Detecting and approximating fault lines from randomly scattered data

Andrew Crampton, John C. Mason

Research output: Contribution to journalArticle

11 Citations (Scopus)

Abstract

Discretely defined surfaces that exhibit vertical displacements across unknown fault lines can be difficult to approximate accurately unless a representation of the faults is known. Accurate representations of these faults enable the construction of constrained approximation models that can successfully overcome common problems such as over-smoothing. In this paper we review an existing method for detecting fault lines and present a new detection approach based on data triangulations and discrete Gaussian curvature (DGC). Furthermore, we show that if the fault line can be described non-parametrically, then accurate support vector machine (SVM) models can be constructed that are independent of the type of triangulation used in the detection algorithms. We shall also see that SVM models are particularly effective when the data produced by the detection algorithms are noisy. We compare the performances of the various new and established models.

LanguageEnglish
Pages115-130
Number of pages16
JournalNumerical Algorithms
Volume39
Issue number1-3
DOIs
Publication statusPublished - Jul 2005

Fingerprint

Scattered Data
Fault
Line
Triangulation
Support vector machines
Support Vector Machine
Discrete Curvature
Constrained Approximation
Total curvature
Model
Smoothing
Vertical
Unknown

Cite this

@article{1d92000f7c7647e18bebae9746657d71,
title = "Detecting and approximating fault lines from randomly scattered data",
abstract = "Discretely defined surfaces that exhibit vertical displacements across unknown fault lines can be difficult to approximate accurately unless a representation of the faults is known. Accurate representations of these faults enable the construction of constrained approximation models that can successfully overcome common problems such as over-smoothing. In this paper we review an existing method for detecting fault lines and present a new detection approach based on data triangulations and discrete Gaussian curvature (DGC). Furthermore, we show that if the fault line can be described non-parametrically, then accurate support vector machine (SVM) models can be constructed that are independent of the type of triangulation used in the detection algorithms. We shall also see that SVM models are particularly effective when the data produced by the detection algorithms are noisy. We compare the performances of the various new and established models.",
keywords = "Detection and approximation algorithms, Discrete Gaussian curvature (DGC), Fault lines, Support vector machines (SVMs)",
author = "Andrew Crampton and Mason, {John C.}",
year = "2005",
month = "7",
doi = "10.1007/s11075-004-3624-y",
language = "English",
volume = "39",
pages = "115--130",
journal = "Numerical Algorithms",
issn = "1017-1398",
publisher = "Springer Netherlands",
number = "1-3",

}

Detecting and approximating fault lines from randomly scattered data. / Crampton, Andrew; Mason, John C.

In: Numerical Algorithms, Vol. 39, No. 1-3, 07.2005, p. 115-130.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Detecting and approximating fault lines from randomly scattered data

AU - Crampton, Andrew

AU - Mason, John C.

PY - 2005/7

Y1 - 2005/7

N2 - Discretely defined surfaces that exhibit vertical displacements across unknown fault lines can be difficult to approximate accurately unless a representation of the faults is known. Accurate representations of these faults enable the construction of constrained approximation models that can successfully overcome common problems such as over-smoothing. In this paper we review an existing method for detecting fault lines and present a new detection approach based on data triangulations and discrete Gaussian curvature (DGC). Furthermore, we show that if the fault line can be described non-parametrically, then accurate support vector machine (SVM) models can be constructed that are independent of the type of triangulation used in the detection algorithms. We shall also see that SVM models are particularly effective when the data produced by the detection algorithms are noisy. We compare the performances of the various new and established models.

AB - Discretely defined surfaces that exhibit vertical displacements across unknown fault lines can be difficult to approximate accurately unless a representation of the faults is known. Accurate representations of these faults enable the construction of constrained approximation models that can successfully overcome common problems such as over-smoothing. In this paper we review an existing method for detecting fault lines and present a new detection approach based on data triangulations and discrete Gaussian curvature (DGC). Furthermore, we show that if the fault line can be described non-parametrically, then accurate support vector machine (SVM) models can be constructed that are independent of the type of triangulation used in the detection algorithms. We shall also see that SVM models are particularly effective when the data produced by the detection algorithms are noisy. We compare the performances of the various new and established models.

KW - Detection and approximation algorithms

KW - Discrete Gaussian curvature (DGC)

KW - Fault lines

KW - Support vector machines (SVMs)

UR - http://www.scopus.com/inward/record.url?scp=17444403905&partnerID=8YFLogxK

UR - https://link.springer.com/journal/11075

U2 - 10.1007/s11075-004-3624-y

DO - 10.1007/s11075-004-3624-y

M3 - Article

VL - 39

SP - 115

EP - 130

JO - Numerical Algorithms

T2 - Numerical Algorithms

JF - Numerical Algorithms

SN - 1017-1398

IS - 1-3

ER -