Please use this identifier to cite or link to this item:
|Title:||Cloud removal in optical remote sensing imagery based on direct translation from SAR to optical image using deep learning||Authors:||Zhou, Hao||Keywords:||Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
Engineering::Electrical and electronic engineering
|Issue Date:||2021||Publisher:||Nanyang Technological University||Source:||Zhou, H. (2021). Cloud removal in optical remote sensing imagery based on direct translation from SAR to optical image using deep learning. Master's thesis, Nanyang Technological University, Singapore. https://hdl.handle.net/10356/154639||Project:||ISM-DISS-02473||Abstract:||Synthetic Aperture Radar (SAR) is an indispensable remote sensing technology nowadays. However, due to the different imaging theory applied in Synthetic Aperture Radars, the interpretation of SAR images may come out as extremely different from conventional optical satellite images. Thus, to tackle with the interpretation problem, a specialized SAR-optical image translation model is developed to directly translate the original SAR images into equivalent optical satellite images. This model is implemented with a novel two-step Generative Adversarial Network architecture. To present the performance of proposed model on SAR-Optical image translation task, remote sensing data acquired from Sentinel-1 and Sentinel-2 is utilized for the model training and validation phase. The final results indicate a promising performance both on enhancing the human perception of translated optical images and increasing the statistical indices of PNSR and SSIM, which have reached at 19.09 dB and 0.4211 respectively.||URI:||https://hdl.handle.net/10356/154639||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Theses|
Updated on Jan 20, 2022
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.