Classification of Footwear Outsole Patterns using Fourier Transform and Local Interest Points

Classification of Footwear Outsole Patterns using Fourier Transform and Local Interest Points

Highlights • Query images vary in media, transfer mechanism, enhancement technique and substrate. • Automated retrieval methods were compared using CMC and ROC curves. • Of all methods considered, phase-only correlation showed superior database retrieval. • Using phase-only correlation, all ROC curv...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Forensic science international 2017-06, Vol.275, p.102-109
Hauptverfasser: Richetelli, Nicole, Lee, Mackenzie C, Lasky, Carleen A, Gump, Madison E, Speir, Jacqueline A
Format: Artikel
Sprache:eng
Schlagworte:
Algorithms
Automation
Blood
Classification
Crime
Crime scenes
Databases, Factual
Dust
Fluorescent Dyes
Footwear
Forensic science
Forensic sciences
Forensic Sciences - methods
Fourier Analysis
Fourier transforms
Fourier–Mellin transform
Gentian Violet
Humans
Image Processing, Computer-Assisted
Image quality
Investigations
Mellin transforms
Noise
Pathology
Pattern recognition
Pattern Recognition, Automated
Phase only correlation
RANSAC
Shoes
Shoes & boots
SIFT
Stochasticity
Substrates
Surface Properties
Tiles
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Highlights • Query images vary in media, transfer mechanism, enhancement technique and substrate. • Automated retrieval methods were compared using CMC and ROC curves. • Of all methods considered, phase-only correlation showed superior database retrieval. • Using phase-only correlation, all ROC curve AUCs were above 0.80.
ISSN:0379-0738
1872-6283
DOI:10.1016/j.forsciint.2017.02.030