基于MRF随机场和广义混合模型的遥感图像分级聚类

有限混合模型FM的分级聚类已广泛应用于不同领域,然而,它的计算复杂度与观测数据的平方成正比,因此,在海量数据方面的应用就受到了限制。另一方面,多光谱图像数据中同时包含有空间和光谱两类信息,但大多数基于像素的多光谱图像聚类方法,仅使用了其频谱信息而忽视了空间信息。本文提出了一种新的基于广义有限混合模型GFM的分级聚类方法,该算法把MRF随机场和GFM模型结合在一起,分类数可以通过PLIC准则自动确定。算法在执行过程中,采用K均值聚类方式获得过分类图像,分级聚类从过分类图像开始,代替原来从单点类开始的方式,这样可以方便获取GFM模型成分密度的初始参数。最后,采用由Gibbs采样器生成的仿真测试图对算法的精度进行了定量评价,通过与K均值聚类和FM聚类的比较说明了本文算法的优越性,同时用荷兰Flevoland农业地区的极化SAR图像验证了本文算法的有效性。     

混合像元线性分解的精度估算

遥感图像中普遍存在着混合像元,对混合像元进行分解是遥感图像处理中的难点。混合像元线性分解技术是进行高光谱影像处理的常用方法。本文针对混合像元线性分解的基本原理与算法作了简要分析,并通过实验的方法估算了混合像元线性分解模型的精度。     

基于高斯混合模型的建筑物QuickBird多光谱影像数据分类研究

针对任何一种遥感影像数据的信息提取都有其无法逾越的理论极限,正确认识这种极限有利于明确相关算法研究的方向,降低工程应用成本.制约影像信息提取精度的"同物异谱"现象以及与之相关的影像对象"光谱异质性"问题正是科学认识这种极限的关键和切入点.城市下垫面中的建筑物屋顶材料不同,光谱反射率也不同,"同物异谱"现象严重.基于高斯混合模型的期望最大(Expectation Maximization,EM)估计算法,能为分析建筑物类内以及同一建筑物对象内部光谱异质性程度提供科学依据, 进而提高分类精度.本文以QuickBird多光谱影像为实证研究数据, 利用高斯混合模型及其EM估计算法拟合出不同材料屋顶的密度分布, 实现建筑物影像对象分类, 得到优于传统监督、非监督分类的结果.     

基于超像素分割与CRF的高分辨率遥感影像变化检测

针对目前高分辨率遥感影像变化检测算法对于光谱变化过敏感问题,本文提出了一种基于超像素分割与条件随机场（CRF）的遥感影像变化检测算法。首先采用空间约束的t混合模型驱动的分割模型,获得同质性超像素块,实现良好的边界附着性和亮度均匀性。然后计算分割得到的双时相影像块之间的特征差异性,获取变化幅度图像。最后利用模糊聚类算法（FCM）对变化幅度图像进行聚类,得到隶属度图像作为CRF一阶势,并利用光谱-空间相似度约束的函数构建CRF二阶势。试验结果表明,与现有方法相比,该方法检测精度可提高5%,错检率和漏检率可降低3%,能较好地应对输入图像的光谱变化,并保持变化检测结果的边缘细节。     

遥感云分类不确定性的多维混合熵模型评价

针对遥感影像提取信息分类过程中存在随机不确定性和模糊不确定性两种噪声,影响分类结果的准确性问题,该文在多光谱遥感影像处理中,通过对传统的混合熵模型进行多维化改进,提出多维混合熵的不确定性评价模型。采用云算法对遥感影像进行解译分类,获取相应的不确定性模型参数计算出信息熵和模糊熵,从像元和类别两个尺度构建出遥感云分类不确定性的多维混合熵评价模型。结果表明,多维混合熵模型能够充分考虑多光谱遥感数据的多维性,可以从不同尺度对遥感分类的随机不确定性和模糊不确定性进行有效全面地评价。     

基于马尔可夫随机场的大面积遥感影像聚类初值化问题

马尔可夫随机场聚类模型,是利用混合像元所描述物体的时间和空间上的信息,来提高遥感影像的分类精度,例如多光谱偏振合成孔径雷达(SAR)影像。然而,这种影像的分类太过于依赖初始条件的选择,如分类种数和种类参数的选择。本文将会介绍一种基于MRF聚类模型的遥感影像分类的初值化方法。其适当的初始聚类参数来自一系列相似的区域,种类数目的估计由PLIC得到。适用于由许多大面积相似的区域组成的影像,例如农业收获区域。实验采用荷兰Flevoland地区的偏振SAR影像,结果表明比其它方法如模糊C-means和ICM聚类算法好。     

