Single-image superresolution based on local regression and nonlocal self-similarity

Abstract

Abstract. The challenge of learning-based superresolution (SR) is to predict the relationships between low-resolution (LR) patches and their corresponding high-resolution (HR) patches. By learning such relationships from external training images, the existing learning-based SR approaches are often affected by the relevance between the training data and the LR input image. Therefore, we propose a single-image SR method that learns the LR-HR relations from the given LR image itself instead of any external images. Both the local regression model and nonlocal patch redundancy are exploited in the proposed method. The local regression model is employed to derive the mapping functions between self-LR-HR example patches, and the nonlocal self-similarity gives rise to a high-order derivative estimation of the derived mapping function. Moreover, to fully exploit the multiscale similarities inside the LR input image, we accumulate the previous reconstruction results and their corresponding LR versions as additional example patches for the subsequent estimation process, and adopt a gradual magnification scheme to achieve the desired zooming size step by step. Extensive experiments on benchmark images have validated the effectiveness of the proposed method. Compared to other state-of-the-art SR approaches, the proposed method provides photorealistic HR images with sharp edges. © The Authors.Published by SPIE under a Creative Commons Attribution 3.0 Unported License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.

Second-Order Regression-Based MR Image Upsampling

“…On the other hand, ( 3 ) also reveals that, in order to reconstruct the HR patch q , multiorders of the mapping function derivatives should be estimated first. Several authors [ 16 , 17 ] have argued in their work that, for image reconstruction, a second-order derivative estimation is able to adequately balance detail preservation and computation time. Therefore, we propose a second-order derivation of the mapping function in this method.…”

Single image super-resolution via blind blurring estimation and dictionary learning

“…NLSS in natural images can be exploited as useful prior knowledge for various image restoration tasks, such as super resolution [10,[15][16][17][18], image denoising [19], deblurring [20] and inpainting [21] etc. However, it is not necessary to consider the redundancy of all patches most of the time according to our previous observations.…”

Single image super-resolution via blind blurring estimation and anchored space mapping

“…NLSS has been employed in a lot of computer vision fields such as super resolution [12,[18][19][20][21], denoising [22], deblurring [23], and inpainting [24] etc. Further, Zontak and Irani [18] quantified this property by relating it to the spatial distance from the patch and the mean gradient magnitude |grad| of a patch.…”