混合核特征加权SVM遥感湿地空间分类

针对湿地空间信息的复杂性和SVM的分类性能,设计一种基于混合核函数的特征加权SVM分类模型,综合利用多种特征信息,避免被弱相关特征所支配,从而提供更佳的映射性能和泛化能力。实验结果表明,该分类模型兼具良好的外推和内推能力,能够有效地融合不同信息源特征,得到更完整和准确的分类结果,在总体精度、Kappa系数等多项指标上都表现出更高的水平。     

AAC算法的大气校正复合改进算法

利用模拟数据,评价Autonomous Atmospheric Compensation(AAC)算法的抗噪性,认为AAC算法的抗噪性较弱。基于TASI实测数据,利用AAC算法开展反演计算时,计算结果呈现出多样性问题。结合In-scene Atmospheric Compensation(ISAC)算法中黑体像元的标定方法,提出了一种复合改进算法。首先,利用ISAC算法反演的大气透过率和路径辐射,重新计算AAC算法中大气透过率之比(Tr)和相邻两强弱吸收通道的路径辐射之差(Pd),再次,运用经验公式获得稳定的大气反演结果(大气透过率和路径辐射),有效解决了计算结果多样性的问题。利用复合改进算法,开展的温度与发射率分离实验,证明反演得到的发射率波谱更接近野外实测波谱。     

时变参数灰色模型背景值重构及实现

针对时变参数灰色模型PGM(1,1)的背景值重构收敛速度及稳定性问题,该文运用积分的方法综合序列在Δt内不同变化趋势,导出背景值模型的准确表达式。实现了反映序列对新老信息偏爱程度的最优权值介于(0,1)内,且非不能越过某一阈值;给出了背景值重构模型最优解准确求取的具体算法步骤。基于MATLAB语言的实验结果表明:改进模型预测精度高,易于实现;所研究的带权灰色模型GM(1,n)背景值模型的重构及计算方法验证了PGM(1,1)模型重构及计算实现方法的有效性和实用性。     

三重加权变形监测预测模型及应用

针对GM(1,1)建模过程存在背景值、时间因素和初始条件3方面的不足,该文提出三重加权TPGM(1,1)预测模型。通过对背景值进行加权生成新的背景值,建立PGM(1,1)模型;在PGM(1,1)基础上考虑到时间因素,在求解灰参数时进行第2次加权建立DPGM(1,1)模型;最后考虑到初始条件对预测模型的影响,在DPGM(1,1)基础上进行第3次加权,建立TPGM(1,1)模型。通过实例分析,比较GM(1,1)、PGM(1,1)、DPGM(1,1)、TPGM(1,1)4种模型在变形监测数据处理中的拟合和预测结果,表明三重加权TPGM(1,1)模型拟合效果更好、预测精度更高;该模型具有前3种模型的优点,同时弥补了传统GM(1,1)存在的不足。     

局部均值分解与支持向量回归的大坝变形预测

针对现有变形预测方法对于大坝变形的预测效果不理想的问题,该文利用局部均值分解方法获取生产函数分量并进行支持向量回归建模,用此方法对大坝变形进行多尺度分析。通过局部均值分解对大坝变形序列进行分解得到其乘积函数分量,然后利用支持向量机回归进行外推预测,再把各乘积函数分量的预测结果进行叠加重构生成,进而获得大坝变形预测值。通过实例分析,比较GM(1,1)、支持向量机和该文方法3种模型在变形监测数据处理中的拟合和预测结果,表明该文方法充分发掘数据本身所蕴含的物理机制和物理规律,提高了大坝变形多尺度预测精度。     

改进支持向量机的高分遥感影像道路提取

针对支持向量机受分类数的限制在高分辨率遥感影像中无法直接获取高精度道路网信息的问题,该文提出一种新的混合的基于支持向量机的方法:首先,利用模糊C均值聚类方法将输入的遥感影像分为3类,以减少支持向量机的错分现象;其次,运用支持向量机将不同类别的像素分为道路类和非道路类;最后,应用马尔科夫随机场对分类结果进行噪声去除,并采用形态学进行后处理,进而得到精确道路网信息。实验结果表明:该算法不仅能够从高分辨率遥感影像中提取出道路网,而且精度优于直接使用支持向量机算法以及对比算法。     

A support vector machine classifier based on a new kernel function model for hyperspectral data

