One-class classification techniques in image recognition problems

The one class classification problem is different from the binary/multi-class classification problem in the sense that the negative class is not present or not properly characterized. The problem of classifying positive (target) cases in the absence of appropriately defined negative cases (outliers) has been given an increased attention during the last few years. In this paper we give a brief overview about the methods used for one class classification applied in three fields of interest: (a) remote sensing, (b) 3D image retrieval and modeling and (c) medical image/data analysis/processing. We also present results of the application of one of the methods (one-class SVM) for the detection of vegetated areas in a hyperspectral remote sensing image.