In this work efficient geometric algorithms are provided for the linear approximation of digital signals under the uniform norm. Given a set of n points (xi, yi)i=1..n, with xi < xj if i < j, we give a new method to find the optimum linear approximation in O(n). Given also an error bound, we demonstrate how to construct in O(n) a non continuous piecewise solution such that the number k of segments is optimal. Furthermore we show that for such number of segments, the solution that is l∞ optimal can also be found in O(n) provided that n/k = O(1).

Efficient (Piecewise) Linear Minmax Approximation of Digital Signals

DALAI, Marco
Methodology
;
LEONARDI, Riccardo
Conceptualization
2004-01-01

Abstract

In this work efficient geometric algorithms are provided for the linear approximation of digital signals under the uniform norm. Given a set of n points (xi, yi)i=1..n, with xi < xj if i < j, we give a new method to find the optimum linear approximation in O(n). Given also an error bound, we demonstrate how to construct in O(n) a non continuous piecewise solution such that the number k of segments is optimal. Furthermore we show that for such number of segments, the solution that is l∞ optimal can also be found in O(n) provided that n/k = O(1).
2004
0780384849
File in questo prodotto:
File Dimensione Formato  
DL_ICASSP-2004_full-text.pdf

gestori archivio

Tipologia: Full Text
Licenza: NON PUBBLICO - Accesso privato/ristretto
Dimensione 199.92 kB
Formato Adobe PDF
199.92 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
DL_ICASSP-2004_post-print.pdf

accesso aperto

Descrizione: DL_ICASSP-2004_post-print
Tipologia: Documento in Post-print
Licenza: Creative commons
Dimensione 82.3 kB
Formato Adobe PDF
82.3 kB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11379/15014
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 3
  • ???jsp.display-item.citation.isi??? 0
social impact