基于高光谱Hyperion数据的线性光谱混合模型与神经网络模型的比较

混合像元问题是定量遥感中的热点问题之一,为了改进从遥感数据中提取定量信息,人们建立了各种混合光谱分解技术,其中线性光谱混合模型和神经网络模型就是两种比较成熟的方法。以陕西省横山地区的高光谱Hyperion数据为研究基础,通过最小噪声变换(MNF)、像元纯度指数(PPI)转换和RMS误差分析的迭代方法相结合提取影像中的纯净像元作为终端端元。分别运用神经网络模型和线性光谱混合模型对影像进行光谱分解,得到各个组分的分解图像。以标准植被指数(NDVI)影像为衡量标准,选取训练样本点,分别对两种模型进行回归分析,结果显示NDVI影像与线性光谱混合模型植被分解图像的判定系数(R2=0.91)要大于其与神经网络模型的判定系数(R2=0.81)。进一步分析表明在一般情况下,线性光谱混合模型具有比神经网络模型略高的分离精度,但是神经网络模型对细部信息的提取的效果要好于线性光谱混合模型,最后提出了端元均方根误差(EAR)指数,一种新的混合像元分解的思路。     

一种高光谱遥感图像快速谱聚类算法

为获得分类效果更优良的遥感图像分类方式并解决高光谱遥感图像分类运算速度缓慢的问题，集成Lanczos算法与谱聚类算法，探讨了高光谱遥感图像谱聚类算法应用于遥感图像分类的可行性，提出了一种面向高光谱遥感图像的快速谱聚类算法；通过对比美国圣地亚哥机场高光谱遥感图像K-均值算法与谱聚类算法的分类结果，发现面向高光谱遥感图像的谱聚类算法易于识别线性地物，且分类的速度能得到较大提升。     

一种云模型和期望最大聚类的遥感影像分割算法

为解决遥感影像分割中存在的不确定性问题和传统层次聚类算法中存在的时间复杂度高、缺乏可再分性等缺陷,基于云模型和期望最大聚类提出了一种新的遥感影像分割算法。该算法首先使用峰值法云变换从影像中抽取底层概念,然后通过EM算法对底层概念进行聚类,最后通过极大判别法完成遥感影像分割。实验证明,EM算法进行概念聚类能够快速地将概念分类为指定个数,并估计出高阶云概念的数学特征,相比于传统的基于云模型的遥感影像分割算法具有更好的分割效果。     

基于独立分量分析的遥感影像非监督分类方法

利用独立分量分析的方法，从图像信号分离的角度出发，将每个波段像元的光谱特征看成是由相互独立的不同地物类型光谱信号混合而成。通过ETM^-遥感影像数据的分类试验，验证了该方法应用于多光谱遥感影像非监督分类的有效性。     

A New Semi-supervised Classification Method Based on Mixture Model Clustering for Classification of Multispectral Data

A new method for semi-supervised classification of remotely-sensed multispectral image data is developed in this study. It consists of unsupervised-clustering for data labelling and supervised-classification of clusters in multispectral image data (MID) using spectral signatures. Mixture model clustering, based on model selection, is proposed for finding the number and determining the structures of clusters in MID. The best mixture model, for the best clustering of data, finds the number and determines the structure of clusters in MID. The number of elements in the best mixture model fits to the number of clusters in MID. The elements of the best mixture model fits to the structure of clusters in MID. Clusters in MID is supervised-classified using spectral signatures. Euclidean distance is used as the discrimination function for the supervised-classification method. The values of Euclidean distances are used as decision rule for the supervised-classification method.     

Hierarchical clustering of multispectral images using combined spectral and spatial criteria

An agglomerative hierarchical clustering method, which uses both spectral and spatial information for the aggregation decision, is proposed here. The method is suitable for large multispectral images, provided that an unsupervised classification is previously applied. The method is tested on a synthetic image and on a satellite image of the coastal zone.     

