This paper presents a novel application of advanced machine learning techniques for Mars terrain image classification. Fuzzy-rough feature selection (FRFS) is employed in conjunction with Support Vector Machines (SVMs) to construct image classifiers. These techniques are for the first time, integrated to address problems in space engineering where the images are of many classes and large-scale. The use of FRFS allows the induction of low-dimensionality feature sets from feature patterns of a much higher dimensionality. Experimental results demonstrate that FRFS helps to enhance the efficacy of the conventional classifiers. The resultant SVM-based classifiers which utilise FRFS-selected features generally outperform K-Nearest Neighbours and Decision Tree based clas- sifiers and those which use PCA-returned features.
|Publication status||Published - 25 Oct 2011|