Abstract:
Multi-modal image super-resolution (MISR) uses an HR image from another modality to find the HR version of a low-resolution (LR) image. This paper designs a model-based MISR deep network architecture.
A novel joint multi-modal dictionary learning (JMDL) algorithm models cross-modality dependency. JMDL simultaneously learns three dictionaries and two transform matrices to combine modalities. We then create a deep coupled ISTA network from the JMDL model by unfolding the iterative shrinkage and thresholding algorithm (ISTA).
We propose a layer-wise optimization algorithm (LOA) to initialize network parameters before back-propagation since network initialization is crucial to deep network training. Convex optimization solves the network initialization as a multi-layer dictionary learning problem.
The LOA effectively reduces training loss and improves reconstruction accuracy. Finally, we compare our MISR method to other top methods. Our method consistently outperforms others at different upscaling factors for multi-modal scenarios.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
