Single Image Super Resolution using Direction let Transform and Directional Variance

Abstract

Super-resolution  technique is defined as recovering missing spatial frequency information from a low resolution image  to retrieve its high resolution image(HR) which is equivalent to the original scene . Similarities between low resolution image  and high resolution image blocks are learned, and are used in the low resolution input  image to retrieve the high-resolution version. In this article the authors propose a  single frame super resolution method using a skewed anisotropic wavelet transform , called Directionlets Transform. The proposed method can be used for live image processing applications. Available method [2]  which is  based on  directionlet transform is computationally very intensive. In this paper using directional variance method, the directionlet method is modified which further reduces the computation and time.