【作者】 姜玲；

【导师】 黄家柱；

【作者基本信息】 南京师范大学 ， 地图学与地理信息系统， 2008， 硕士

【摘要】 论文在开展柬埔寨洞里萨湖地区砂矿找矿项目的基础上,深入进行地貌制图研究。该项目目标为:确定已知砂矿点的地貌类型,进行区域地貌类型解译,绘制地貌解译图,提供砂矿找矿远景靶区。论文针对柬埔寨地质资料严重缺乏及洞里萨湖地区地貌特点,采用TM、ALOS遥感影像数据结合DEM数据开展地貌及微地貌制图研究,实现数字地貌制图。采用TM影像数据直接选取端元波谱采用波谱角和最大似然法对地表覆被进行监督分类;采用ALOS影像数据,选取植被指数RVI、DVI、TSAVI及拉伸后的DEM数据参与非监督分类,实现对地表覆被的细分,为数字地貌制图建立基础;利用数字地形分析技术,运用德国eCognition软件的图像分割功能,将由光照投影图或遥感影像分割获得的矢量多边形数据代替传统的格网作为统计范围,改良传统地形起伏度、平均高程、平均坡度等地形因子的计算方法,实现基于SRTM DEM地形因子的提取。针对洞里萨湖地区特殊的堆积地貌环境,采用先对SRTM DEM数据进行负值转换再提取水系的方法,取得了很好的效果。分析比对上述研究结果,选择合适的分类图像和专题图参与数字地貌制图,并将面向对象影像分类技术引入数字地貌制图中。本文利用遥感图像提取地貌和微地貌信息并将DEM作为一个分量,将遥感影像、遥感分类图像、地形因子专题图等共同参与图像分割与分类,最终实现地貌类型的划分及数字地貌制图。本文主要结论和创新点如下:1.将由光照投影图或遥感影像分割获得的矢量多边形数据代替传统的格网作为统计范围,可以明显提高地形因子对地形的表现力,并且保持了地貌边界与实际地表的吻合性。2.将DEM数据处理后的图像,如拉伸后的DEM、平均高程图、地形起伏度图等与遥感数据叠合,共同参与遥感图像分类。该方法实现在遥感图像分类过程中兼顾高程信息,误分、漏分斑块明显减少,分类效果得到明显改善,可识别的地貌类别得以增加。地貌分类界线兼具土地覆被差异性与地形因子的差异性。3.将面向对象影像分类的分类方法引入地貌分类制图,可以明显减少像元孤岛及分类图中的斑点或空洞,有效去除由遥感数据质量引起的“椒盐”噪声带来的影响。更多还原

【Abstract】 The paper focus on the work of finding mineral in the area of The Tonle Sap, Cambodia, starting on landform research, nailing down the landform type of possible existent placer and its distribution, picking up the configuration and modality of landform type, and protracting landform information graphs. According to the landform type in the area of the Tonle Sap, Cambodia, we try to mainly accumulate fluvial landform characteristics of landform and fluvial corrosive landform. So we can start on the research of landform and micro-landform charting to carry out digital landform charting based on the remote sensing image of pixel by using TM, ALOS data and SRTM DEM data.We used the TM image data, we directly select pixel spectrum, monitoring and classifying surface cover by methods of Spectral Angle Mapper and Maximum likelihood. For ALOS image data, we select Vegetation index RVI、DVI、TSAVI and DEM data participate in unsupervised classification finished the classification of land-cover and built a basis for digital landform charting.Using digital terrain analysis technique to realize the extraction of tenain, based on SRTM DEM. During the process of the terrain factor extracting, by the using of the image segmentation function of the German software eCognition, take the light projection or segmentation of remote sensing images obtained polygon vector data instead of the traditional grid as Statistics. Modify traditional ups and downs of the tenain, with an average elevation, the average slope of tenain, and other methods of calculation. Aiming at Sap Lake area for the accumulation of special landscape environment, the first use of anti-SRTM DEM data to extract water from the method and achieved very good results. Analyze the results of the research, and choose the appropriate classification of images and thematic maps participation in the digital landscape mapping, and object-oriented image classification technology into digital mapping of the landscape. Use remote sensing images from micro-topography and landscape, and take DEM as a component of remote sensing images, pixel-based classification of remote sensing imagery, tenain of thematic maps such as joint participation image segmentation and classification, and eventually realize the types of landscapes and figures landscape mapping, obtain topography, micro-topography thematic map.The results of the research and the innovative points:1. The vector polygon data which obtains by the illumination projection or the remote sensing phantom division will replace traditional the graticule mesh to take the statistical scope, may enhance the tenain factor obviously to the tenain expressive force, and maintained the landform boundary and the actual surface tallies the nature.2. The remote sensing data with DEM conelation images, such as stretched DEM, average elevation chart, tenain prominency chart together fellowship the image classification. This mehod help the researchers can giving consideration to the elevation Information when they do the remote sensing image classification.The result of the classification has a good classifying quality, with less mistakes, more recognized kinds.The taxonomic boundary line get in this way with the diversity of land-cover and terrainfactor.3. Introducing the object-oriented remote sensing image classification method into the cartography of geomorphology classification,the method can helping clearing the isolated points, protruding points, and removaling the influence of salt and pepper noise.更多还原