Abstract Three-dimensional (3D) semantic segmentation of point clouds is important in many scenarios, such as automatic driving, robotic navigation, while edge computing indispensable the devices. Deep learning methods based on sampling prove to be computation and memory efficient tackle large-scale (e.g. millions points). However, some local features may abandoned sampling. In this paper, We p...

We examine a variant of the clasic nearest neighbor problem, where given a set of k points, we preprocess the points so that we can quickly answer the distance to the nearest neighbor to a query point. We specifically focus on the case where all points—both the k original points as well all query points—are confined to an n×n integer grid in the plane. Furthermore, we only require the distance ...

For a given point set in Euclidean space we consider the problem of finding (approximate) nearest neighbors of a query point but restricting only to points that lie within a fixed cone with apex at the query point. Apart from being a rather natural question to ask, solutions to this problem have applications in surface reconstruction and dimension detection. We investigate the structure of the ...

We suggest a simple modification to the kd-tree search algorithm for nearest neighbor search resulting in an improved performance. The Kd-tree data structure seems to work well in finding nearest neighbors in low dimensions but its performance degrades even if the number of dimensions increases to more than three. Since the exact nearest neighbor search problem suffers from the curse of dimensi...

We describe the implementation of a fast, dynamic, approximate, nearest-neighbor search algorithm that works well in fixed dimensions (d ≤ 5), based on sorting points coordinates in Morton (or z-) ordering. Our code scales well on multi-core/cpu shared memory systems. Our implementation is competitive with the best approximate nearest neighbor searching codes available on the web, especially fo...

This paper considers two kinds of algorithms. (i) If # is a binary code of length n, a ‘‘soft decision’’ decoding algorithm for # changes an arbitrary point of Rn into a nearest codeword (nearest in Euclidean distance). (ii) Similarly a decoding algorithm for a lattice Λ in Rn changes an arbitrary point of Rn into a closest lattice point. Some general methods are given for constructing such alg...

The hubness phenomenon, as it was recently described, consists in the observation that for increasing dimensionality of a data set the distribution of the number of times a data point occurs among the k nearest neighbors of other data points becomes increasingly skewed to the right. As a consequence, so-called hubs emerge, that is, data points that appear in the lists of the k nearest neighbors...

Comparisons made in two studies of 21 methods for finding prototypes upon which to base the nearest prototype classifier are discussed. The criteria used to compare the methods are by whether they: (i) select or extract point prototypes; (ii) employ preor post-supervision; and (iii) specify the number of prototypes a priori, or obtain this number “automatically”. Numerical experiments with 5 da...