基于DEM小流域复杂网络的黄土高原地貌自动识别研究

地貌识别,对于人类建设,地质构造研究,环境治理等相关领域都有着重要意义。传统的基于像素单元或面向对象的地貌识别方法存在局限性。由于流域小单元具有表面形态的完整性,在地貌演化中具有明确的地理意义,基于流域小单元的地貌识别成为了该领域的一个新热点。然而,基于传统地形因子的地貌识别方法使用的因子往往较为单一或者在地学描述上存在重复性,目前尚无针对流域小单元进行空间结构描述和拓扑关系特征量化的地貌识别研究。基于此,本文基于DEM进行水文分析并通过坡谱方法解决了小流域稳定面积难以确定的问题,在黄土高原样区提取了181个稳定小流域。根据复杂网络理论和地貌学原理提出了流域加权复杂网络的概念和相应的8个定量指标用于流域空间结构的模拟和量化描述。最后采用了基于决策树的XGBoost机器学习算法进行地貌识别,实验对于黄土高原主要地貌类型的识别显现出较好的效果,Kappa系数为86.00%,总体精度达到了88.33%。对于地貌形态特征明显的地貌,复杂网络方法其顾及空间结构和拓扑特征的特性导致了其较高的识别性能,精度和召回率都在90%~100%之间。通过与前人的研究进行对比,其识别结果亦呈现出较高的精度,这些都验证了流域加权复杂网络是一种基于流域小单元地貌识别的高精度且有效的方法。     

黄土地貌类型的坡谱自动识别分析

地貌形态特征识别与分类,对生态环境、水文研究及地质构造分析等地学研究具有重要意义,已成为现代地貌学的一个研究热点。传统的统计模式识别方法精度较低,难以解决线性不可分的模式分类问题。人工方法虽然识别精度高,但因各人认知偏差导致的识别误差难以控制。人工神经网络作为一种动态信息处理系统,能有效解决线性不可分的地貌类型识别问题。坡谱是利用微观地形定量指标来反映宏观地形特征的有效方法,在地貌学研究中正受到广泛的关注。本文以陕北黄土高原8个不同地貌类型区的数字高程模型（DEM）为实验数据,以流域为分析单元,提取坡谱及其特征指标作为描述地形特征的定量因子,并通过BP神经网络的构建与学习,进行黄土地貌类型自动识别。实验结果表明,在8种地貌类型的样本数据中,第1次实验正确识别率平均值达70%;第2次和第3次实验中,去除相似度较高的峁状丘陵沟壑或峁梁状丘陵沟壑任一种地貌类型后,正确识别率平均提升为80%和85%。经Kappa系数验证,该方法能以DEM数据有效识别不同类型的黄土地貌。     

地质界线图层的自动化生成方法

地质界线要素是地质图空间数据库的基本要素之一。地质界线图层通常是地质图空间数据库中图元最多的一个图层,其数字化和质量检验工作较为繁重。针对已有地质体面图层,而地质界线图层缺失的情况,本文提供一种快速补充地质界线图层的方法。该方法也可对原生地质界线图层进行属性数据的自动化质量检验。其主要处理环节包括：首先,基于地质体面图层,构建地质体邻接关系图,可实现对地质界线高效、准确地提取;其次,基于产状数据与断层数据,运用地层接触关系识别规则,可分别实现对基本地层接触关系和断层接触关系的自动识别;最后,针对两邻接地质体面实体多段界线间的接触关系识别结果不一致的情形,运用“平行不整合优先”原则和“长度优先”原则,进行了合理的综合判别。基于1:50 000庐山幅地质界线图层的自动化生成实验表明,本文方法可高效、快捷、准确地生成地质界线图层,既能够满足自动化补充地质界线图层的需要,又可作为检验地质界线图层属性的有效工具。     

