Locally adaptive tree-based thresholding using the treethresh package in R

Evers, L. and Heaton, T. (2017) Locally adaptive tree-based thresholding using the treethresh package in R. Journal of Statistical Software, 78, Code S 2. (doi: 10.18637/jss.v078.c02)

Abstract

This paper introduces the treethresh package offering accurate estimation, via thresholding, of potentially sparse heterogeneous signals and the denoising of images using wavelets. It gives considerably improved performance over other estimation methods if the underlying signal or image is not homogeneous throughout but instead has distinct regions with differing sparsity or strength characteristics. It aims to identify these different regions and perform separate estimation in each accordingly. The base algorithm offers code which can be applied directly to any one-dimensional potentially sparse sequence observed subject to noise. Also included are functions which allow two-dimensional images to be denoised following transformation to the wavelet domain. In addition to reconstructing the underlying signal or image, the package provides information on the believed partitioning of the signal or image into its differing regions.

• Item Type: Articles
• Status: Published
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Evers, Dr Ludger
• Authors: Evers, L., and Heaton, T.
• College/School: College of Science and Engineering > School of Mathematics and Statistics > Statistics
• Journal Name: Journal of Statistical Software
• Publisher: Foundation for Open Access Statistics
• ISSN: 1548-7660
• ISSN (Online): 1548-7660
• Published Online: 06 July 2017
• Copyright Holders: Copyright © 2017 The Authors
• First Published: First published in Journal of Statistical Software 78:Code Snippet 2
• Publisher Policy: Reproduced under a Creative Commons License