自动空三点特征高精度影像匹配研究

发布时间：2018-01-19 14:27

本文关键词： 影像匹配自动空三高精度匹配最小二乘匹配光束法平差　出处：《西安科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：数字摄影测量技术的不断发展,已经日益成为获取和更新国家基础地理信息数据的一种重要手段,为测制和更新国家地理信息数据,提供丰富的信息来源。同时,数字摄影测量技术在国民基础经济建设中起着重要的作用,被广泛应用于土地资源的调查研究、自然灾害监测、城市规划、数字城市建设等领域。影像匹配技术一直是数字摄影测量的关键技术,是数字摄影测量和计算机视觉领域共同研究的核心内容。影像匹配技术在摄影测量、计算机视觉、城市三维建模以及军事应用等多个领域都有着广阔的应用前景。在数字摄影测量领域中,影像匹配技术是自动空中三角测量的基础,是空间数据产品生产的基础,也是建筑物等人工地物目标三维重建的基础。现阶段影像匹配的研究算法也越来越多,对匹配精度要求也越来越高,针对航空影像的自动空三,影像匹配的高质量与高精度要求显得尤为重要。航空影像采集不可避免的会受到自然场景以及天气变化的影响,因此,部分特征点匹配算法会产生误匹配点,影响匹配质量,降低匹配精度,最终导致空三精度的降低。本文对航空影像自动空三中的点特征高精度匹配问题进行了研究和探讨,重点围绕常规影像以及倾斜影像的高精度匹配这两个主要方面进行了算法研究和论证实验。针对常规影像高精度匹配的问题,本文提出了一种基于高程平面约束的物方匹配方法,该方法基于高程平面约束匹配,在匹配过程中引入像点偏移量来搜索匹配,同时像点偏移量也可为最小二乘匹配提供可靠稳定的初值;针对倾斜影像的高精度匹配问题,本文结合物方面元匹配方法的特点提出了一种适应于倾斜影像高精度匹配的物方面元匹配方法,该方法通过计算物方平面与像平面之间的H矩阵,将像方信息由H矩阵变换到物方平面上并结合最小二乘匹配进行高精度匹配计算。实验中,为了更好的验证本文算法的正确性,针对常规影像的高精度匹配,本文分别设计了共线条件约束的多片最小二乘匹配算法,铅垂线轨迹法(VLL法)与本文算法的对比实验,最终实验表明,本文算法相比其他算法在匹配结果以及光束法平差结果方面都具有优势。同样针对倾斜影像的高精度匹配,本文也设计了对比实验,结果表明,本文提出的基于物方面元的倾斜影像高精度匹配方法改善了匹配结果,提高了光束法平差的精度。
[Abstract]:With the continuous development of digital photogrammetry technology, it has become an important means to obtain and update national basic geographic information data, for measuring and updating national geographic information data. At the same time, digital photogrammetry technology plays an important role in the national basic economic construction, and is widely used in the investigation and research of land resources, natural disaster monitoring, urban planning. Image matching is the key technology of digital photogrammetry and the core of the research in the field of digital photogrammetry and computer vision. Image matching technology is used in photogrammetry. In the field of digital photogrammetry, image matching is the basis of automatic aerial triangulation. It is the basis of the production of spatial data products, and also the basis of 3D reconstruction of objects on buildings and other human sites. At present, there are more and more algorithms for image matching, and the accuracy of image matching is becoming more and more high. The requirement of high quality and high precision of image matching is very important for the automatic space 3 of aerial image. The acquisition of aerial image will inevitably be affected by natural scene and weather change. The partial feature point matching algorithm will produce mismatch points, which will affect the matching quality and reduce the matching accuracy. Finally, the accuracy of space three is reduced. This paper studies and discusses the problem of high accuracy matching of point feature in automatic space three of aerial image. Focusing on the conventional image and the tilt image of high-precision matching these two main aspects of the algorithm research and demonstration experiment, aiming at the conventional image high-precision matching problem. In this paper, a method of object square matching based on height plane constraints is proposed. This method is based on height plane constraint matching. In the process of matching, image offset is introduced to search for matching. At the same time, the image offset can also provide a reliable and stable initial value for least square matching. Aiming at the problem of high precision matching of inclined images, this paper presents a new method of object aspect element matching, which is suitable for high precision matching of tilted images according to the characteristics of object aspect element matching method. By calculating the H matrix between the object square plane and the image plane, the image square information is transformed from the H matrix to the object square plane and the high-precision matching calculation is carried out by combining the least square matching. In order to better verify the correctness of this algorithm, aiming at the high precision matching of conventional images, this paper designs a collinear conditional constraint multi-slice least squares matching algorithm. The comparison experiment between the method of lead vertical trace and the algorithm of this paper shows that. Compared with other algorithms, this algorithm has advantages in matching results and beam adjustment results. For the high-precision matching of tilted images, a comparative experiment is also designed in this paper, and the results show that. In this paper, the high precision matching method based on the object aspect element is proposed to improve the matching result and the accuracy of the beam method adjustment.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P231

【参考文献】