Convolution Neural Network (CNN) based High R...

Convolution Neural Network (CNN) based High Resolution Image Reconstruction

Author Name : Mr. Ritesh Kumar Singh, Roshan Banbariya

ABSTRACT

DL modern techniques have shown promising outcomes in SISR image reconstruction from noisy and blurry LR data of images. SR is the process of creating HR images from LR images. This example considers SISR, where the goal is to recover one HR image from one LR image. SISR is challenging because high-frequency image content typically cannot be recovered from the LR image. Without high-frequency information, the quality of the HR image is limited. Further, SISR is an ill-posed problem because one LR image can yield several possible high-resolution images. In this paper we proposed VDSR network for single super resolution imaging. Apart from that we have done comparative study between two methods of resolution that is bicubic interpolation and CNN based VDSR network for single super resolution imaging.

Keywords:SISR, deep learning, neural networks, and objective function