Bi-Local Region Based Iris Segmentation Framework for Less-Constrained Visible Wavelength Images

Iris recognition systems often degrade in performance when the subjects' cooperation is not expected in a less-constrained environment captured from a greater distance by visible wavelength imaging system. Algorithms in this field have reported inaccurate segmentations due to non-circular geome...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:7th International Conference on Imaging for Crime Detection and Prevention (ICDP 2016) s. 22
Hlavní autoři: Chai, T.Y, Goi, B.M, Tay, Y.H, Teng, K.S, Yeo, I
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: Stevenage, UK IET 2016
The Institution of Engineering & Technology
Témata:
Algorithms
Biometric recognition systems
Computer vision and image processing techniques
Contours
Image processing
Image recognition
Image segmentation
Interpolation and function approximation (numerical analysis)
Object recognition
Performance degradation
Shape
bilocal region based iris segmentation framework
NICE.II database
splines (mathematics)
less-constrained visible wavelength image
NICE.I database
noncooperative subject
image processing step
visible wavelength imaging system
image segmentation
B-spline formulation
iris recognition often system
active contour based method
contour evolution
computational power
iris noncircular geometry
local region based active contour model
iris recognition
ISBN:1785614002, 9781785614002
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Iris recognition systems often degrade in performance when the subjects' cooperation is not expected in a less-constrained environment captured from a greater distance by visible wavelength imaging system. Algorithms in this field have reported inaccurate segmentations due to non-circular geometry of the iris and various noise factors introduced by non-cooperative subjects. In this paper, we propose a local region based active contour model to segment non-circular iris shape from visible wavelength images which are normally noisy and inhomogeneous. Accurate segmentation can be achieved through the proposed bi-local neighborhood approach which allows contour evolution using local region based terms while simultaneously avoiding occlusions without the need of separate image processing steps. B-spline formulation in this approach ensures simplicity and efficiency of this algorithm, thereby overcoming the limitations of active contour based methods in terms of computational power. The proposed algorithm has demonstrated good segmentation accuracy on NICE.I and NICE.II databases.
Bibliografie:ObjectType-Article-1
ObjectType-Feature-2
SourceType-Conference Papers & Proceedings-1
content type line 22
ISBN:1785614002
9781785614002
DOI:10.1049/ic.2016.0090