Reliability-based Mesh-to-Grid Image Reconstruction
This work addresses image reconstruction for tasks such as super-resolution and virtual view generation, offering incremental improvements in quality.
The paper tackles the problem of reconstructing images from non-integer sample positions, common in applications like super-resolution and multi-camera systems, by introducing a reliability-based content-adaptive framework that refines initial estimates with denoising, achieving over 2 dB PSNR improvement over the initial estimate and up to 0.7 dB gain over state-of-the-art methods.
This paper presents a novel method for the reconstruction of images from samples located at non-integer positions, called mesh. This is a common scenario for many image processing applications, such as super-resolution, warping or virtual view generation in multi-camera systems. The proposed method relies on a set of initial estimates that are later refined by a new reliability-based content-adaptive framework that employs denoising in order to reduce the reconstruction error. The reliability of the initial estimate is computed so stronger denoising is applied to less reliable estimates. The proposed technique can improve the reconstruction quality by more than 2 dB (in terms of PSNR) with respect to the initial estimate and it outperforms the state-of-the-art denoising-based refinement by up to 0.7 dB.