基于DEM纹理特征的月貌自动识别方法探究
发布时间：2019-03-25 19:02
【摘要】：月海和月陆是两种最主要的月貌单元,对于月海及月陆快速准确地识别是进行各项月球研究的重要基础。目前,月海和月陆的识别大多采用DEM结合其派生地形因子建立指标体系的方法。这种方法虽然可在宏观尺度对月海和月陆进行识别和提取,但仍存在2个问题:(1)可扩展性差,不同地区难以共用同一套地形因子构建指标体系;(2)指标体系中各因子权重设置具有较大的主观性。针对以上问题,本文以"嫦娥一号"探测器获取的全月球DEM数据,从月表地形纹理特征的角度出发,提出一种以月表DEM数据识别月海、月陆的自动快速的方法。首先,利用灰度共生矩阵模型,以DEM数据为基础,实现对典型月海、月陆地形纹理特征的量化,然后,对量化指标的筛选,构建能有效区分两类月表形貌单元的特征向量。在此基础上,选用离差平方和作为识别器,最终实现对月海和月陆的自动识别。本文识别方法的整体识别率达到85.7%;综上可知,该方法既能克服原有方法中因子权重设置的主观性,又具有较好的通用性。
[Abstract]:Lunar sea and lunar land are the two most important lunar feature units. The fast and accurate recognition of lunar sea and lunar land is an important basis for all kinds of lunar studies. At present, the identification of lunar sea and lunar land mostly uses DEM combined with its derived terrain factors to establish the index system. Although this method can be used to identify and extract lunar sea and lunar land on a macro scale, there are still two problems: (1) poor scalability, it is difficult to share the same terrain factors in different areas to construct an index system; (2) the weight setting of each factor in the index system is subjective. In order to solve the above problems, this paper presents an automatic and rapid method to identify the moon sea and lunar land with the lunar surface DEM data from the point of view of the topography texture features of the moon surface, based on the all-moon DEM data obtained by Chang'e-1 spacecraft. Firstly, the gray-level co-occurrence matrix model is used to quantify the texture features of the typical lunar sea and land topography based on the DEM data. Then, the feature vectors which can effectively distinguish the two types of monthly surface topography units are constructed by screening the quantized indexes. On this basis, the sum of squared deviation is selected as the discriminator to realize the automatic recognition of lunar sea and lunar land. The overall recognition rate of this method is up to 85.7%, which shows that the method can not only overcome the subjectivity of factor weight setting in the original method, but also has better universality.
【作者单位】： 南京师范大学地理科学学院;
【基金】：国家自然科学基金项目(41171320) 江苏省高校自然科学基金重大项目(13KJA170001) 江苏省研究生科研创新计划项目(KYLX_0701)
【分类号】：P184;P208
[Abstract]:Lunar sea and lunar land are the two most important lunar feature units. The fast and accurate recognition of lunar sea and lunar land is an important basis for all kinds of lunar studies. At present, the identification of lunar sea and lunar land mostly uses DEM combined with its derived terrain factors to establish the index system. Although this method can be used to identify and extract lunar sea and lunar land on a macro scale, there are still two problems: (1) poor scalability, it is difficult to share the same terrain factors in different areas to construct an index system; (2) the weight setting of each factor in the index system is subjective. In order to solve the above problems, this paper presents an automatic and rapid method to identify the moon sea and lunar land with the lunar surface DEM data from the point of view of the topography texture features of the moon surface, based on the all-moon DEM data obtained by Chang'e-1 spacecraft. Firstly, the gray-level co-occurrence matrix model is used to quantify the texture features of the typical lunar sea and land topography based on the DEM data. Then, the feature vectors which can effectively distinguish the two types of monthly surface topography units are constructed by screening the quantized indexes. On this basis, the sum of squared deviation is selected as the discriminator to realize the automatic recognition of lunar sea and lunar land. The overall recognition rate of this method is up to 85.7%, which shows that the method can not only overcome the subjectivity of factor weight setting in the original method, but also has better universality.
【作者单位】： 南京师范大学地理科学学院;
【基金】：国家自然科学基金项目(41171320) 江苏省高校自然科学基金重大项目(13KJA170001) 江苏省研究生科研创新计划项目(KYLX_0701)
【分类号】：P184;P208
【参考文献】
相关期刊论文 前9条
1 欧阳自远;月球地质学[J];地球科学进展;1994年02期
2 薄华;马缚龙;焦李成;;图像纹理的灰度共生矩阵计算问题的分析[J];电子学报;2006年01期
3 刘凯;汤国安;陶e，
本文编号：2447230
本文链接：https://www.wllwen.com/kejilunwen/tianwen/2447230.html
