Single-frame super-resolution using superpixel based dictionary
Taku Nakahara, Kazunori Uruma, Tomohiro Takahashi, Toshihiro Furukawa · 2016
This paper proposes a single-frame super-resolution algorithm based on the dictionary learning technique. A lot of studies of the dictionary learning based super-resolution algorithm achieve a high performance. However, the dictionary learning based algorithm often demands the sample image set to construct the dictionaries, and the performance of the algorithm depend heavily on the sample image set. Then, this paper proposes to construct an appropriate dictionary without using sample image set, and a new super resolution algorithm is proposed. In order to construct an appropriate dictionary using an input low resolution image, this paper introduces the superpixel based image segmentation technique. Experiment results show the effectiveness of the proposed algorithm.