无人机影像数据快速配准和自动拼接系统的设计与实现
发布时间：2019-04-18 19:48
【摘要】：由于无人机航测遥感系统具有灵活、成本低、大比例尺高精度的特点，在小区域和飞行困难地区快速获取高分辨率影像方面有明显优势。因此，无人机航测遥感技术已经成为提高测绘成果现势性的有力手段，是增强测绘应急保障能力的捷径。无人机影像由于受飞行高度、相机视角的影响，单张无人机影像所覆盖的区域面积不大，在特定任务中需要对多张影像进行拼接，有效覆盖所有工作区。影像匹配从提出到现在，经过了无数次的改进和发展，无论是匹配点精度还是匹配速度都有了质和量的飞跃，但是由于无人机影像具有像幅小、重叠度变化大、旋偏角大、影像畸变大、噪声和遮挡严重等特性，需要寻找一种对各种畸变、噪声都具有良好鲁棒性的一种算法，而基于特征匹配的影像拼接算法能很好的满足，因而被广泛的应用。本文主要以基于尺度不变特征（SFIT）的影像匹配算法为主要内容，学习无人机影像快速自动拼接的关键技术，并利用C++编程实现，，论文的研究工作主要包括以下几个部分： 1.总结了无人机影像拼接技术的意义和国内外研究现状，确定了本文无人机影像拼接的技术流程。 2.利用C++语言编程，VS2010平台编译实现了无地理坐标的无人机遥感影像快速自动拼接。采用SIFT算法进行影像特征点提取、同名点匹配，然后利用RANSAC剔除误匹配点对，完成精匹配。最后利用同名点对解算影像间的几何变换模型，完成一条航带多张影像拼接，输出成果。 3.采用直接加权平均法对拼接影像进行融合，处理色差、光照差异、拼接缝等问题。 研究结果表明，采用SIFT算法能够有效提取大量的特征点用于影像匹配，对缺乏地面控制点的无人机影像拼接效果良好。但是由于特征点过多会影响计算速度，需要寻求有效方法进行过滤，更好的满足无人机遥感的时效性要求。
[Abstract]:Because UAV aerial survey remote sensing system has the characteristics of flexibility, low cost and high precision on large scale, it has obvious advantages in obtaining high-resolution images quickly in small area and difficult area of flight. Therefore, UAV aerial remote sensing technology has become a powerful means to improve the present situation of surveying and mapping achievements, and it is a shortcut to enhance the ability of emergency support of surveying and mapping. Due to the influence of flying altitude and camera angle, the area covered by single UAV image is not large, so it is necessary to splice multiple images to cover all the working areas effectively in a given task. Since it was put forward, image matching has been improved and developed countless times. Both the precision of matching point and the speed of matching have made a leap in quality and quantity. However, due to the small image size, large overlap and large rotation angle, the UAV image has a small image amplitude, a large degree of overlap, and a large rotation angle. Because of its large distortion, serious noise and occlusion, it is necessary to find an algorithm that has good robustness to all kinds of distortion and noise. However, the image mosaic algorithm based on feature matching can be satisfied very well, so it has been widely used. In this paper, the image matching algorithm based on scale invariant feature (SFIT) is taken as the main content, and the key technology of fast automatic stitching of UAV image is studied, and realized by C programming. The research work of this paper mainly includes the following parts: 1. This paper summarizes the significance of UAV image mosaic technology and the research status at home and abroad, and determines the technical flow of UAV image splicing in this paper. 2. By programming in C language and compiling on VS2010 platform, the remote sensing image of unmanned aerial vehicle (UAV) without geographical coordinates can be automatically stitched quickly and automatically. The SIFT algorithm is used to extract the feature points and match the same name points. Then the mismatched point pairs are eliminated by RANSAC to complete the fine matching. Finally, using the geometric transformation model of the same-named point pair to solve the image, we complete the multi-image splicing of one airstrip and output the result. 3. The direct weighted average method is used to fuse the splicing image, deal with the color difference, illumination difference, splicing seam and so on. The results show that the SIFT algorithm can effectively extract a large number of feature points for image matching and has a good effect on UAV image mosaic without ground control points. However, because too many feature points will affect the computing speed, it is necessary to find an effective method to filter and better meet the requirements of the time-effectiveness of UAV remote sensing.
