Super-Resolution Image Reconstruction Using Wavelet Based Patch and Discrete Wavelet Transform

Abstract

This paper proposes a novel method that combines the discrete wavelet transform (DWT) and example-based technique to reconstruct a high-resolution from a low-resolution image. Although previous interpolation- and example-based methods consider the reconstruction adaptive to edge directions, they still have a problem with aliasing and blurring effects around edges. In order to address these problems, in this paper, we utilize the frequency sub-bands of the DWT that has the feature of lossless compression. Our proposed method first extracts the frequency sub-bands (Low-Low, Low-High, High-Low, High-High) from an input low-resolution image by the DWT, and then the low-resolution image is inserted into the Low-Low sub-band. Since information in high-frequency sub-bands (Low-High, High-Low, and High-High) might be lost in the low-resolution image, they are reconstructed or estimated by using example-based method from image patch database. After that, we make a high-resolution image by performing the inverse DWT of reconstructed frequency sub-bands. In experimental results, we can show that the proposed method outperforms previous approaches in terms of edge enhancement, reduced aliasing effects, and reduced blurring effects.