Study of spatiotemporal processing algorithms for video and image processing

The present paper discusses an important task of noise suppression in images and video encountered in many applications including hydroacoustics. We provide the review of spatiotemporal methods and algorithms of noise suppression. We suggest a new technique for the noise suppression problem based on...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Pattern recognition and image analysis Jg. 27; H. 4; S. 686 - 694
Hauptverfasser: Kaplun, D. I., Voznesenskiy, A. S., Klionskiy, D. M., Geppener, V. V., Serzhenko, F. L.
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Moscow Pleiades Publishing 01.10.2017
Springer Nature B.V
Schlagworte:
Algorithms
Analysis
Computer Science
Filtration
Image processing
Image Processing and Computer Vision
Noise
Parameter estimation
Pattern Recognition
Processing
Representation
Understanding of Images
image processing
spatiotemporal processing
noise suppression
movement detector
non-local means algorithm
video processing
ISSN:1054-6618, 1555-6212
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The present paper discusses an important task of noise suppression in images and video encountered in many applications including hydroacoustics. We provide the review of spatiotemporal methods and algorithms of noise suppression. We suggest a new technique for the noise suppression problem based on movement detector and the NLM algorithm (non-local means algorithm). The performance of the algorithms is considered using the filtering procedure of images and video. We also study filtering quality as a function of input parameters and we provide recommendations for selecting input parameters. The technique of estimating the filtering quality is suggested by combining such metrics as peak signal-to-noise ratio and the index of structural similarity.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1054-6618
1555-6212
DOI:10.1134/S1054661817040095