数字摄影测量中最小二乘影像匹配的探讨与研究
发布时间：2018-07-20 16:55
【摘要】：影像匹配作为图像处理领域的重要技术之一，是当前摄影测量学与计算机视觉领域的重要课题，它被广泛应用于矿产研究、医用人体影像配准、个人指纹识别、雷达跟踪监测、无人机导航探测等领域。现阶段影像匹配的研究越来越多，它的精度已经达不到许多工作的要求，最小二乘影像匹配由于它的高精度受到广泛关注。 本文是在参与研发地面摄影测量系统的最小二乘影像匹配的模块的情况下，通过研究核线影像生成、点特征提取、影像匹配和最小二乘影像匹配的各种算法，采用相关系数算法实现最小二乘影像匹配，使得影像可以达到子像素级。 本文根据原始影像的特点，选择了基于数字纠正的核线影像生成方法，研究了点特征提取算法以便于提供影像匹配的数据，在论文中分析了常用的Harris算子、Moravec算子和Forstner算子并进行比较，分析各个算子的优点和缺点，选取Harris算子提取特征点来作为影像匹配数据源。此外，在使用相关系数算法进行影像匹配和最小二乘影像匹配时，，分析了影像的纹理信息、搜索窗口的大小和阈值的选择对于匹配结果的影响，尽可能使得影像的匹配效率更高，匹配结果更好。
[Abstract]:Image matching, as one of the most important techniques in the field of image processing, is an important subject in the field of photogrammetry and computer vision. It is widely used in mineral research, medical human body image registration, personal fingerprint identification, radar tracking and monitoring. UAV navigation and detection and other fields. At present, there are more and more research on image matching, and its precision has not reached the requirements of many work. Because of its high accuracy, the least square image matching has been paid more and more attention. This paper is involved in the research and development of the least square image matching module of ground photogrammetry system, through the study of kernel line image generation, point feature extraction, image matching and least square image matching algorithms. The correlation coefficient algorithm is used to realize the least square image matching, so that the image can reach sub-pixel level. According to the characteristics of the original image, this paper chooses the kernel line image generation method based on digital correction, and studies the point feature extraction algorithm in order to provide image matching data. In this paper, the common Harris operator Moravec operator and Forstner operator are analyzed and compared, and the advantages and disadvantages of each operator are analyzed. Harris operator is selected to extract feature points as image matching data source. In addition, when using correlation coefficient algorithm for image matching and least square image matching, the texture information of image, the influence of the size of search window and the selection of threshold value on the matching results are analyzed. As far as possible, the image matching efficiency is higher and the matching result is better.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P231.5
本文编号：2134154
[Abstract]:Image matching, as one of the most important techniques in the field of image processing, is an important subject in the field of photogrammetry and computer vision. It is widely used in mineral research, medical human body image registration, personal fingerprint identification, radar tracking and monitoring. UAV navigation and detection and other fields. At present, there are more and more research on image matching, and its precision has not reached the requirements of many work. Because of its high accuracy, the least square image matching has been paid more and more attention. This paper is involved in the research and development of the least square image matching module of ground photogrammetry system, through the study of kernel line image generation, point feature extraction, image matching and least square image matching algorithms. The correlation coefficient algorithm is used to realize the least square image matching, so that the image can reach sub-pixel level. According to the characteristics of the original image, this paper chooses the kernel line image generation method based on digital correction, and studies the point feature extraction algorithm in order to provide image matching data. In this paper, the common Harris operator Moravec operator and Forstner operator are analyzed and compared, and the advantages and disadvantages of each operator are analyzed. Harris operator is selected to extract feature points as image matching data source. In addition, when using correlation coefficient algorithm for image matching and least square image matching, the texture information of image, the influence of the size of search window and the selection of threshold value on the matching results are analyzed. As far as possible, the image matching efficiency is higher and the matching result is better.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P231.5
【参考文献】
相关期刊论文 前5条
1 靳建立;范永弘;纪松;;基于面特征的整体影像匹配方法[J];测绘通报;2008年09期
2 王舒鹏;方莉;;利用Moravec算子提取特征点实现过程分析[J];电脑知识与技术;2006年26期
3 徐彦君,杜利民,侯自强,金贵昌;基于相位的尺度自适应立体匹配方法[J];电子学报;1999年07期
4 耿则勋;影像匹配中核线影像的快速生成算法[J];测绘学院学报;2000年03期
5 李芳芳;贾永红;肖本林;张谦;;利用线特征和SIFT点特征进行多源遥感影像配准[J];武汉大学学报(信息科学版);2010年02期
本文编号：2134154
本文链接：https://www.wllwen.com/kejilunwen/dizhicehuilunwen/2134154.html
