On Super-Resolution versus Image Interpolation

This work presents a discussion about multi-frame Super-Resolution Reconstruction (SRR) and single-frame interpolation. SRR has been extensively studied in the last decades but it is not yet as popular as image interpolation. Since both methods are directed to improve image quality, comparisons are inevitable and desirable. However, fundamental misunderstandings can still be found in the literature on the choice between these two approaches. Recently, some mistaken conclusions have been reached regarding the SRR performance. This paper clarifies important SRR issues in image reconstruction, in contrast to interpolation. For reasons further discussed, two algorithms are considered: bicubic interpolation and LMS-SRR.