The farthest point strategy for progressive image sampling

By Y. Eldar, M. Lindenbaum, M. Porat, Yehoshua Y. Zeevi. 1994

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

A Point Set Generation Network for 3D Object Reconstruction from a Single Image (2016, methodology): “To compute CD and EMD, their predicted and ground truth volumes are sampled by iterative farthest point sampling [7] to a discrete set of points with the same cardinality as ours.”
Text-to-3D using Gaussian Splatting (2023, methodology): “To avoid too many vertices in the provided shape, we use farthest point sampling (Eldar et al., 1997) for point clouds and uniform surface sampling for meshes to extract a subset of the original shape instead of directly using all the vertices or points.”
GraspNet-1Billion: A Large-Scale Benchmark for General Object Grasping (2020, background): “We subsample M points with farthest point sampling [10] to cover the whole scene.”
Monte Carlo convolution for learning on non-uniformly sampled point clouds (2018, methodology): “It has favorable scalability compared to the state-of-the-art Farthest Point [Eldar et al. 1997] sampling and allows to bound the maximal number of samples in a receptive field.”
Partial Similarity of Objects, or How to Compare a Centaur to a Horse (2009, methodology): “The shapes were represented as binary images and sub-sampled using the farthest point strategy [34, 21, 33] at approximately 3000 points.”
PointGrid: A Deep Network for 3D Shape Understanding (2018, background): “furthest sampling [9], which may provide better representation of the local shape.”
Image denoising with nonlocal spectral graph wavelets (2010, methodology): “This heuristic is an example of the “farthest point” strategy developed previously in the context of progressive image sampling by Eldar [24], and also”
Recent Advances in Remeshing of Surfaces (2008, background): “The farthest point paradigm [49] advocates inserting one sample point at a time, as far as possible from previously placed samples, i.”
ARCTIC: A Dataset for Dexterous Bimanual Hand-Object Manipulation (2022, methodology): “Similarly, we pre-define 3D landmarks for objects using farthest point sampling [15, 16] on the object mesh.”
On Bending Invariant Signatures for Surfaces (2003, methodology): “This subsampling procedure is also known as iterative Voronoi sampling, see, e.g., [16].”