Pso and Ga Based Neighbor Emedding Super Resolution

In this paper a novel technique for Neighbor embedding single image super resolution (SR) is proposed. Given a low-resolution image, its high-resolution image is reconstructed from a set of training images, which are composed of one or more low-resolution and corresponding highresolution image pairs. In this paper we propose a new approach to a single image super-resolution through neighbor embedding using Genetic Algorithms (GA) and Particle Swarm Optimization (PSO). GAs and PSO are used for patch size, overlap and K nearest neighbor parameters tuning of neighbor embedding super resolution by maximizing the PSNR as a fitness value. Experiments show that the use of GA and PSO for finding the parameters of neighbor embedding method is more accurate than setting the parameters as random. Also, it can be seen from the results that the proposed method increases the average of PSNR 2.2db in comparing with Bicubic interpolation, but the PSNR differences between PSO and GAs are not significant.