基于POI数据的城市建筑功能分类方法研究

对建筑物进行建模与分析是智慧城市建设的重要任务之一。将城市中数量庞大的建筑物按功能分类,辅助认知城市内部空间结构,对政府部门开展人口估计,土地管理,城市规划等工作具有重要意义。本文以蕴含丰富语义信息的兴趣点（POI, Point of Interest ）作为主要信息源,针对POI分布稀疏导致大量建筑物无法识别出功能的问题,改进了传统的城市功能区定量识别方法。该方法为建筑物内部及周边一定区域范围内的POI赋予反距离权重,通过计算不同类型POI的加权频数密度比例来识别建筑物功能类型。文中以北京市西四环中路附近5000多栋建筑物为例进行实验验证,实现了将目标区域内的建筑物按功能类型划分为居住、商业、公服和3种混合类型,识别率达93.04%,与人工判别的结果对比得出总体分类精度达91.18%。该方法采用易于获取的互联网POI数据,可以实现大范围建筑物功能类型的快速自动化识别,丰富了城市建筑模型语义属性,扩展了POI数据的应用范围。     

图像数据融合的地貌类型识别分类与制图

计算机遥感地貌制图是利用航空像片或者卫星影像进行识别制图;另是利用DEM数据融合计算提取。对此,本文介绍了一种对区域基本地貌形态类型进行计算机自动分类的方法。它通过识别标志在影像上对地貌分布区进行数字化,把TM影像中的地貌信息和从DEM中提取出来的地貌信息结合,以划分出详细的地貌类型:如河北省沽源县的台地、河谷平原、开阔平原、丘陵、低山和中山6大类。最后,通过一定的算法进行分类成图。     

基于房屋轮廓与纹理的三维建筑模型分层次聚类研究

目前三维建筑模型已广泛应用于城市规划,导航和虚拟地理环境等领域.不同细节的模型是LOD( Level of detail )技术的基础,由于三维模型的生产成本高昂,模型自动化简逐渐引起了学者的关注.三维模型化简包括单模型化简和多模型综合2方面,目前单个模型的化简研究比较多,而模型群组综合的研究仍然处于起步阶段.本文主要研究模型群组的聚类综合,提出一种基于房屋轮廓与纹理的分层次聚类算法:首先,基于房屋的底面轮廓构建约束Delaunay三角网,以道路为基准对三角网进行划分,通过可视分析构建初始的邻接图,使建筑群组分类符合城市形态学;其次,将房屋纹理引入三维模型群聚类的过程,使用SOM( Self-organizing Map )智能分类算法对纹理进行分析,然后分割邻接图;最后,以最邻近距离对邻接图构造最小生成树,并进行线性检测,将离散的建筑合并到已聚类的群组中,最终完成模型的合并.本文利用纹理辅助轮廓特征,实现三维建筑模型的聚类,符合人类的视觉习惯,实验结果证明了本文方法的有效性.     

黄土高原地貌类型界线划定研究进展

区分地理实体最直接有效的方式在于对其界线作出划定。目前,黄土高原地貌类型界线划定多是在分类基础上按照分类界线、自然区划界线来界定。基于不同数据源及其表达方式,本文追踪前人对黄土高原地貌类型界线划定的研究进展,从形态成因的地貌分类、数字地貌分类等分类体系中总结了黄土地貌类型界线的内涵,分析了基于自然语言和数字环境下定量描述的优缺点和存在的问题;并梳理了黄土地貌类型界线的表示方法以及基于数字地形分析技术的地貌类型定量识别及其划分方法;从地貌界线确定与分类体系的关系、地貌界线划定的理论与方法参考、地貌界线划定的尺度效应3个方面对地貌类型界线做出了讨论分析与展望,以期为黄土地貌区划的相关理论研究提供背景基础,为当地实践工作等提供理论依据和支撑。     

基于多源数据和船舶停留轨迹语义建模的港口目标识别

港口目标识别是海事船舶监管的重中之重,船舶自动识别系统（Automatic Identification System,AIS）所获取的船舶活动信息,可为港口目标识别提供高时相和高精度的船舶航行数据。为了探究AIS数据在港口目标识别中的应用,提出一种基于多源数据和船舶停留轨迹语义建模的港口目标识别方法。通过数据挖掘和语义信息增强构建船舶停留轨迹语义模型,识别船舶港口停留轨迹;建立基于随机森林的船舶停留方式分类模型,分类船舶泊位停留轨迹和船舶锚地停留轨迹,并利用空间逐级合并方法提取港口泊位和港口锚地;综合船舶泊位停留轨迹、道路、海岸线、水深、土地利用与土地覆盖等数据,顾及情景-领域知识实现港口目标识别。基于2017年96 790艘船舶的超8300万条AIS轨迹记录,应用本文方法识别南海研究区的港口目标。实验结果表明,本文方法对于船舶轨迹停留行为总体分类精度为0.9477, Kappa系数为0.8948。提取出南海研究区447个港口区域,与Google Earth影像叠加验证结果表明,提取结果均位于真实的港口影像内,相较于Natural Earth数据集中包含的南海区域24个港口点位,提取结果的完整性大大增强。因此,基于多源数据和船舶停留语义建模的港口目标识别方法对于港口目标识别具有较高的准确性和完整性。此外,该方法提取的港口区域可为基于遥感影像的港口目标识别提供靶区,从而提高大区域甚至全球范围内港口目标动态识别的效率。     

