The choice of vantage points in vp-tree construction for copy-move detection in digital images

Binary space partitioning trees are a known way to speed up searches in multidimensional spaces. VP-trees have a high efficiency among various types of binary space partitioning trees and have proven their advantages in solving the problem of finding duplicate fragments. In this paper, we investigate several ways to select vantage points in vp-tree construction. The study was conducted using several different types of images and feature extraction techniques.