A comparison of techniques for super-resolution evaluation
Monica A. Trifas, Jeremy Straub · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2011
This paper reviews the perception of image quality by human respondents in the context of super-resolution. Specifically, the desirability of an automated mechanism for generating a metric that correlates well with human perceptions of image quality is discussed. Common metrics are evaluated to ascertain whether they demonstrate suitable correlation with human perception of image quality. It is found that the commonly used pixel-difference evaluation technique outperforms a threshold-based technique; however, neither is demonstrated to correlate well with human perception.