Single image super resolution by adaptive K-means clustering

In recent days imaging systems have seen considerable extended usage due to their ease of use and reasonable price. However, they have weaknesses lies in image resolution. In order to increase the quality of the images, due to the technical limitations and costs of hardware parts, software techniques like the super-resolution is used, which means increasing the density of pixels in the image. The super-resolution is broken down into two categories; super-resolution using a single image and super-resolution using multiple images. In this paper, a method for increasing image quality, based on the Dong method has been proposed. In the proposed method, which is based on only one image, tries to improve the quality of image, based on the Dong method and optimizing it using a compatible selection of a vocabulary, which is based on the concept of inherent sparseness of images and appropriate adjustment statements. In this method, we have tried to present the best clustering procedure with the highest precision for selection of patches. The proposed method has been applied on different pictures from different databases. The results have been compared by using SSIM and PSNR metrics. The simulations results show that the proposed method outperforms the currently available methods.