Due to the visual difference between the Synthetic Aperture Radar (SAR) image and optical image, it is difficult to accurately label the SAR image, leading to the fact that there are only few labeled images with lots of unlabeled images. In this situation, the self-learning, which utilizes both the labeled samples and unlabeled samples to learn an optimal classifier, performs well. For self-learning, the performance of the initial classifier has a great influence on the following learning. During the self-leaning procedure, the classifier might easily get incorrect predictions of unlabeled samples provided by itself especially in original rounds when the accuracy of classifier is undesirable. Therefore, the performance of self-learning is usually unstable. Based on the active sample proposal and modified self-learning, this paper gives a novel semi-supervised method for SAR target discrimination. Firstly, an undirected graph is constructed by using all the unlabeled samples and the most informative samples are selected to be labeled by man. Secondly, confidence of the classifier’s prediction on unlabeled samples is achieved by both the discrimination result and local geometrical information. Experimental results on the measured SAR dataset illustrate the proposed semi-supervised discrimination method can still obtain good discrimination performance with few labeled samples.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.