Baptiste Magnier presents Smoothing Rotating Filters,
On 2011-05-30 10:00
at G102A, Karlovo náměstí 13, Praha 2
Smoothing Rotating Filters
Baptiste Magnier
baptiste.magnier@mines-ales.fr
Ecole des mines d'Alès, LGI2P, Parc Scientifique G. Besse
30035 Nîmes Cedex 1
FRANCE
http://www.lgi2p.ema.fr/~magnier/
ABSTRACT
What type of segmentation can we do with anisotropic linear filtering? In this
talk, I will present you a new anisotropic Gaussian filter. The novelty of this
approach resides of ideas coming from directional recursive linear filtering. A
peculiarity of our anisotropic filter is that it is based on the use of two
elongated and oriented filters in two different directions.
Using these rotating kernels allows to :
- classify a pixel as a part of contour or region
- compute two different directions at each pixel used in a diffusion scheme
- compute roof edges
- extract edges even in noisy images
We have used this filter in order to remove the texture in images. Indeed,
applying this filter at each pixel of the image enables to classify if a pixel
belongs to a homogeneous region, a contour or a textured regions. Moreover, as
the filter is composed of two elongated and oriented filters in two different
directions, this directions are used in our anisotropic diffusion scheme.
A new work of ridge/valley (roof edges) extraction in images has been developed.
This work is based on the difference of rotating Gaussian semi-filters. The
novelty of this approach resides in mixing of ideas coming both from directional
filters and DoG method. We obtain a new ridge/valley anisotropic DoG detector
enabling very precise detection of ridge/valley points. Moreover, this detector
performs correctly at crest lines even if highly bended, and is precise on
junctions.
The derivative kernel into two parts allows to compute edges. We have developed
a new perceptual edge detector based on anisotropic linear filtering and local
maximization. The novelty of this approach resides in the mixing of ideas coming
both from perceptual grouping and directional recursive linear filtering. We
obtain new edge operators enabling very precise detection of edge points which
are involved in large structures.
LABORATORY
Created in 1994, LGI2P (Laboratory of Computer Engineering and Production
Engineering) develops and implements concepts, methods and tools to identify and
raise technical obstacles, technological and organizational facing industries
and services. The activity is part of LGI2P science and information technology
and communication. It focuses on the design, implementation, optimization of
technical systems, sociotechnical systems and piloting collaborative processes.
This targeted research results in many industrial and academic partnerships at
regional, national and European level.
The LGI2P is now organized into two research teams:
– ISOE (Interoperable System Organization and Engineering)
– KID (Knowledge Representation and Image Analysis for Decision)
http://www.lgi2p.ema.fr:8090/spip/
Baptiste Magnier
baptiste.magnier@mines-ales.fr
Ecole des mines d'Alès, LGI2P, Parc Scientifique G. Besse
30035 Nîmes Cedex 1
FRANCE
http://www.lgi2p.ema.fr/~magnier/
ABSTRACT
What type of segmentation can we do with anisotropic linear filtering? In this
talk, I will present you a new anisotropic Gaussian filter. The novelty of this
approach resides of ideas coming from directional recursive linear filtering. A
peculiarity of our anisotropic filter is that it is based on the use of two
elongated and oriented filters in two different directions.
Using these rotating kernels allows to :
- classify a pixel as a part of contour or region
- compute two different directions at each pixel used in a diffusion scheme
- compute roof edges
- extract edges even in noisy images
We have used this filter in order to remove the texture in images. Indeed,
applying this filter at each pixel of the image enables to classify if a pixel
belongs to a homogeneous region, a contour or a textured regions. Moreover, as
the filter is composed of two elongated and oriented filters in two different
directions, this directions are used in our anisotropic diffusion scheme.
A new work of ridge/valley (roof edges) extraction in images has been developed.
This work is based on the difference of rotating Gaussian semi-filters. The
novelty of this approach resides in mixing of ideas coming both from directional
filters and DoG method. We obtain a new ridge/valley anisotropic DoG detector
enabling very precise detection of ridge/valley points. Moreover, this detector
performs correctly at crest lines even if highly bended, and is precise on
junctions.
The derivative kernel into two parts allows to compute edges. We have developed
a new perceptual edge detector based on anisotropic linear filtering and local
maximization. The novelty of this approach resides in the mixing of ideas coming
both from perceptual grouping and directional recursive linear filtering. We
obtain new edge operators enabling very precise detection of edge points which
are involved in large structures.
LABORATORY
Created in 1994, LGI2P (Laboratory of Computer Engineering and Production
Engineering) develops and implements concepts, methods and tools to identify and
raise technical obstacles, technological and organizational facing industries
and services. The activity is part of LGI2P science and information technology
and communication. It focuses on the design, implementation, optimization of
technical systems, sociotechnical systems and piloting collaborative processes.
This targeted research results in many industrial and academic partnerships at
regional, national and European level.
The LGI2P is now organized into two research teams:
– ISOE (Interoperable System Organization and Engineering)
– KID (Knowledge Representation and Image Analysis for Decision)
http://www.lgi2p.ema.fr:8090/spip/