭借其独有的高机动灵活性、高效率及低消耗等优点,成功成为传统遥感平台和常规航空摄影测量技术的合理代替,同时也受到相关行业的密切关注和重点研发;尤其在小范围的高精度定位数据和高分影像快速获取、灾害应急响应及危险区域的空中监测等方面具有明显的优势,开发针对无人机影像的快速处理算法,面对地理国情普查、国家应急救灾及数字城市建设等方面都具有极为关键的优势。在低空无人机遥感影像处理系统中,影像的正射纠正及拼接是无人机遥感影像处理的关键。选择适当的高效的方法将无人机航摄影像进行正射影像校正及拼接对于实现数字正射影像的自动化生产具有重要意义。本文的研究重点是无人机影像正射纠正与拼接技术,主要从以下几个方面进行研究:(1)详细论述了影像正射纠正的基本原理的处理流程,分析总结了上述原理与方法的优缺点及不同条件下的兼容性;(2)分析无人机正射影像拼接方法,主要包括Dijkstra算法、动态规划法拼接线、Twin snake算法镶嵌线检测和蚁群算法镶嵌线等影像拼接算法的原理、作业流程及优缺点;(3)结合长春净月和德惠同太乡苇子村工程实例,进行无人机影像正射纠正及拼接的实验验证及分析。净月实验区主要由丘陵地区及居民区构成,无人机平台携载索尼a7r相机摄取两条航带、59幅航摄像片航摄影像,摄区范围:43°42′01.6488′′N-43°42′08.23824′′N,125°19′14.8314′′E-125°20′17.7738′′E,飞行方向大致为东西走向;德惠市同太乡苇子村实验区地势平坦,以平原地势为主,包含部分居民区,无人机平台携载索尼ILCE-QX1相机摄取20条航带、884幅航摄像片,摄区范围:经纬度为44°23′09.276′′N-44°25′58.8′′N,125°20′46.068′′E-125°25′20.928′′E,飞行方向大致为南北走向。对这两个实验区的无人机影像结合POS数据进行空三加密,逐一正射纠正单幅影像,最后拼接生成正射影像图。在德惠市同太乡苇子村实验区选取100个检校点对所处理的正射影像进行精度检验,得到平面点位中误差为12.2cm,满足精度要求;另外,选取局部正射影像与该区域已有的DLG进行叠加分析验证正射影像的质量,由此发现,居民区及平原地区的正射影像质量最高,丘陵地区次之。
[Abstract]:In recent years, because people pay more and more attention to the research and development of rapid mapping and high-precision mapping, compared with the traditional space remote sensing platform, it is easy to be strongly affected by the external environment, and the operation time is longer. Due to its unique advantages of high mobility, high efficiency and low consumption, UAV has become a reasonable replacement of traditional remote sensing platform and conventional aerial photogrammetry technology. At the same time, it also receives the close attention and the key research and development of the related industries; Especially in the small range of high-precision positioning data and high-score image rapid acquisition disaster emergency response and aerial monitoring in dangerous areas and other aspects have obvious advantages to develop a rapid processing algorithm for UAV images. Facing the geographical situation survey, the national emergency relief and the digital city construction and so on, all have the extremely vital superiority, in the low altitude unmanned aerial vehicle remote sensing image processing system. The key of UAV remote sensing image processing is to correct and concatenate the orthophoto image of UAV. Selecting proper and efficient method to correct and assemble aerial aerial image of UAV image can realize the automatic production of digital orthophoto image. This paper focuses on orthophoto correction and splicing technology of UAV images. The processing flow of the basic principle of orthophoto correction is discussed in detail, and the advantages and disadvantages of the above principles and methods and the compatibility under different conditions are analyzed and summarized. 2) analyze the orthophoto mosaic method of UAV, including Dijkstra algorithm and dynamic programming method. The principle, operation flow, advantages and disadvantages of the image mosaic algorithms, such as Twin snake algorithm mosaic line detection and ant colony algorithm mosaic line algorithm; Combined with the engineering examples of Jingyue in Changchun and Weizi Village in Tongtai Township of Dehui, the experimental verification and analysis of orthophoto correction and splicing of UAV images are carried out. The experimental area of Jingyue is mainly composed of hilly areas and residential areas. The UAV platform carries a Sony a7r camera to capture 59 aerial photographs of two airstrips. The range of shooting area is: 43 掳42 / 01.6488 / N-43 掳42 / 08.23824 / N. 125 掳19m 14.8314m E-125 掳20 掳17.7738C, the direction of flight is roughly east-west direction; The experimental area of Weizi Village in Tongtai Township of Dehui City is flat, mainly plain terrain, including some residential areas. The UAV platform carries Sony ILCE-QX1 camera to take 884 aerial photographs of 20 flight belts. Area: longitude and latitude: 44 掳23 / 09.276 / N, N-44 掳25N / 58.8N. 125 掳20 掳46.068 掳E-125 掳25 掳20.928 E. The flight direction is approximately north-south direction. The UAV images of the two experimental areas are encrypted with POS data, and one by one orthography is used to correct a single image. At last, the orthophoto map was constructed. 100 calibration points were selected to check the accuracy of the processed orthophoto images in Weizi Village, Tongtai Township, Dehui City, and the median error of plane points was 12.2cm. Meet the precision requirements; In addition, the quality of orthophoto images is verified by superposition analysis between local orthophoto images and existing DLG in this area. It is found that the quality of orthophoto images is the highest in residential areas and plain areas, followed by hilly areas.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P23

【参考文献】