This article is cited in 5 papers

IMAGE PROCESSING, PATTERN RECOGNITION

Adaptive interpolation based on optimization of the decision rule in a multidimensional feature space

M. V. Gashnikov<sup>ab

a</sup> Samara National Research University, Moskovskoye Shosse 34, 443086, Samara, Russia
^b IPSI RAS – Branch of the FSRC "Crystallography and Photonics" RAS, Molodogvardeyskaya 151, 443001, Samara, Russia

Abstract: An adaptive multidimensional signal interpolator is proposed, which selects an interpolating function at each signal point by means of the decision rule optimized in a multidimensional feature space using a decision tree. The search for the dividing boundary when splitting the decision tree vertices is carried out by a recurrence procedure that allows, in addition to the search for the boundary, selecting the best pair of interpolating functions from a predetermined set of functions of an arbitrary form. Results of computational experiments in nature multidimensional signals are presented, confirming the effectiveness of the adaptive interpolator.

Keywords: multidimensional signal, adaptive interpolation, multidimensional feature, optimization, interpolation error.

Received: 05.11.2019
Accepted: 29.11.2019

DOI: 10.18287/2412-6179-CO-661