A survey of image super-resolution algorithms based on convolutional neural networks for smart city health care applications

Image super-resolution (SR) is a technique for improving an image by increasing its spatial resolution. A super-resolution approach can improve an image's pixel intensity to the next optimized level. The SR schemes have a challenging task in several applications like remote sensing, medical imaging, and biological detection. The SR technique is divided into single-image SR (SISR) and multi-image SR (MISR) based on the number of input images (MISR). This work reviews SISR with diverse image datasets like Urban 100, Set5, Set4, and DIV2k. Convolution neural networks (CNN) are considered one of the boosting solutions to implement the superresolution and contribute to remarkable progress. This survey deals with the review of diverse convolution neural networkbased SR techniques used in various applications. The purpose of this proposed survey is to investigate several convolution neural network-based SR approaches that are employed in various applications. This study analyses the characteristics of various convolution neural network-based image super-resolution algorithms in terms of peak signal to noise ratio (PSNR) and structural similarity index measure (SSIM). According to the survey, the 3D dilated convolutional encoder-decoder network is the best qualitative & quantitative analysis method for brain MRI super-resolution.