高分辨率遥感影像多尺度均值漂移分割算法研究

根据高分辨率遥感影像信息提取过程中对影像的对象化分割的需求,分析了均值漂移分割算法的原理,并对其多尺度分割方法进行了设计与实现。实验证明该算法具有较好的影像分割精度。     

基于LiDAR数据和航空影像的水体自动提取

提出将LiDAR数据对水体的敏感性与航空影像的高分辨率特征相结合的水体自动提取方法。利用SIFT算法对LiDAR强度图像和航空影像进行配准,在LiDAR高程图像上提取无回波信号的黑色区域,构建几何约束条件,排除由遮挡产生的无效区域;将水体初始位置映射到航空影像上,结合边缘信息进行区域生长,并对生长区域进行数学形态学运算,最终获取水体区域。实验结果表明,该方法可以获得很好的水体提取效果。     

Bound and mobile soil water isotope ratios are affected by soil texture and mineralogy,whereas extraction method influences their measurement

Questions persist about interpreting isotope ratios of bound and mobile soil water pools, particularly relative to clay content and extraction conditions. Interactions between pools and resulting extracted water isotope composition are presumably related to soil texture, yet few studies have manipulated the bound pool to understand its influence on soil water processes. Using a series of drying and spiking experiments, we effectively labelled bound and mobile water pools in soils with varying clay content. Soils were first vacuum dried to remove residual water, which was then replaced with heavy isotope-enriched water prior to oven drying and spiking with heavy isotope-depleted water. Water was extracted via centrifugation or cryogenic vacuum distillation (at four temperatures) and analysed for oxygen and hydrogen isotope ratios via isotope ratio mass spectrometry. Water from centrifuged samples fell along a mixing line between the two added waters but was more enriched in heavy isotopes than the depleted label, demonstrating that despite oven drying, a residual pool remains and mixes with the mobile water. Soils with higher clay + silt content appeared to have a larger bound pool. Water from vacuum distillation samples have a significant temperature effect, with high temperature extractions yielding progressively more heavy isotope-enriched values, suggesting that Rayleigh fractionation occurred at low temperatures in the vacuum line. By distinctly labelling bound and mobile soil water pools, we detected interactions between the two that were dependent on soil texture. Although neither extraction method appeared to completely extract the combined bound and mobile (total water) pool, centrifugation and high temperature cryogenic vacuum distillations were comparable for both δ²H and δ¹⁸O of soil water isotope ratios.     

基于决策树分类的大屯矿区地物信息提取及矿区污染分析

本文在对国内外遥感图像分类方法充分研究分析的基础上,选择决策树分类法对大屯矿区的Landsat 8遥感图像进行分类研究。选取样本提取并分析研究区典型地类光谱特征曲线,依据光谱曲线特征和归一化植被指数建立了土地利用分类决策树模型,通过反复试验和修正,筛选出适宜大屯矿区地物分类的决策树最优阈值,对研究区进行分类和精度评价,最后通过分类结果对研究区的水体污染状况进行简要分析。     

基于光谱角匹配算法的水体信息提取研究

针对水体信息提取时,传统算法不能较好地解决山体阴影被误提为水体的问题,引入多用于高光谱数据处理的光谱角匹配算法,以Land-sat8 OLI(band1-band7)为数据源,以黄河小浪底水库周边区域为实验区,开展水体信息提取研究,结果显示:水体指数法和波段关系提取的结果中有大量的山体阴影信息,提取精度不超过30%,而光谱角匹配法提取结果受阴影的影响较小,提取精度在99%以上。     

基于CIPS软件探讨高山地DOM数据的生产方式

在DOM数据生产中,山地、高山地数据一直以来都是重点和难点。本文分析了高山地影像数据的特点,提出一种与山地、高山地数据相适应的处理方式。最后基于CIPS软件,按需求设计对应的软件处理流程,并使用满足实验要求的高山地影像数据,对这种方式加以实现和验证。     

Geochemical fractionation of Ni,Cu and Pb in the deep sea sediments from the Central Indian Ocean Basin: An insight into the mechanism of metal enrichment in sediment

Metal speciation study in combination with major element chemistry of deep sea sediments provided possible metal enrichment pathways in sediments collected from environmentally different locations of Central Indian Ocean Basin (CIB). Metal speciation study suggests that Fe–Mn oxyhydroxide phase was the major binding phase for Ni, Cu and Pb in the sediments. The second highest concentrations of all these metals were present within the structure of the sediments. Easily reducible oxide phase (within the Fe–Mn oxyhydroxide binding phases) was the major host for all the three metals in the studied sediments. Major element chemistry of these sediments revealed that there was an increased tendency of Cu and Ni to get incorporated into the deep sea sediment via the non-terrigenous Mn-oxyhydroxide fraction, whereas, Pb gets incorporated mostly via amorphous Fe-hydroxides into the sediment from the CIB. This is the first attempt to provide an insight into the mechanism of metal enrichment in sediment that host vast manganese nodule.     

开放式地理实体关系抽取的Bootstrapping方法

从网络文本中抽取地理实体间空间关系和语义关系要求高时效性和强鲁棒性。本文提出一种开放式地理实体关系的自动抽取方法,通过bootstrapping技术统计词语的词性、位置和距离特征来计算语境中词语权值,据此确定描述地理实体关系的关键词,最终组织成结构化实例,并使用百度百科和Stanford CoreNLP开展了试验。研究结果表明,本文方法能自动挖掘自然语言的部分词法特征,无须领域专家知识和大规模标注语料,适用于未知关系类型的信息抽取任务;较之经典的Frequency、TFIDF和PPMI频率统计方法,精度和召回率分别提升约5%和23%。     

一种四叉树和测地线活动轮廓模型相结合的海陆影像分割方法

海陆影像分割对于后续的海岸线提取、潮间带地形反演、海岸演化状况分析等都具有十分重要的意义。本文在分析了四叉树、测地线活动轮廓(GAC)模型和Canny边缘检测算子等在海陆影像分割中优缺点的基础上,提出了一种四叉树、Canny算子和GAC模型相结合的海陆影像分割方法。该方法综合利用上述各方法的优点,将Canny算子边缘检测结果融入到基于四叉树初分割的GAC模型中,重构边界停止函数,演化水平集方程,实现海陆影像分割。试验结果表明,该方法具有海陆影像分割速度快、精度高、可靠性强和自动化程度高等优点,对于弱边缘以及严重凹陷边缘,都能实现自动和准确分割。     

一种新的窗户特征提取方法研究

在建筑物立面重建中,窗户提取、结构识别是学者们研究的热点。针对建筑物特征提取的应用需求,文中提出一种基于局部搜索算法——爬山法的窗户信息提取方法,介绍研究中采用的一些原理、方法,从而可以准确有效地从激光扫描数据中提取出建筑物的窗户信息。利用本方案对地面激光扫描系统获取的建筑物实验数据进行具体的处理,并给出可视化表达结果。