期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2006
卷号：XXXVI Part 7
出版社：Copernicus Publications
摘要：Support Vector Machines (SVMs) are a statistical learning theory based techniques and have been applied to different fields. For the pattern recognition case, SVMs have been used for isolated handwritten digit recognition, object recognition, charmed quark detection, face detection in images and text categorization. SVM have been shown to perform well for density estimation also where the probability distribution function of the feature vector can be inferred from a random sample. In this work SVM has been used for density estimation, and it uses Mean Field (MF) theory for developing an easy and efficient learning procedure for the SVM. In SVM a kernel function determines the characteristic of an SVM. The kernel functions used in SVM are defined as local kernels, global kernels and spectral kernels. In the case of local kernel only the data that are close or in the proximity of each others have an influence on the kernel values. In global kernel samples that are far away from each other still have an influence on the kernel value. A spectral kernel uses the spectral knowledge into SVM classification, which reduces false alarms for thematic classification. In this paper the effect of different mixed kernels generated while taking spectral kernel with local or global kernels have been studied on overall sub-pixel classification accuracy of remote sensing data using Fuzzy Error Matrix (FERM)
关键词：Sub-Pixel; Support Vector Machine (SVM); Mixed Kernel Function; Density Estimation; Fuzzy Error Matrix ; (FERM).
