Moving object area detection using normalized self adaptive optical flow

Optical flow estimation is one of the oldest and still most active research domains in computer vision. This paper proposes a novel and efficient method of moving object area detection in the video sequence employing the normalized self-adaptive optical flow. This new approach first performs smoothi...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Optik (Stuttgart) Jg. 127; H. 16; S. 6258 - 6267
Hauptverfasser: Sengar, Sandeep Singh, Mukhopadhyay, Susanta
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Elsevier GmbH 01.08.2016
Schlagworte:
Adaptive algorithms
Adaptive threshold
Algorithms
Computer vision
Frames
Gaussian
Gaussian filter
Motion detection
Noise threshold
Normalization
Optical flow
Smoothing
Video data
Motion detection
Normalization
Adaptive threshold
Gaussian filter
Optical flow
ISSN:0030-4026, 1618-1336
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Optical flow estimation is one of the oldest and still most active research domains in computer vision. This paper proposes a novel and efficient method of moving object area detection in the video sequence employing the normalized self-adaptive optical flow. This new approach first performs smoothing on the individual frame of the video data using Gaussian filter, then determines the optical flow field with an existing optical flow algorithm, next filters out the noise using adaptive threshold approach, after that normalize, morphology operation, and the self adaptive window approach is applied to identify the moving object areas. The proposed work is accurate for detecting the moving object areas with varying object size. The proposed scheme has been formulated, implemented and tested on real video data sets that provides an effective and efficient way in a complex background environment.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:0030-4026
1618-1336
DOI:10.1016/j.ijleo.2016.03.061