基于DEM的地形特征线提取方法研究与应用

发布时间：2019-05-10 18:38

【摘要】：地形特征线的提取研究作为地理信息技术中的基础环节是不容忽视的,在基于样本的地形合成等领域都有很好的应用。在数字高程模型中包含了很多重要的地形特征信息,其中山脊线和山谷线是地形特征信息中最重要的特征线。能否准确的从地形中提取出山脊线和山谷线的特征信息,直接影响着基于样本的地形合成系统的效果。因此,本文针对地形特征线提取方法中可以改进的方面,设计了一种新的显著度可控的地形特征线提取算法,与现有的算法进行实验对比。同时,将本文的算法应用于一套基于样本的地形合成系统。本文的创新点如下:(1)提出了一种基于显著度值的细线化算法。该算法从特征点的特征显著程度出发,使特征线雏形细化过程中挑选待剔除的特征点时,能够保留特征更为明显的特征点,以此达到细化后的特征线能够保持较强特征性。同时,相比于传统的细线化算法,本文算法在对特征线细化后能够保持更好的特征线形态。(2)提出了一种环路检测与破环算法。该算法按照扫描线的顺序,利用深度优先遍历方法对非特征像素点进行数值填充,使得非特征像素点以区域形式进行划分。通过对区域的分类,能够实现对环路区域与非环路区域的分类,而区域内部非特征点数量的统计,能够实现对大型环路与小型环路的归类。最后,根据对每一个小型环路区域边界线段的检测,筛选特征性最弱的边界线段进行剔除,从而实现小型环路的拆除以及对大型环路的保留。(3)提出了一种特征图分解及特征线筛选算法。该算法以特征线分支(除端点外,组成分支的特征点度为2)为基本单元对特征图进行分解,从特征性最强分支开始筛选特征性最相似分支组成特征线。相比于传统的方法,本文方法既避免了特征线的过度延伸,又适用于带环路的特征图,使得特征线呈现的效果更接近于实际观察所得。基于以上创新点,本文设计并实现了一套地形特征线提取系统,该系统填补了传统方法中无法识别特殊地形的空白,且提取的特征线效果更加符合实际。同时,本文还实现了一套基于样本的地形合成系统,将本文提出的地形特征线提取方法应用于系统中,使得合成过程中能找到更合适的特征匹配块。
[Abstract]:As the basic link of geographic information technology, the extraction of terrain feature lines can not be ignored, and has good applications in sample-based terrain synthesis and other fields. There are many important topographic feature information in the digital elevation model, among which ridge line and valley line are the most important feature lines in terrain feature information. Whether the feature information of ridge line and valley line can be extracted accurately from terrain directly affects the effect of sample-based terrain synthesis system. Therefore, a new terrain feature line extraction algorithm with controllable saliency is designed in this paper, which can be improved in the terrain feature line extraction method, and the experimental results are compared with the existing algorithms. At the same time, the proposed algorithm is applied to a sample-based terrain synthesis system. The innovations of this paper are as follows: (1) A fine line algorithm based on saliency value is proposed. Based on the feature saliency of the feature points, the algorithm can preserve the feature points with more obvious features when the feature points to be eliminated are selected in the process of feature line refinement, so that the refined feature lines can maintain strong characteristics. At the same time, compared with the traditional fine line algorithm, this algorithm can maintain a better feature line shape after thinning the feature line. (2) A loop detection and loop breaking algorithm is proposed. According to the order of scanning lines, the depth-first traverse method is used to populate the non-feature pixels, so that the non-feature pixels are divided in the form of regions. Through the classification of the region, the classification of the loop area and the non-loop area can be realized, and the statistics of the number of non-characteristic points in the region can realize the classification of the large loop and the small loop. Finally, according to the detection of the boundary line segment of each small loop area, the boundary segment with the weakest characteristic is screened out. Thus, the removal of small loop and the retention of large loop are realized. (3) A feature graph decomposition and feature line screening algorithm is proposed. In this algorithm, the feature line branch (except the end point, the characteristic point degree of the component branch is 2) is used as the basic unit to decompose the feature graph, and the characteristic line is selected from the strongest characteristic branch to form the characteristic line. Compared with the traditional method, this method not only avoids the overextension of the feature line, but also is suitable for the feature map with loop, so that the effect of the feature line presentation is closer to the actual observation. Based on the above innovations, a set of terrain feature line extraction system is designed and implemented in this paper, which fills in the blank that the traditional method can not recognize the special terrain, and the effect of the extracted feature line is more in line with the reality. At the same time, this paper also implements a set of sample-based terrain synthesis system, and applies the terrain feature line extraction method proposed in this paper to the system, so that more suitable feature matching blocks can be found in the synthesis process.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P208

【参考文献】