The peculiarities of remote sensing images make RSI classification a hard task. The aim is to propose a kind of boost-classifier adapted to multi-scale segmentation. It uses the paradigm of boosting, whose principle is to combine weak classifiers to build an efficient global one. Each weak classifier is trained for one level of the segmentation and one region descriptor. It proposes and tests weak classifiers based on linear Support Vector Machines(SVM)and region distances provided by descriptors. It shows in this paper that the approach based on boosting can detect the scale and set of features best suited to a particular training set. It also shows that hierarchical multi-scale analysis is able to reduce training time and to produce a stronger classifier. The results show that the proposed methods outperform the baseline.
; 王秀娟
修改评论