Region-based super-resolution using adaptive diffusion regularization

In conventional super-resolution algorithms, a single scheme is usually adopted and applied for a whole image, regardless of region characteristics. Since an image consists of various regions having different characteristics, these algorithms may not provide equally good performance for all regions. To alleviate this fundamental drawback of the conventional approach, we propose a region-based super-resolution algorithm. In the algorithm, an image is first analyzed and segmented into smooth, intermediate, and edge regions. According to the region type, a different diffusion-based regularization term and different reconstruction parameters are then used. For regularization in edge regions in particular, we adopt directional smoothing based on adaptive anisotropic diffusion, rather than omni-directional smoothing, to restore the continuity and sharpness of edges and simultaneously reduce undesired noise. The reconstruction parameters are independently managed, depending on the region type, to maximize the visual quality of the whole image. Experimental results show that the proposed method improves the objective quality as well as the subjective visual quality of reconstructed high-resolution images.