| Ref.No: | 95037900 |
| Start date: | 13.12.2024 |
| End date: | 12.12.2025 |
| Approval date: | 13.12.2024 |
| Department: | RURAL & SURVEYING ENGINEERING |
| Sector: | TOPOGRAPHY SCOPE |
| Financier: | ΤΑΜΕΙΑΚΑ ΥΠΟΛΟΙΠΑ, AUTOHRIMATODOTOUMENA |
| Budget: | 14.236,05 € |
| Public key: | 61ΟΘ46ΨΖΣ4-ΧΛΟ |
| Scientific Responsible: | Prof. KARATHANASI |
| Email: | karathan@survey.ntua.gr |
| Description: | MANY DEEP LEARNING ARCHITECTURES HAVE BEEN DEVELOPED TO SUPER RESOLVE IMAGES. HOWEVER, THEIR APPLICATION TO REMOTE SENSING IMAGES REQUIRES SPECIFIC ADAPTATION OF THE NETWORKS ,AS WELL AS THEIR TRAINING PROCESSES TO ENSURE THE PRESERVATION OF SPECTRAL INFORMATION. THE PROJECT WE EVALUATE VARIOUS NETWORKS, INCLUDING TSRCNN (SUPER-RESOLUTION CONVOLUTIONAL NEURAL NETWORK), SRGAN (SUPER-RESOLUTION GENERATIVE ADVERSARIAL NETWORK), ESRGAN (ENHANCED SUPER-RESOLUTION GAN), PIX2PIX (CONDITIONAL GAN (CGAN)), AND SWINIR (SWIN TRANSFORMER FOR IMAGE RESTORATION) IN TERMS OF IMPROVING THE SPATIAL RESOLUTION OF MULTISPECTRAL IMAGERY WHILE MAINTAINING THE SPECTRAL INFORMATION. |
