Optimising Data for Exemplar-Based Inpainting

Optimised Sparse Image Reconstruction Via Densification and Relocation

“…the higher the PSNR is, the lower the MSE is (better). We experiment with classic images that are often encountered in the context of reconstruction-based image compression [27,38,42,45]. All images are scaled to 256 × 256 and are greyscale unless otherwise stated.…”

Reconstruction of Smooth 3D Color Functions from Keypoints: Application to Lossy Compression and Exemplar-Based Generation of Color LUTs

“…The search for a minimal subset of keypoints for the reconstruction of image data is a problem that has already been addressed in the literature, e.g. for the purpose of compression or inpainting of 2D natural images, as in [7,13,15,17]. In these papers, the keypoint determination methods are stochastic and can therefore generate different sets of points at each run, from the same input image.…”

“…and only improved sets are kept over time, until convergence towards a stable state. In [15], a bottom-up approach based on a dithering method of the error map is proposed to add new keypoints, followed by a pixel exchange step. In [13], a similar idea is used, with stochastic sparsification of image laplacian halftoning.…”

“…Our algorithm has not been compared to other bottom-up approaches to determining key points (e.g [13,15]). Indeed, the stochastic aspect of these methods means that the number of 3D reconstructions will generally be greater.…”

Exemplar-Based Image Inpainting With Multi-Resolution Information and the Graph Cut Technique

“…In [15], [32], the minimization problem of total variation was solved by split Bregman algorithm, however, the filling of large texture region art still blurry. Karos et al [18] proposed optimizing data for inpainting of exemplar-based technique, new pixels are added by shaking the current error map, which use data optimization strategies for inpainting.…”