Support vector machines for spectral unmixing
暂无分享,去创建一个
Mixture modelling is becoming an increasingly important tool in the remote sensing community as researchers attempt to resolve sub-pixel, area information. This paper describes an approach based on a relatively new technique, support vector machines (SVMs), and compares this with more established algorithms such as linear spectral mixture models (LSMMs) and artificial neural networks (ANNs). In the simplest case, the mixture regions formed by the linear SVM and the LSMM are equivalent. Extensions to the basic SVM algorithm allow the technique to be applied to data sets that exhibit spectral confusion and to data sets that have non-linear mixture regions. The paper highlights the key advantage offered by the SVM approach in that it selects end-members (pure pixels) automatically and the potential of the SVM method is demonstrated using a Landsat TM data set.
[1] J. Settle,et al. Linear mixing and the estimation of ground cover proportions , 1993 .
[2] Marti A. Hearst. Trends & Controversies: Support Vector Machines , 1998, IEEE Intell. Syst..
[3] Martin Brown,et al. Linear spectral mixture models and support vector machines for remote sensing , 2000, IEEE Trans. Geosci. Remote. Sens..