基于高分遥感影像的农村居民点提取研究

发布时间：2018-08-29 09:42

【摘要】：当今是3S技术高速发展的时代,高分一、二、三号卫星的发射和高分数据的生产,促进了3S技术之一遥感(Remote Sensing)技术的进步。近几年,数据更新速度加快,卫星遥感影像空间分辨率已达到亚米级,高分遥感影像的应用不断增加,而传统的目视解译专题信息的方法和提取精度不高的自动、半自动提取方法,已经不能满足应用需求,因此,精度高、流程化、智能化的提取方法成为解决这一问题的关键。在各提取方法中,能应用于平原地区的农村居民点提取较少,能应用于南方丘陵区的农村居民点提取的几乎没有,因此,加强对丘陵区农村居民点提取的研究具有重要意义。本文基于高分二号和高分一号卫星遥感影像进行丘陵区农村居民点的提取研究,以四川省绵阳市三台县的秋林镇为例,分三步提取农村居民点,第一步提取农村居民点房屋,第二步提取农村居民点附属用地(房屋周围的林盘、晒坝等用地),第三步合并农村居民点房屋和附属用地。本文使用ENVI系列遥感处理软件进行数据预处理、增强处理以提高数据显示效果、减少噪声、提高辨识度,再挖掘分析像元光谱信息、空间形状特征、纹理信息、空间关系特征、地形特征等,通过对监督分类、非监督分类、基于专家知识的决策树分类、基于规则和样本的面向对象信息分割和提取等方法进行实验研究,提出基于规则的面向对象特征提取法是提取农村居民点房屋的最优方法,并提取出农村居民点房屋,再在Arc GIS中利用农村居民点房屋与附属用地的相邻位置关系特征和形状特征,提取出农村居民点的附属用地,再将两者合并为农村居民点,从而实现对目标地物农村居民点的提取。最后将提取结果与同年度土地调查调绘的农村居民点矢量图进行叠加来评价其提取精度,并应用于现状分析。主要研究成果如下:(1)建立基于规则面向对象特征提取法的高分二号农村居民点房屋提取模型,以及高分一号城镇居民点提取模型。(2)结合基于规则面向对象特征提取法和GIS(地理信息系统)空间分析法,较精准地提取出农村居民点的位置和形状,与同年度土地调查调绘矢量数据相比,其位置精度、形状精度以及综合精度达到85%以上。(3)建立精度评价体系,包括位置精度、形状精度以及综合精度,并在Arc GIS中建立精度评价模型。(4)将该研究方法提取的数据应用于农村居民点的现状分析,发现农村居民点用地存在村庄分布分散和村庄用地规模小的问题。本文创新点如下:(1)使用高分二号卫星遥感影像对丘陵区的农村居民点进行提取研究,并结合基于规则面向对象特征提取法与GIS空间分析法实现智能化地提取农村居民点。(2)在挖掘特征信息中,除了对光谱特征、纹理特征、空间形状特征进行分析,还在地形因子以及空间关系方面加强了分析。随着高分遥感数据不断普及,基于高分遥感影像的研究以及对农村居民点的研究,在国土监测、地理国情普查、城市规划、住房建设、现代农业、灾害评估、环境变化预测等方面提供科学的依据,为城乡可持续发展提供决策性支持。
[Abstract]:Today is the era of rapid development of 3S technology. The launching of high-resolution satellites and the production of high-resolution data have promoted the progress of remote sensing technology, one of the 3S technologies. The method of visual interpretation of thematic information and the method of automatic and semi-automatic extraction with low precision can no longer meet the needs of application. Therefore, the key to solve this problem is the method of high precision, flowing and intelligent extraction. There are few rural settlements in Hilly area, so it is very important to strengthen the research on the extraction of rural settlements in Hilly area. Based on the remote sensing images of Gaofen-2 and Gaofen-1, this paper studies the extraction of rural settlements in Hilly area. Taking Qiulin Town of Santai County, Mianyang City, Sichuan Province, as an example, it extracts rural settlements in three steps. The first step is to extract rural residential housing, the second step is to extract rural residential ancillary land (forest pan around the house, sun dam and other land), and the third step is to merge rural residential housing and ancillary land. Identity, then mining and analyzing pixel spectral information, spatial shape feature, texture information, spatial relationship feature, terrain feature, etc. Through supervised classification, unsupervised classification, decision tree classification based on expert knowledge, rule-based and sample-based object-oriented information segmentation and extraction methods for experimental research, rule-based orientation is proposed. Object feature extraction method is the best method to extract rural residential housing, and extract rural residential housing, and then in Arc GIS using the adjacent location and shape characteristics of rural residential housing and ancillary land, extract rural residential land, and then merge the two into rural residential areas, so as to achieve the goal. The main research results are as follows: (1) Establish the extraction model of high-grade 2 rural residential housing based on rule-oriented object feature extraction method, and use the method to evaluate the extraction accuracy. (2) Combining the rule-based object-oriented feature extraction method and the spatial analysis method of GIS (Geographic Information System), the location and shape of rural residential areas are extracted more accurately. Compared with the vector data of land survey in the same year, the location precision, shape precision and comprehensive precision of rural residential areas are more than 85%. (3) Establish a precision evaluation system, including position precision, shape precision and comprehensive precision, and establish a precision evaluation model in Arc GIS. (4) Apply the data extracted by this research method to the analysis of the current situation of rural residential areas, and find the problems of scattered village distribution and small scale of rural residential land. The following: (1) Using high-resolution satellite remote sensing image to extract rural residential areas in Hilly areas, and combining rule-based object-oriented feature extraction method and GIS spatial analysis method to achieve intelligent extraction of rural residential areas. (2) In mining feature information, in addition to spectral features, texture features, spatial shape features analysis, but also. With the popularity of high-resolution remote sensing data, research on high-resolution remote sensing images and rural settlements, scientific basis is provided for land monitoring, geographic survey, urban planning, housing construction, modern agriculture, disaster assessment and environmental change prediction. Urban and rural sustainable development provides decision support.
【学位授予单位】：四川师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P237

【参考文献】