Abstract:
Single image super-resolution (SISR) is a difficult inverse problem because it requires reconstructing high-frequency details from low-resolution images. Most CNN-based super-resolution (SR) methods directly learn the mapping from low-resolution to high-resolution using complex convolutional neural networks.
However, blindly increasing network depth improves performance marginally but costs a lot. Integrating image prior knowledge into the model helps image reconstruction more efficiently. The soft-edge is an important image feature in many computer vision tasks.
This paper proposes a Soft-edge assisted Network (SeaNet) to reconstruct high-quality SR images using image soft-edge. Three subnets—RIRN, Edge-Net, and IRN—make up SeaNet. Reconstruction has two phases. Stage-I RIRN and Edge-Net reconstruct the rough SR feature maps and soft-edge.
Stage-II fuses the outputs of previous stages and feeds them to the IRN for high-quality SR image reconstruction. Extensive experiments show that SeaNet converges quickly and performs well with image soft-edge.
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
