Gradient-based edge preserving interpolation and its application to super-resolution

Super-resolution is a process for obtaining high-quality, high-resolution images from one or a set of low-resolution images. The most practical methods for image super-resolution are reconstruction-based methods, which minimize the difference between observed low-resolution images and the estimate for high-resolution images. Therein, the interpolation step plays a key role in the estimated high-resolution image quality. Usually, the conventional bilinear or bicubic methods are used in reconstruction-based super-resolution. However, these conventional interpolations generally lead to blurring in edge regions and need more time for convergence in the reconstruction-based super-resolution method. Therefore, this paper proposes a gradient-based edge-preserving interpolation method, which can reduce not only artifact noise but also blurring near the edge regions in the estimated high-resolution image. Furthermore, our proposed interpolation method can also solve high-complexity, time-consuming problems in the recently developed new edge-directed interpolation, which usually can achieve sharp edges in the high-resolution reconstructed image. Experiments confirm that our proposed interpolation method for image super-resolution is more effective than the conventional interpolation methods.