A Residual Dense Generative Adversarial Network For Pansharpening With Geometrical Constraints
| Autor: | Anais Gastineau, Jean-François Aujol, Christian Germain, Yannick Berthoumieu |
|---|---|
| Přispěvatelé: | Institut de Mathématiques de Bordeaux (IMB), Université Bordeaux Segalen - Bordeaux 2-Université Sciences et Technologies - Bordeaux 1-Université de Bordeaux (UB)-Institut Polytechnique de Bordeaux (Bordeaux INP)-Centre National de la Recherche Scientifique (CNRS), Laboratoire de l'intégration, du matériau au système (IMS), Centre National de la Recherche Scientifique (CNRS)-Institut Polytechnique de Bordeaux-Université Sciences et Technologies - Bordeaux 1, ANR-18-CE92-0050,SUPREMATIM,Super-résolution d'images multi-échelles en sciences des matériaux avec des attributs géométriques(2018) |
| Rok vydání: | 2020 |
| Předmět: |
Computer science
Multispectral image ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION 0211 other engineering and technologies Pansharpening 02 engineering and technology Residual Regularization (mathematics) Generative Adversarial Network Image (mathematics) Panchromatic film regularization remote sensing Computer Science::Computer Vision and Pattern Recognition residual dense network 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Point (geometry) [MATH]Mathematics [math] Algorithm Image resolution 021101 geological & geomatics engineering |
| Zdroj: | ICIP 27th IEEE international conference on image processing (ICIP 2020) 27th IEEE international conference on image processing (ICIP 2020), Oct 2020, Abou Dabi, United Arab Emirates |
| DOI: | 10.1109/icip40778.2020.9191230 |
| Popis: | International audience; The pansharpening problem consists in fusing a high resolution panchromatic image with a low resolution multispectral image in order to obtain a high resolution multispectral image. In this paper, we adapt a Residual Dense architecture for the generator in a Generative Adversarial Network framework. Indeed, this type of architecture avoids the vanishing gradient problem faced when training a network by re-injecting previous information thanks to dense and residual connections. Moreover, an important point for the pansharpening problem is to preserve the geometry of the image. Hence, we propose to add a regularization term in the loss function of the generator: it preserves the geometry of the target image so that a better solution is obtained. In addition, we propose geometrical measures that illustrate the advantages of this new method. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|