The block-matching with 3D transform domain collaborative filtering (BM3D) achieves very good performance in image denoising. However, BM3D becomes ineffective when an image is h...
We propose an effective color image denoising method that exploits ltering in highly sparse local 3D transform domain in each channel of a luminance-chrominance color space. For e...
Kostadin Dabov, Alessandro Foi, Vladimir Katkovnik...
Abstract We review the evolution of the nonparametric regression modeling in imaging from the local Nadaraya-Watson kernel estimate to the nonlocal means and further to transform-d...
Vladimir Katkovnik, Alessandro Foi, Karen Egiazari...
In this paper we study interest point descriptors for image matching and 3D reconstruction. We examine the building blocks of descriptor algorithms and evaluate numerous combinati...
This paper presents a 3D approach to multi-view object class detection. Most existing approaches recognize object classes for a particular viewpoint or combine classifiers for a f...