Feature-Guided SAR-to-Optical Image Translation

Jiexin Zhang,Jianjiang Zhou,Xiwen Lu

doi:10.1109/access.2020.2987105

Abstract

The powerful performance of Generative Adversarial Networks (GANs) in image-to-image translation has been well demonstrated in recent years. However, most methods are focused on completing an isolated image translation task. With the complex scenes in optical images and high-frequency speckle noise in SAR images, the quality of generated images is often unsatisfactory. In this paper, a feature-guided method for SAR-to-optical image translation is proposed to better take the unique attributes of images into account. Specifically, in view of the diversity of structure features and texture features, VGG-19 network is used as the feature extractor in the task of cross-modal image translation. To ensure the acquisition of multilayer features in the process of image generation, feature matching is carried out on different layers. Loss function based on Discrete Cosine Transform is designed to filter out the high-frequency noise. The generated images show better performance in feature preservation and noise reduction, and achieve higher Image Quality Assessment scores compared with images generated by some famous methods. The superiority of our algorithm is also demonstrated by being applied to different networks.

Highlights

With the rapid development of aerial remote sensing technology, researches concerned with earth observation system and remote sensing technology used for monitoring the state of industry, agriculture and forestry have gradually become prevailing in earth system science and information science [1]–[3]
We firstly look for the superior network for Synthetic Aperture Radar (SAR)-to-optical task to extract the features of SAR images and optical remote sensing images, and take feature matching to guarantee the feature similarity
2) We use a novel multilayer feature matching module based on VGG’s superior ability of cross-modal feature extraction, which can significantly preserve more information than networks without it when applied to the task of SAR-to-optical image translation

Summary

Introduction

With the rapid development of aerial remote sensing technology, researches concerned with earth observation system and remote sensing technology used for monitoring the state of industry, agriculture and forestry have gradually become prevailing in earth system science and information science [1]–[3]. INDEX TERMS SAR-to-optical image translation, feature extraction, high-frequency noise, generative adversarial networks. J. Zhang et al.: Feature-Guided SAR-to-Optical Image Translation pix2pixHD [20] based on pix2pix framework in 2018, which can generate high-resolution images.

Results

Conclusion