首页 >  , Vol. , Issue () : -

摘要

全文摘要次数: 1456 全文下载次数: 2187
引用本文:

DOI:

10.11834/jrs.20211454

收稿日期:

2021-06-30

修改日期:

2021-09-13

PDF Free   EndNote   BibTeX
结合影像和DEM的撞击坑检测方法综述
刘慧珍, 李大刚, 徐懿
澳门科技大学
摘要:

撞击坑在月球及类地行星表面普遍存在,是开展天体研究的重要数据来源。目前已有很多种撞击坑检测的方法,主要基于撞击坑的影像数据或者DEM(Digital Elevation Model);而随着数据的不断丰富,近年来也有综合使用这两类数据的趋势出现。本文聚焦并全面梳理了现有综合使用影像和DEM的这部分撞击坑检测方法并将之分为四类:第一类是以影像数据为主DEM为辅的方法;第二类是DEM为主辅以影像信息的方法;第三类是组合使用两类数据的方法,该类方法从影像数据和DEM中分别检测撞击坑,然后将两者的结果对照合并从而得到最终的结果;最后一类我们称为融合的方法,用覆盖同一区域的影像和DEM收集训练样本,将两类样本先融合再作为撞击坑检测算法的输入,从而得到统一的检测模型。本文整理了这四类方法的代表性工作,分析总结了不同方法的优势和适用范围,最后指出了该领域的研究热点及未来的发展方向。

A survey on impact crater detection methods using both image data and DEMs
Abstract:

Impact craters are ubiquitous on the surface of the moon and most planets, which is an important data source for celestial body research. At present, there are many impact crater detection methods, and most early ones are based on image data or DEM alone. There is a trend of using both images and DEMs in recent years. In this paper, we examine the existing impact crater detection methods using both images and DEMs and classify them into four categories: first, image based DEM supplemented method; second, DEM based image supplemented method. All of these two methods have their advantages in impact crater detection, for example, they take one data as the main data and the other as the supplemental data to optimize the impact crater detection results, and the accuracy can be improved; Third, combined method, in which impact craters are first detected using image data and DEM separately, and then combined to generate the final results. By this way can make up for the shortcomings of using only one data and make good use of the respective advantages of image data and DEM as well as enrich the impact crater data sets; and finally the fusion method, in which image and DEM covering the same area are used together as training samples, and the samples are fused as the input to the algorithm, so as to obtain an unified detection model. The advantages of this method are that the fusion of image data and DEM before the crater detection helps to train the unified detection model, it can input a single type of data to detect craters and meet the requirements of the transfer learning. In this paper we summarize the representative works of these four categories, and analyze the advantages and scope of applicability of different methods, and finally point out the research hotspots and future development directions in this field.

本文暂时没有被引用!

欢迎关注学报微信

遥感学报交流群