彩色扫描地图点状符号的自动识别

本文初步探讨了彩色扫描地图上点状符号自动识别的方法及软件设计的模型和框架.彩色扫描地图上规则符号的主要视觉特征是颜色、形状及大小.识别符号建立在符号模式库的基础上,根据模式库中提供的颜色类型,通过颜色分层方法去寻找彩色扫描地图上的目标,然后提取其轮廓线,获取形状和大小等特征量,最后用符号特征相量与模式库中的特征相量相匹配,对符号进行判别分类,并通过坐标转换,输出符号的中心点地理坐标及符号的其他对应属性.本文最后以彩色地图上电信号强度符号的识别为实例,论述了具体应用系统的设计和开发.     

基于DEM的岷江上游流域构造活动强度分析

目前,利用DEM衍生的地貌参数对大面积区域构造活动进行量化分析的研究相对较少。本文以岷江上游流域为例,利用ASTER GDEM数据和GIS技术,提取了该流域面积高程积分值（HI）、河长坡降指数（SL ）、流域盆地不对称度（AF）、盆地形状指数（BS）、谷底宽度与谷肩高度比（VF）和水系分维数（FD）6种地貌参数,对其进行了等级划分,并量化了研究区相对构造活动强度（IAT）。结果显示：岷江上游流域有较高的SL、AF、BS值,较低的VF及FD值,中等HI值,其IAT值为1.33,分类等级为1级,表明该流域侵蚀程度强烈, 流域形态狭长, 河谷呈V型形态,受构造活动影响显著,地貌发育处于幼年期晚期,中年期早期。研究结果与区域地质背景较为吻合,该方法可为进一步探讨区域构造活动差异性提供新的手段。     

Spatial structure characteristics detecting of landform based on improved 3D Lacunarity model

The spatial structure characteristics of landform are the foundation of geomorphologic classification and recognition.This paper proposed a new method on quantifying spatial structure characteristics of terrain surface based on improved 3D Lacunarity model.Lacunarity curve and its numerical integration are used in this model to improve traditional classification result that different morphological types may share the close value of indexes based on global statistical analysis.Experiments at four test areas with different landform types show that improved 3D Lacunarity model can effectively distinguish different morphological types per texture analysis.Higher sensitivity in distinguishing the tiny differences of texture characteristics of terrain surface shows that the quantification method by 3D Lacu-narity model and its numerical integration presented in this paper could contribute to improving the accuracy of land-form classifications and relative studies.     

Automatic recognition of loess landforms using Random Forest method

The automatic recognition of landforms is regarded as one of the most important procedures to classify landforms and deepen the understanding on the morphology of the earth. However, landform types are rather complex and gradual changes often occur in these landforms, thus increasing the difficulty in automatically recognizing and classifying landforms. In this study, small-scale watersheds, which are regarded as natural geomorphological elements, were extracted and selected as basic analysis and recognition units based on the data of SRTM DEM. In addition, datasets integrated with terrain derivatives(e.g., average slope gradient, and elevation range) and texture derivatives(e.g., slope gradient contrast and elevation variance) were constructed to quantify the topographical characteristics of watersheds. Finally, Random Forest(RF) method was employed to automatically select features and classify landforms based on their topographical characteristics. The proposed method was applied and validated in seven case areas in the Northern Shaanxi Loess Plateau for its complex andgradual changed landforms. Experimental results show that the highest recognition accuracy based on the selected derivations is 92.06%. During the recognition procedure, the contributions of terrain derivations were higher than that of texture derivations within selected derivative datasets. Loess terrace and loess mid-mountain obtained the highest accuracy among the seven typical loess landforms. However, the recognition precision of loess hill, loess hill–ridge, and loess sloping ridge is relatively low. The experiment also shows that watershed-based strategy could achieve better results than object-based strategy, and the method of RF could effectively extract and recognize the feature of landforms.     

