Texture and edge preserving multiframe super-resolution

• Published in: IET image processing 2014-09, Vol.8 (9), p.499-508
• Main Authors: Turgay, Emre, Akar, Gozde B
• Format: Article
• Language: English
• Subjects: Applied sciences; Detection, estimation, filtering, equalization, prediction; discrete cosine transform; discrete cosine transforms; edge preservation; edge preserving multiframe super resolution; Estimates; Estimators; Exact sciences and technology; frequency selective characteristics; Gabor filters; Gabor fllters; Image processing; Image reconstruction; image resolution; image texture; Information, signal and communications theory; maximum likelihood estimation; maximum‐a‐posteriori estimators; multiple SR reconstruction; overlapping aliased low‐resolution observation; pixel‐to‐pixel difference; Post-processing; post‐processing step; Preserving; regularisation parameters; restoration step; Signal and communications theory; Signal processing; Signal, noise; SR literature; State of the art; super resolution image reconstruction; Surface layer; Telecommunications and information theory; Texture; texture preserving multiframe super resolution; texture‐speciflc regularisation; two stage method
• ISSN: 1751-9659; 1751-9667; 1751-9667
• DOI: 10.1049/iet-ipr.2013.0342

Super-resolution (SR) image reconstruction refers to methods where a higher resolution image is reconstructed using a set of overlapping aliased low-resolution observations of the same scene. Although edge preservation has been a widely explored topic in SR literature, texture-specific regularisation has recently gained interest. In this study, texture-specific regularisation is handled as a post-processing step. A two stage method is proposed, comprising multiple SR reconstructions with different regularisation parameters followed by a restoration step for preserving edges and textures. In the first stage, two maximum-a-posteriori estimators with two different amounts of regularisation are employed. In the second stage, pixel-to-pixel difference between these two estimates is post-processed to restore edges and textures. Frequency selective characteristics of discrete cosine transform and Gabor filters are utilised in the post-processing step. Experiments on synthetically generated images and real experiments demonstrate that the proposed methods give better results compared with the state-of-the-art SR methods especially on textures and edges.