Normalization of Relative Position Descriptors

Loading...
Thumbnail Image

Date

2018-08-29

Authors

Jazouli, Manal

Journal Title

Journal ISSN

Volume Title

Publisher

University of Guelph

Abstract

A relative position descriptor (RPD) is a quantitative representation of the position of a spatial object relative to another. RPDs have been used in applications such as scene matching, human-robot interaction, and geospatial information retrieval. RPDs are usually not invariant under geometric transformations (i.e., the RPD changes when a geometric transformation is applied to the objects). However, some RPDs can be normalized and made invariant under these transformations − and this is a desirable property. Here, we review and introduce new normalization methods for well-known RPDs, namely, the force histogram and the φ−descriptor.

Description

Keywords

Relative Position Descriptor, Force histogram, Phi-descriptor, image descriptors, Normalization, affine invariance

Citation