总体最小二乘法在相机标定中的应用

对比总体最小二乘方法与最小二乘方法在相机标定中的适用性及优越性。在相机标定中,由于像点坐标和对应的地面点坐标均存在误差,因此采用总体最小二乘方法对误差方程中的系数矩阵及观测向量同时改正,能够建立更加合理的计算模型。文中以相机标定两步法为例,通过实例解算,证明利用总体最小二乘法能够得到精度更高的相机标定参数解。     

GPS 高程转换的总体最小二乘组合方法

在现有GPS高程转换的研究中,通常只采用单一的解算方法,然而,不同的计算方法由于适用的条件不同,在对同一组数据进行解算时,解算的精度也各有优劣。针对以上问题,文中提出GPS高程转换的总体最小二乘组合方法,对多种不同的高程转换方法进行组合,运用总体最小二乘法计算不同方法的权值,结合实际工程数据,验证该方法的可行性。计算结果表明,提出的方法能够有效地提高高程转换精度。     

改进的距离约束最小二乘模糊度搜索算法

针对短基线最小二乘模糊度搜索算法搜索效率低的缺点,该文提出了一种改善模糊度搜索空间以提高模糊度搜索效率的方法。最小二乘搜索算法通过基线长度范围确定模糊度搜索空间,完成模糊度的搜索和确认。距离约束的短基线模糊度搜索空间是一个空心椭球,各模糊度之间具有相关性,影响模糊度搜索效率。该文通过最小二乘去相关调整方法降低模糊度之间的相关性,减少搜索次数。实验证明,该算法的搜索效率提高了25%以上,且随着基线长度的增加,搜索效率有更多的提升。     

一种裸露土壤湿度反演方法

针对目前土壤湿度反演方法研究较少且缺少实时性的现状,该文提出一种土壤湿度反演方法——最小二乘支持向量机技术。以积分方程模型为正向算法,数值模拟不同雷达参数(频率、入射角及极化)下后向散射系数随土壤含水量和地表粗糙度的变化情况。经过数据敏感性分析,选取C-波段和X-波段、小入射角下的同极化后向散射系数作为支持向量回归的训练样本信息;经过适当的训练,利用支持向量回归技术对土壤含水量进行了反演研究;并考虑通过多频率、多极化、多入射角数据的组合,消除地表粗糙度的影响,提高反演精度。模拟结果表明,该方法反演土壤湿度具有较高的精度和较好的实时性;同时,与人工神经网络方法的结果比较,证明了该方法的有效性,为土壤湿度的反演研究提供了一种方法。     

基于优化经验模态分解和聚类分析的滑坡位移智能预测研究

针对三峡库区"阶跃式"滑坡的变形特征,提出了一种新的滑坡位移预测方法。以白水河滑坡ZG118和XD-01监测点位移数据为例,采用基于软筛分停止准则的经验模态分解(SSSC-EMD)将累计位移-时间曲线和影响因子时间序列自适应地分解为多个固有模态函数(IMF),并采用K均值(K-Means)聚类法对其进行聚类累加,得到有物理含义的位移分量(趋势性位移、周期性位移以及随机性位移)和影响因子分量(高频影响因子和低频影响因子)。使用最小二乘法对趋势性位移进行拟合预测;采用果蝇优化-最小二乘支持向量机(FOA-LSSVM)模型对周期性位移和随机性位移进行预测。将各位移分量预测值进行叠加处理,实现滑坡累计位移的预测。研究结果表明,所提出的(SSSC-EMD)-K-Means-(FOA-LSSVM)模型能够预测"阶跃式"滑坡的位移变化规律,且预测精度高于传统的支持向量机回归(SVR)、最小二乘支持向量机(LSSVM)模型;并通过改变训练集长度,进行单因素分析,发现其与预测精度之间呈正相关关系。     

地震预警现地PGV连续预测的最小二乘支持向量机模型

为提升现地仪器地震烈度预测的准确性与连续性,研究面向地震预警的PGV连续预测模型.以中国仪器地震烈度标准的计算参数:0.1～10 Hz带通滤波三分向矢量合成速度峰值PGV为预测目标,利用日本K-net与KiK-net台网P波触发后1?10 s强震数据,基于人工智能中的机器学习方法-最小二乘支持向量机,选取7种特征参数作...     

Mapping of moraine dammed glacial lakes and assessment of their areal changes in the central and eastern Himalayas using satellite data

The relatively rapid recession of glaciers in the Himalayas and formation of moraine dammed glacial lakes(MDGLs) in the recent past have increased the risk of glacier lake outburst floods(GLOF) in the countries of Nepal and Bhutan and in the mountainous territory of Sikkim in India. As a product of climate change and global warming, such a risk has not only raised the level of threats to the habitation and infrastructure of the region, but has also contributed to the worsening of the balance of the unique ecosystem that exists in this domain that sustains several of the highest mountain peaks of the world. This study attempts to present an up to date mapping of the MDGLs in the central and eastern Himalayan regions using remote sensing data, with an objective to analyse their surface area variations with time from 1990 through 2015, disaggregated over six episodes. The study also includes the evaluation for susceptibility of MDGLs to GLOF with the least criteria decision analysis(LCDA). Forty two major MDGLs, each having a lake surface area greater than 0.2 km2, that were identified in the Himalayan ranges of Nepal, Bhutan, and Sikkim, have been categorized according to their surface area expansion rates in space and time. The lakes have been identified as located within the elevation range of 3800 m and6800 m above mean sea level(a msl). With a total surface area of 37.9 km2, these MDGLs as a whole were observed to have expanded by an astonishing 43.6% in area over the 25 year period of this study. A factor is introduced to numerically sort the lakes in terms of their relative yearly expansion rates, based on their interpretation of their surface area extents from satellite imageries. Verification of predicted GLOF events in the past using this factor with the limited field data as reported in literature indicates that the present analysis may be considered a sufficiently reliable and rapid technique for assessing the potential bursting susceptibility of the MDGLs. The analysis also indicates that, as of now, there are eight MDGLs in the region which appear to be in highly vulnerable states and have high chances in causing potential GLOF events anytime in the recent future.     

利用加权整体最小二乘DSM提取DEM的方法

随着无人机航摄软硬件技术的快速发展,通过倾斜摄影三维建模快速获取DSM的成本越来越低。在小比例尺的地形图测量中,DSM若能通过快速处理得到较为精确的DEM,即可节省大量的人力物力。针对林地范围内DSM数据的特点,采用加权整体最小二乘法对DEM进行提取。     

融合罗德里格矩阵和整体最小二乘的双目机器人手眼标定算法

针对传统双目式机器人手眼标定方法中未知量多、计算复杂的特点,研究并提出了一种基于罗德里格矩阵和整体最小二乘相融合的改进算法。其中罗德里格矩阵实现了对未知量的降维;整体最小二乘顾及了系数矩阵的误差。通过大量数据分析表明,改进后的手眼标定算法较传统方法在稳定性和精度上均有大幅提高,满足了机器人视觉系统在工业检测和制造领域中的高精度需求。