【学位授予单位】：中国地质大学（北京）
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P231;P237
本文编号：2460289
[Abstract]:Because UAV aerial survey remote sensing system has the characteristics of flexibility, low cost and high precision on large scale, it has obvious advantages in obtaining high-resolution images quickly in small area and difficult area of flight. Therefore, UAV aerial remote sensing technology has become a powerful means to improve the present situation of surveying and mapping achievements, and it is a shortcut to enhance the ability of emergency support of surveying and mapping. Due to the influence of flying altitude and camera angle, the area covered by single UAV image is not large, so it is necessary to splice multiple images to cover all the working areas effectively in a given task. Since it was put forward, image matching has been improved and developed countless times. Both the precision of matching point and the speed of matching have made a leap in quality and quantity. However, due to the small image size, large overlap and large rotation angle, the UAV image has a small image amplitude, a large degree of overlap, and a large rotation angle. Because of its large distortion, serious noise and occlusion, it is necessary to find an algorithm that has good robustness to all kinds of distortion and noise. However, the image mosaic algorithm based on feature matching can be satisfied very well, so it has been widely used. In this paper, the image matching algorithm based on scale invariant feature (SFIT) is taken as the main content, and the key technology of fast automatic stitching of UAV image is studied, and realized by C programming. The research work of this paper mainly includes the following parts: 1. This paper summarizes the significance of UAV image mosaic technology and the research status at home and abroad, and determines the technical flow of UAV image splicing in this paper. 2. By programming in C language and compiling on VS2010 platform, the remote sensing image of unmanned aerial vehicle (UAV) without geographical coordinates can be automatically stitched quickly and automatically. The SIFT algorithm is used to extract the feature points and match the same name points. Then the mismatched point pairs are eliminated by RANSAC to complete the fine matching. Finally, using the geometric transformation model of the same-named point pair to solve the image, we complete the multi-image splicing of one airstrip and output the result. 3. The direct weighted average method is used to fuse the splicing image, deal with the color difference, illumination difference, splicing seam and so on. The results show that the SIFT algorithm can effectively extract a large number of feature points for image matching and has a good effect on UAV image mosaic without ground control points. However, because too many feature points will affect the computing speed, it is necessary to find an effective method to filter and better meet the requirements of the time-effectiveness of UAV remote sensing.
【学位授予单位】：中国地质大学（北京）
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P231;P237
【参考文献】
相关期刊论文 前10条
1 李原福;王学明;段金辉;;基于特征点检测的重叠图像拼接算法[J];吉林大学学报(信息科学版);2010年06期
2 雷小群;李芳芳;肖本林;;一种基于改进SIFT算法的遥感影像配准方法[J];测绘科学;2010年03期
3 邱建国;张建国;李凯;;基于Harris与Sift算法的图像匹配方法[J];测试技术学报;2009年03期
4 陈裕;刘庆元;;基于SIFT算法和马氏距离的无人机遥感图像配准[J];测绘与空间地理信息;2009年06期
5 李波;一种基于小波和区域的图像拼接方法[J];电子科技;2005年04期
6 徐光著;朱冰莲;丰建军;;一种改进的动态场景拼接算法[J];电子科技;2011年07期
7 卫征;方俊永;张兵;;非量测相机镜头光学畸变的改正[J];光学技术;2007年06期
8 宫本旭;李亮;;无人机遥感数据的获取和在矿山监测中的处理方法[J];贵州地质;2011年03期
9 程红;陈文剑;;基于SIFT算法的图像匹配剔点方法研究[J];地理与地理信息科学;2012年06期
10 马超;赵西安;王青松;;基于均匀特征匹配的无人机影像拼接[J];北京建筑工程学院学报;2013年04期
本文编号：2460289
本文链接：https://www.wllwen.com/kejilunwen/dizhicehuilunwen/2460289.html
