Abstract
To make vision possible, the visual nervous system must represent the most informative features in the light pattern captured by the eye. Here we use Gaussian scale-space theory to derive a multiscale model for edge analysis and we test it in perceptual experiments. At all scales there are two stages of spatial filtering. An odd-symmetric, Gaussian first derivative filter provides the input to a Gaussian second derivative filter. Crucially, the output at each stage is half-wave rectified before feeding forward to the next. This creates nonlinear channels selectively responsive to one edge polarity while suppressing spurious or "phantom" edges. The two stages have properties analogous to simple and complex cells in the visual cortex. Edges are found as peaks in a scale-space response map that is the output of the second stage. The position and scale of the peak response identify the location and blur of the edge. The model predicts remarkably accurately our results on human perception of edge location and blur for a wide range of luminance profiles, including the surprising finding that blurred edges look sharper when their length is made shorter. The model enhances our understanding of early vision by integrating computational, physiological, and psychophysical approaches. © ARVO.
Original language | English |
---|---|
Article number | 7 |
Pages (from-to) | 1-21 |
Number of pages | 21 |
Journal | Journal of Vision |
Volume | 7 |
Issue number | 13 |
DOIs | |
Publication status | Published - 19 Oct 2007 |
Fingerprint
Bibliographical note
Creative Commons Attribution Non-Commercial No Derivatives LicenseKeywords
- blur
- edges
- feature analysis
- Gaussian derivatives
- half-wave rectification
- human vision
- scale-space theory
- spatial filters
- visual cortex
Cite this
}
From filters to features : Scale-space analysis of edge and blur coding in human vision. / Georgeson, Mark A.; May, Keith A.; Freeman, Tom C. A.; Hesse, Gillian S.
In: Journal of Vision, Vol. 7, No. 13, 7, 19.10.2007, p. 1-21.Research output: Contribution to journal › Article
TY - JOUR
T1 - From filters to features
T2 - Scale-space analysis of edge and blur coding in human vision
AU - Georgeson, Mark A.
AU - May, Keith A.
AU - Freeman, Tom C. A.
AU - Hesse, Gillian S.
N1 - Creative Commons Attribution Non-Commercial No Derivatives License
PY - 2007/10/19
Y1 - 2007/10/19
N2 - To make vision possible, the visual nervous system must represent the most informative features in the light pattern captured by the eye. Here we use Gaussian scale-space theory to derive a multiscale model for edge analysis and we test it in perceptual experiments. At all scales there are two stages of spatial filtering. An odd-symmetric, Gaussian first derivative filter provides the input to a Gaussian second derivative filter. Crucially, the output at each stage is half-wave rectified before feeding forward to the next. This creates nonlinear channels selectively responsive to one edge polarity while suppressing spurious or "phantom" edges. The two stages have properties analogous to simple and complex cells in the visual cortex. Edges are found as peaks in a scale-space response map that is the output of the second stage. The position and scale of the peak response identify the location and blur of the edge. The model predicts remarkably accurately our results on human perception of edge location and blur for a wide range of luminance profiles, including the surprising finding that blurred edges look sharper when their length is made shorter. The model enhances our understanding of early vision by integrating computational, physiological, and psychophysical approaches. © ARVO.
AB - To make vision possible, the visual nervous system must represent the most informative features in the light pattern captured by the eye. Here we use Gaussian scale-space theory to derive a multiscale model for edge analysis and we test it in perceptual experiments. At all scales there are two stages of spatial filtering. An odd-symmetric, Gaussian first derivative filter provides the input to a Gaussian second derivative filter. Crucially, the output at each stage is half-wave rectified before feeding forward to the next. This creates nonlinear channels selectively responsive to one edge polarity while suppressing spurious or "phantom" edges. The two stages have properties analogous to simple and complex cells in the visual cortex. Edges are found as peaks in a scale-space response map that is the output of the second stage. The position and scale of the peak response identify the location and blur of the edge. The model predicts remarkably accurately our results on human perception of edge location and blur for a wide range of luminance profiles, including the surprising finding that blurred edges look sharper when their length is made shorter. The model enhances our understanding of early vision by integrating computational, physiological, and psychophysical approaches. © ARVO.
KW - blur
KW - edges
KW - feature analysis
KW - Gaussian derivatives
KW - half-wave rectification
KW - human vision
KW - scale-space theory
KW - spatial filters
KW - visual cortex
UR - http://www.scopus.com/inward/record.url?scp=35448988583&partnerID=8YFLogxK
UR - http://www.journalofvision.org/content/7/13/7
U2 - 10.1167/7.13.7
DO - 10.1167/7.13.7
M3 - Article
C2 - 17997635
VL - 7
SP - 1
EP - 21
JO - Journal of Vision
JF - Journal of Vision
SN - 1534-7362
IS - 13
M1 - 7
ER -