The Main Ethiopian Rift(MER)is an area of extreme topography underlain by post-Miocene volcanic rocks,Jurassic limestone and a Precambrian basement.A prime concern is the rapid expansion of wide gullies that are impin...The Main Ethiopian Rift(MER)is an area of extreme topography underlain by post-Miocene volcanic rocks,Jurassic limestone and a Precambrian basement.A prime concern is the rapid expansion of wide gullies that are impinging on agricultural land.We investigate the potential contribution of Advanced Space-borne Thermal Emission and Reflection Radiometer(ASTER)data and geomorphologic parameters to discern patterns and features of gully erosion in the MER.Maximum Likelihood Classifica-tion(MLC),Support Vector Machine(SVM),and Minimum Distance(MD)classifiers are used to extract different gully shapes and patterns.Several spatial textures based on Grey Level Co-occurrence Matrices(GLCMs)are then generated.Afterwards,the same classifiers are applied to the ASTER data combined with the spatial texture information.We used geomorphologic parameters ex-tracted from SRTM and ASTER DEMs to describe the geomorphologic setting and the gullies' shapes.The classifications show accuracies varying between 67% and 89%.Maps derived from this quantitative analysis allow the monitoring and mapping of land degradation as a direct result of gully-widening.This study reveals the utility of combining ASTER data and spatial textural infor-mation in discerning areas affected by gully erosion.展开更多
针对高光谱图像中光谱信息提取时高维特征向量由于部分邻域叠加造成数据缺损,以及图像局部区域像素点在空间结构信息中存在同谱异类现象和密度差异的问题,提出了一种基于空谱超像素融合核极限学习机(SSKELM)的高光谱图像分类算法.对光...针对高光谱图像中光谱信息提取时高维特征向量由于部分邻域叠加造成数据缺损,以及图像局部区域像素点在空间结构信息中存在同谱异类现象和密度差异的问题,提出了一种基于空谱超像素融合核极限学习机(SSKELM)的高光谱图像分类算法.对光谱空间第一主成分分量进行超像素分割,每个超像素被看作一个形状自适应区域。利用空间信息、超像素内及像元间的核权重融合,获取像素点类别标签;同时,借助核函数在高维超平面数据中线性可分能力、极限学习机随机隐藏层输出矩阵及其优化算法的限制条件少等优势,将空谱像素点融合训练并形成新的矩阵样本输出.使用University of Pavia和Indian Pines两个数据集进行实验,总体准确率OA值较其他算法分别提高了1.76%和2.80%,有效验证本文提出方法在图像分类中具有一定价值.展开更多
基金Supported by the German Academic Exchange Service
文摘The Main Ethiopian Rift(MER)is an area of extreme topography underlain by post-Miocene volcanic rocks,Jurassic limestone and a Precambrian basement.A prime concern is the rapid expansion of wide gullies that are impinging on agricultural land.We investigate the potential contribution of Advanced Space-borne Thermal Emission and Reflection Radiometer(ASTER)data and geomorphologic parameters to discern patterns and features of gully erosion in the MER.Maximum Likelihood Classifica-tion(MLC),Support Vector Machine(SVM),and Minimum Distance(MD)classifiers are used to extract different gully shapes and patterns.Several spatial textures based on Grey Level Co-occurrence Matrices(GLCMs)are then generated.Afterwards,the same classifiers are applied to the ASTER data combined with the spatial texture information.We used geomorphologic parameters ex-tracted from SRTM and ASTER DEMs to describe the geomorphologic setting and the gullies' shapes.The classifications show accuracies varying between 67% and 89%.Maps derived from this quantitative analysis allow the monitoring and mapping of land degradation as a direct result of gully-widening.This study reveals the utility of combining ASTER data and spatial textural infor-mation in discerning areas affected by gully erosion.
文摘针对高光谱图像中光谱信息提取时高维特征向量由于部分邻域叠加造成数据缺损,以及图像局部区域像素点在空间结构信息中存在同谱异类现象和密度差异的问题,提出了一种基于空谱超像素融合核极限学习机(SSKELM)的高光谱图像分类算法.对光谱空间第一主成分分量进行超像素分割,每个超像素被看作一个形状自适应区域。利用空间信息、超像素内及像元间的核权重融合,获取像素点类别标签;同时,借助核函数在高维超平面数据中线性可分能力、极限学习机随机隐藏层输出矩阵及其优化算法的限制条件少等优势,将空谱像素点融合训练并形成新的矩阵样本输出.使用University of Pavia和Indian Pines两个数据集进行实验,总体准确率OA值较其他算法分别提高了1.76%和2.80%,有效验证本文提出方法在图像分类中具有一定价值.