基于随机森林的黄土地貌分类研究

地貌分类在指导人类建设活动的规模与布局中有着重要的意义。然而,传统的基于数字高程模型(DEM)的地貌分类方法使用的地形因子和考虑到的地貌特征往往比较单一。本文提出了一种基于流域单元的地貌分类方法,该方法考虑了流域单元的多方面特征,包括基本地形因子统计量、地形特征点线统计量、小流域特征和纹理特征。本研究首先基于DEM进行水文分析将研究区域划分成不同的小流域。然后利用数字地形分析提取29个不同方面的特征来表征流域的形态,并基于随机森林(RF)算法进行了特征选择和参数标定。RF是一种基于决策树算法的集成分类器,能有效地处理高维数据,分类精度高。最后选择训练集小流域对RF分类器进行训练,使用训练完成的分类器对整个研究区域的地貌进行分类,研究地貌分异的规律。该实验在我国陕北黄土高原典型黄土地貌区域的地貌分类中取得了较好的结果,结果表明不同的地貌之间存在明显的区域界线,特定的地貌类型在空间上表现出明显的聚集性。通过人工判读进行验证的分类精度达到了85%,Kappa系数为0.83。     

A terrain openness index for the extraction of karst Fenglin and Fengcong landform units from DEMs

The Fenglin and Fengcong landform units are considered to be an important representation for defining the degree of development of Karst landforms. However, these terrain features have been proven difficult to delineate and extract automatically because of their complex morphology. In this paper, a new method for identifying the Fenglin and Fengcong landform units is proposed. This method consists of two steps: (1) terrain openness calculation and (2) toe line extraction. The proposed method is applied and validated in the Karst case area of Guilin by using ASTER GDEM with one arc-second resolution. The openness of both the positive and negative terrain and a threshold were used to extract toe lines for segmenting depressions and pinnacles in Fenglin and Fengcong landforms. A comparison between the extracted Fenglin and Fengcong landform units and their real units from high resolution images was carried out to evaluate the capability of the proposed method. Results show the proposed method can effectively extract the Fenglin and Fengcong landform units, and has an overall accuracy of 93.28%. The proposed method is simple and easy to implement and is expected to play an important role in the automatic extraction of similar landform units in the Karst area.     

Quantifying spatial scale of positive and negative terrains pattern at watershed-scale: Case in soil and water conservation region on Loess Plateau

The positive and negative terrains (P-N terrains) widely distributed across China’s Loess Plateau constitute the dual structure characteristic of loess landforms. Analysis of loess P-N terrains at the watershed scale can serve to elucidate the structural characteristics and spatial patterns of P-N terrains, which benefits a better understanding of watershed evolution and suitable scales for loess landform research. The Two-Term Local Quadrat Variance Analysis (TTLQV) is calculated as the average of the square of the difference between the block totals of all possible adjacent pairs of block size, which can be used to detect both the scale and the intensity of landscape patches (e.g., plant/animal communities and gully networks). In this study, we determined the latitudinal and longitudinal spatial scale of P-N terrain patterns within 104 uniformly distributed watersheds in our target soil and water conservation region. The results showed that TTLQV is very effective for examining the scale of P-N terrain patterns. There were apparently three types of P-N terrain pattern in latitudinal direction (i.e., Loess Tableland type, Loess Hill type, and Transitional Form between Sand and Loess type), whereas there were both lower and higher values for P-N terrain pattern scales in all loess landforms in the longitudinal direction. The P-N terrain pattern also clearly presented anisotropy, suggesting that gully networks in the main direction were well-developed while others were relatively undeveloped. In addition, the relationships between the first scales and controlling factors (i.e., gully density, nibble degree, watershed area, mean watershed slope, NDVI, precipitation, loess thickness, and loess landforms) revealed that the first scales are primarily controlled by watershed area and loess landforms. This may indicate that the current spatial pattern of P-N terrains is characterized by internal force. In selecting suitable study areas in China’ Loess Plateau, it is crucial to understand four control variables: the spatial scale of the P-N terrain pattern, the watershed area, the main direction of the watershed, and the loess landforms.