Image interpolation with directionlets

Abstract

We present a novel directionally adaptive image interpolation based on a multiple-direction wavelet transform, called directionlets. The algorithm uses directionlets to efficiently capture directional features and to extract edge information along different directions from the low-resolution image. Then, the high-resolution image is generated using this information to preserve sharpness of details. Our interpolation algorithm outperforms the state-of-the-art methods in terms of both numeric and visual quality of the interpolated image.