PROJECT TITLE :

Semisupervised Discriminant Feature Learning for SAR Image Category via Sparse Ensemble

ABSTRACT:

Terrain scene classification plays an necessary role in varied artificial aperture radar (SAR) image understanding and interpretation. This paper presents a completely unique approach to characterize SAR image content by addressing category with a limited variety of labeled samples. In the proposed approach, every SAR image patch is characterize by a discriminant feature which is generated in an exceedingly semisupervised manner by utilizing a spare ensemble learning procedure. In particular, a nonnegative sparse coding procedure is applied on the given SAR image patch set to generate the feature descriptors initial. The set is combined with a restricted variety of labeled SAR image patches and an abundant range of unlabeled ones. Then, a semisupervised sampling approach is proposed to construct a collection of weak learners, in that every one is modeled by a logistic regression procedure. The discriminant data can be introduced by projecting SAR image patch on every weak learner. Finally, the features of SAR image patches are produced by a sparse ensemble procedure that can scale back the redundancy of multiple weak learners. Experimental results show that the proposed discriminant feature learning approach can achieve a better classification accuracy than several state-of-the-art approaches.