Scale Invariant Super-Resolutions Methods with Application to InSAR Images
| Autor: | Khaled A. Helal, Bardia Barabadi, Nikitas J. Dimopoulos, Amirali Baniasadi |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
Ground truth
Artificial neural network business.industry Computer science 020206 networking & telecommunications Pattern recognition 02 engineering and technology Scale invariance Convolutional neural network Interferometric synthetic aperture radar 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Artificial intelligence Invariant (mathematics) business Scaling Image resolution |
| Zdroj: | APCCAS |
| DOI: | 10.1109/apccas47518.2019.8953090 |
| Popis: | Super-Resolution is the process of generating high-resolution (HR) images from a low-resolution (LR) ones. In learning-based SR algorithms, artificial neural networks (ANN) are used. This is achieved by training the network using HR and LR image pairs and use this network later to create new HR images from LR ones. Our work postulates that the scaling process is invariant across scales. Thus, a model trained at lower scales can be used to reconstruct higher resolution images when the ground truth is not available to train the model. We call this approach Scale Invariant Super-Resolution (SINV) We evaluated SINV using different datasets, and with different upscaling factors11The upscaling factor is the factor by which the image resolution is increased. and showed that it outperforms conventional approaches. We have applied SINV to processing InSAR images. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|