Wavelet-Based Deep Generative Framework for Super Resolution of Low-Resolution Labeled Maps and Weak-Supervised Learning

The unavailability of pixel-level detailed labels is a crucial challenge in the field of remote sensing image analysis. Deep learning (DL) models require a large number of labeled samples for an accurate estimation of a large number of trainable parameters. However, in remote sensing applications usually only a few reliable labeled data are available for the learning of a classifier, whereas often many weak/low-resolution unreliable labeled data can be collected from available land-cover maps. Accordingly, weak-supervised learning (WSL) may overcome the problems by using noisy and low-resolution labels in remote sensing. In this article, we propose a deep adversarial model based on discrete wavelet transform (WT) to exploit weak/lowresolution label information for generating refined super-resolved weak reference maps (WRMs). Our contribution includes the development of a discrete WT-based generator for enhancing the low-resolution labels to generate a refined high-resolution reference map. We also present an efficient framework for multisource image fusion that incorporates the refined superresolved WRM, synthetic aperture radar (SAR) images, and corresponding low-resolution labels. Our findings highlight the effectiveness of the refined super-resolved WRM. In addition, we investigate the impact of the high-resolution reference maps on segmentation accuracy, which reveals their potential in improving the segmentation performance compared with other reference methods.