Multiregression spatially variant blur kernel estimation based on inter‐kernel consistency

Abstract Most single‐image super‐resolution (SR) models suffer from the degradation of image restoration performance when restoring a high‐resolution (HR) low‐resolution (LR) downscaled using an unknown blur kernel. The spatially invariant kernel estimators have been proposed to predict address this issue. Nevertheless, variant exists in real‐world; thus, these are unsuitable for real‐world applications. Although proposed, SR still because do neither consider consistency between surrounding kernels nor refine non‐parametric as parameters. To problem, authors propose multiregression estimation based on inter‐kernel consistency. estimator consists three parts: regression, block, and parametric regression. Specifically, it predicts while considering nearby kernels. Our source codes with pretrained available https://github.com/alsgur0720/multiregression .