利用多级弦长拱高复函数进行面实体综合相似性度量研究

发布时间：2018-06-11 10:13

本文选题：GlobeLand + 多级拱高复函数　；参考：《中国图象图形学报》2017年04期

【摘要】：目的全球最高30 m分辨率的地表覆盖数据GlobeLand30具有高分辨率、高精度等特性,为全球制图提供了重要的数据来源。如何快速准确地识别GlobeLand30与矢量数据中的同名实体,对于空间数据的更新、集成与融合具有重要的意义,针对当前该数据与矢量数据匹配识别方法存在的不足,本文提出一种新的综合相似性度量方法。方法利用面实体轮廓线的多级弦长、拱高和中心距离等特性,构造多级弦长拱高复函数对其整体和细节特征进行描述;然后对面实体轮廓线进行等间隔重采样,通过快速傅里叶变换得到傅里叶描述子对面实体间的形状相似性进行度量;最后将面实体的位置、大小、方向和形状进行加权综合,得到一种综合相似性度量模型。结果将本文综合相似性度量模型应用到GlobeLand30与矢量数据面状水体的匹配中,实验结果为查准率P为100%,查全率Q为97.1%,匹配速度和准确率优于其他文献所提出的方法,当拱高级数为4时,匹配效果达到最优;最后,将该相似性度量模型应用到GlobeLand30数据化简和光滑前后的相似性度量上,也取得了很好的应用效果。结论本文方法适用于GlobeLand30与矢量数据的相似性度量,对于GlobeLand30与矢量数据的集成与融合,对利用GlobeLand30进行矢量数据的生产与更新具有重要的意义。
[Abstract]:Objective Globe Land30, which has the highest resolution of 30 m in the world, has the characteristics of high resolution and high precision, which provides an important source of data for global mapping. How to identify the entities of the same name between Globe Land30 and vector data quickly and accurately is of great significance for the updating, integration and fusion of spatial data. In this paper, a new synthetic similarity measurement method is proposed. Methods based on the characteristics of multi-level chord length, arch height and center distance, the Gao Fu function of multi-chord arch was constructed to describe the whole and details of the contour, and then the contour was resaminated at equal intervals. The shape similarity between the entities on the opposite side of the Fourier description is measured by the fast Fourier transform. Finally, the position, size, direction and shape of the surface entity are weighted synthetically, and a comprehensive similarity measurement model is obtained. Results the synthetic similarity measurement model was applied to the matching between Globe Land30 and vector data surface water body. The experimental results showed that the precision P was 100, the recall rate Q was 97.1.The matching speed and accuracy were better than the methods proposed in other literatures. When the arch number is 4, the matching effect is optimal. Finally, the similarity measurement model is applied to the simplification of Globe Land30 data and the similarity measurement before and after smoothing. Conclusion this method is suitable for measuring the similarity between Globe Land30 and vector data. It is of great significance for the integration and fusion of Globe Land30 and vector data, and for the production and updating of vector data using Globe Land30.
【作者单位】：信息工程大学;
【基金】：国家自然科学基金项目(41571399) 地理信息工程国家重点实验室开放基金项目(SKLGIE2015-M-4-6)~~
【分类号】：P208

【参考文献】