The kernel function is a key factor to determine the performance of a support vector machine (SVM) classifier. Choosing and constructing appropriate kernel function models has been a hot topic in SVM studies. But so far, its implementation can only rely on the experience and the specific sample characteristics without a unified pattern. Thus, this article explored the related theories and research findings of kernel functions, analyzed the classification characteristics of EO-1 Hyperion hyperspectral imagery, and combined a polynomial kernel function with a radial basis kernel function to form a new kernel function model (PRBF). Then, a hyperspectral remote sensing imagery classifier was constructed based on the PRBF model, and a genetic algorithm (GA) was used to optimize the SVM parameters. On the basis of theoretical analysis, this article completed object classification experiments on the Hyperion hyperspectral imagery of experimental areas and verified the high classification accuracy of the model. The experimental results show that the effect of hyperspectral image classification based on this PRBF model is apparently better than the model established by a single global or local kernel function and thus can greatly improve the accuracy of object identification and classification. The highest overall classification accuracy and kappa coefficient reached 93.246% and 0.907, respectively, in all experiments.     

Temporal replication of the national land cover database using active machine learning

Regional and national level land cover datasets, such as the National Land Cover Database (NLCD) in the United States, have become an important resource in physical and social science research. Updates to the NLCD have been conducted every 5 years since 2001; however, the procedure for producing a new release is labor-intensive and time-consuming, taking 3 or 4 years to complete. Furthermore, in most countries very few, if any, such releases exist, and thus there is high demand for efficient production of land cover data at different points in time. In this paper, an active machine learning framework for temporal updating (or backcasting) of land cover data is proposed and tested for three study sites covered by the NLCD. The approach employs a maximum entropy classifier to extract information from one Landsat image using the NLCD, and then replicate the classification on a Landsat image for the same geographic extent from a different point in time to create land cover data of similar quality. Results show that this framework can effectively replicate the land cover database in the temporal domain with similar levels of overall and within class agreement when compared against high resolution reference land cover datasets. These results demonstrate that the land cover information encapsulated in the NLCD can effectively be extracted using solely Landsat imagery for replication purposes. The algorithm is fully automated and scalable for applications at landscape and regional scales for multiple points in time.     

Remote sensing and information value (IV) model for regional mapping of fluvial channels and topographic wetness in the Saudi Arabia

This study aims to map regions of near surface fluvial channels, mega-basins and topographic wetness in Saudi Arabia using remote sensing data and an information value (IV) model, which is a modified approach of weight of evidence. We used the new version of the Shuttle Radar Topographic Mission (SRTM) to delineate the fluvial channels, mega-basin, and slope. These hydrological parameters were used to index the topographic wetness of each mega-basin in the region based on IV in a Geographic Information System. We validated our method using the Space Imaging Radar-C and Landsat 8 images and compared the textural features (fluvial channels) evident from SRTM digital elevation model and to determine whether these patterns were different. Our results revealed that the region is drained by nine tributaries and that the Err Rub Al Khali and Sahba mega-basins have the highest value of the IV and topographic wetness values; the Arran and coastal mega-basins have the lowest value of the IV and topographic wetness values. An integrated approach is timely and economically effective and can be applied throughout the arid and semi-arid regions to help hydrologists and urban developers.     

An efficient application of fusion approach for hot spot detection with MODIS and PALSAR-1 data

Hot spot detection with satellite images, especially with synthetic aperture radar (SAR) images is still a challenging task. Several researchers have used TM/optical data for identification of hot spot but the use of SAR data is very limited for this type of application. The fusion of SAR data with TM/optical data may add additional information which in turn will lead for enhancement of detection capability of the hot spot. Therefore, this study explores the possibility of fusion of Moderate Resolution Imaging Spectroradiometer (MODIS) and Phased Array L-band Synthetic Aperture Radar (PALSAR) satellite images for the hot spot detection. Image fusion is emerging as a powerful tool where information of various sensors can be used for obtaining better results. For this purpose, vegetation greenness and roughness information which is obtained from MODIS and PALSAR satellite images, respectively, are used for fusion, and then, a contextual-based thresholding algorithm is applied to the fused image for hot spot detection. The proposed approach comprises of two steps: (1) application of genetic algorithm-based scheme for image fusion of MODIS and PALSAR satellite images, and (2) classification of the fused image as either hot spot or non-hot spot pixels by employing a contextual thresholding technique. The algorithm is tested over the Jharia Coal Field region of India, where hot spot is one of the major problems and it is observed that the proposed thresholding technique classifies the each pixel of the fused image into two categories: hot spot and non-hot spot and the proposed approach detects the hot spot with better accuracy and less false alarm.