撞击坑识别方法综述  

A Review of Impact-Crater Detection

作　　者：刘宇轩[1,2] 刘建军[1] 牟伶俐[1] 李春来[1] 

机构地区：[1]中国科学院国家天文台,北京100012 [2]中国科学院研究生院,北京100049

出　　处：《天文研究与技术》2012年第2期203-212,共10页Astronomical Research & Technology

基　　金：国家"863"(2010AA122202;2008AA12A214)资助

摘　　要：目前国内外有多种撞击坑识别方法,在一定程度上实现了对撞击坑的识别提取,但是其准确性以及对数据的适应性不尽相同。首先对撞击坑识别研究进展进行概述,再对撞击坑识别方法进行归纳总结,指出不同方法的优缺点和适用条件。最后,对撞击坑识别研究存在的问题进行分析,提出了研究撞击坑识别的重点及解决途径。Currently many methods and algorithms have been applied for the impact-crater detection, and the achieved accuracies and adaptabilities are not quite similar among different approaches and data. This paper first summarizes the current progress of the area, and then discusses the advantages/disadvantages of different approaches and their applicable conditions. We finally analyze the main problems in the research of impact-crater detection by pointing out possible solutions. Current crater-detection approaches fall into four categories ： 1 ） Manual recognition, 2） Shape-profile fitting algorithms including Hough-Transformation, conic- fitting, and template-matching algorithms, 3 ） Machine-learning methods including SVM, genetic algorithm, and neural network methods, and 4 ） Geological-information based analysis using terrain data and spectral data. From the comparison between different approaches, we derive the following conclusions. 1 ） Manual recognition is suitable for cases with only image data available. The accuracy depends on the experience of researchers, and the efficiency is low. 2） The shape-profile fitting algorithms are suitable for the craters with simple structures and clear edges. 3 ） The Geological-information based analysis does not depend on image quality, but instead depends on illumination and resolution. With more and more highly accurate data, the crater detection will tend to use several types of data and combine results from these.

关 键 词：撞击坑识别 人工识别 形态拟合 机器学习 地学信息分析 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]