Tampere University of Technology

TUTCRIS Research Portal

Prediction of signal denoising efficiency for DCT-based filter

Research output: Contribution to journalArticleScientificpeer-review


Original languageEnglish
Pages (from-to)1129-1142
Number of pages14
JournalTelecommunications and Radio Engineering
Issue number13
Publication statusPublished - 2019
Publication typeA1 Journal article-refereed


Signals acquired by different sensors are often noisy and are subject to filtering aimed to reduce noise and preserve important information. Although a great number of different filters exist, their performance does not always satisfy users. There are practical situations when denoising does not lead to expected positive effect which makes it useless. In this paper, we show that a denoising efficiency can be predicted for DCT-based filters. This can be accurately done by an analysis of statistics of DCT-coefficients in a limited number of blocks without execution of denoising itself. Peculiarities of preliminary analysis needed to carry out prediction are discussed. It is shown that a prediction is able to perform well for a wide range of signals and signal-to-noise ratios.


  • Efficiency, Prediction, Signal denoising

Publication forum classification

Field of science, Statistics Finland