Statistical-based edge detector.

Sophie Viseur. ( 2011 )
in: Proc. 31st Gocad Meeting, Nancy

Abstract

Edge detection algorithms are extensively used in computer vision for feature extraction. These algorithms aim at identifying points or lines in a digital image, at which the brightness or the color changes sharply. A wide range of edge detection algorithms have been proposed in litterature. These algorithms all rely on the computation of derivatives (of first or second degrees) and are then sensitive to noise, which commonly leads to use smoothing pre-processes. In this paper, a completely different approach for edge detections is proposed and can be applied on different supports (images, triangulated surfaces). It relies on a particular statistical parameter, the skewness, which characterizes the asymmetric aspect of a probability density function (pdf). This proposed technique has particular properties that are adressed in this paper and results are shown and discussed.

Download / Links

BibTeX Reference

@inproceedings{Viseur2GM2011,
 abstract = { Edge detection algorithms are extensively used in computer vision for feature extraction. These algorithms aim at identifying points or lines in a digital image, at which the brightness or the color changes sharply. A wide range of edge detection algorithms have been proposed in litterature. These algorithms all rely on the computation of derivatives (of first or second degrees) and are then sensitive to noise, which commonly leads to use smoothing pre-processes.
In this paper, a completely different approach for edge detections is proposed and can be applied on different supports (images, triangulated surfaces). It relies on a particular statistical parameter, the skewness, which characterizes the asymmetric aspect of a probability density function (pdf).
This proposed technique has particular properties that are adressed in this paper and results are shown and discussed. },
 author = { Viseur, Sophie },
 booktitle = { Proc. 31st Gocad Meeting, Nancy },
 title = { Statistical-based edge detector. },
 year = { 2011 }
}