Spectral mapping methods applied to LiDAR data: Application to fuel type mapping

Originally developed to classify multispectral and hyperspectral images, spectral mapping methods were used to classify Light Detection and Ranging (LiDAR) data to estimate the vertical structure of vegetation for Fuel Type (FT) mapping. Three spectral mapping methods generated spatially comprehensive FT maps for Cabañeros National Park (Spain): (1) Spectral Mixture Analysis (SMA), (2) Spectral Angle Mapper (SAM), and (3) Multiple Endmember Spectral Mixture Analysis (MESMA). The Vegetation Vertical Profiles (VVPs) describe the vertical distribution of the vegetation and are used to define each FT endmember in a LiDAR signature library. Two different approaches were used to define the endmembers, one based on the field data collected in 1998 and 1999 (Approach 1) and the other on exploring spatial patterns of the singular FT discriminating factors (Approach 2). The overall accuracy is higher for Approach 2 and with best results when considering a five-FT model rather than a seven-FT model. The agreement with field data of 44% for MESMA and SMA and 40% for SAM is higher than the 38% of the official Cabañeros National Park FTs map. The principal spatial patterns for the different FTs were well captured, demonstrating the value of this novel approach using spectral mapping methods applied to LiDAR data. The error sources included the time gap between field data and LiDAR acquisition, the steep topography in parts of the study site, and the low LiDAR point density among others.     

基于分辨率退化模型的全色和多光谱遥感影像融合方法

从影像成像的频率特性出发，提出了一种影像分辨率退化模型，并在此基础上提出了一种新的全色和多光谱遥感影像融合方法。     

Markov-random-field-based super-resolution mapping for identification of urban trees in VHR images

Identification of tree crowns from remote sensing requires detailed spectral information and submeter spatial resolution imagery. Traditional pixel-based classification techniques do not fully exploit the spatial and spectral characteristics of remote sensing datasets. We propose a contextual and probabilistic method for detection of tree crowns in urban areas using a Markov random field based super resolution mapping (SRM) approach in very high resolution images. Our method defines an objective energy function in terms of the conditional probabilities of panchromatic and multispectral images and it locally optimizes the labeling of tree crown pixels. Energy and model parameter values are estimated from multiple implementations of SRM in tuning areas and the method is applied in QuickBird images to produce a 0.6 m tree crown map in a city of The Netherlands. The SRM output shows an identification rate of 66% and commission and omission errors in small trees and shrub areas. The method outperforms tree crown identification results obtained with maximum likelihood, support vector machines and SRM at nominal resolution (2.4 m) approaches.     

Multiresolution fusion of remote sensing images based on resolution degradation model

A new method based on resolution degradation model is proposed to improve both spatial and spectral quality of the synthetic images. Some ETM panchromatic and multispectral images are used to assess the new method. Its spatial and spectral effects are evaluated by qualitative and quantitative measures and the results are compared with those of IHS, PCA, Brovey, OWT(Orthogonal Wavelet Transform) and RWT (Redundant Wavelet Transform). The results show that the new method can keep almost the same spatial resolution as the panchromatic images, and the spectral effect of the new method is as good as those of waveletbased methods.     

北京城市不透水层覆盖度遥感估算

 应用线性光谱混合模型研究城市环境生物物理组成，端元的确定是其关键。城市地表同物异谱现象显著，光谱变异强烈，对于高反照率地物尤其突出。端元的光谱变异对线性光谱混合模型拟合结果产生重要影响。以同种纯净地物光谱曲线形状具有相似性为出发点，提出了一种端元优化选取方法，在此基础上计算了北京城市地表不透水层覆盖度。研究结果表明，该方法能够在一定程度上减小端元光谱变异性对线性光谱混合模型拟合结果的影响，进而提高城市不透水层覆盖度的估算精度。     

DCT Based Pansharpening of Satellite Images Using Adaptive Linear Model

Remote sensing offers a wide variety of image data with different characteristics in terms of spatial and spectral resolutions. For optical sensor systems, imaging systems have a trade-off between high spatial and high spectral resolution, and no single system offers both. Hence, in the remote sensing application, an image with ‘greater quality’ often means higher spatial and higher spectral resolution. It is, therefore, necessary and very useful to merge images with higher spectral information and higher spatial information. Pansharpening combines spatial information from the high-resolution panchromatic image and color information from multispectral bands to create a high-resolution color image. Here we propose Discrete Cosine Transform (DCT) based pansharpening algorithm using Adaptive Linear model which preserves spectral information from Multispectral image and retains spatial resolution of Panchromatic image.