Summary: Internal Report 3/97, Albert­Ludwigs­Universit¨ at Freiburg, IIF­LMB, Germany, December 1997
Local Invariant Feature Histograms
for Texture Classification
S. Siggelkow, H. Burkhardt
Albert­Ludwigs­Universit¨ at Freiburg
Institut f¨ ur Informatik
79085 Freiburg i. Br., Germany
sven.siggelkow@informatik.uni­freiburg.de \Lambda
Abstract. This paper presents a method for texture classification based on invari­
ant gray scale features. These features remain constant if the images are transformed
according to the action of a transformation group. The basic method applied for ex­
tracting invariant features, is given by an integration over the transformation group.
For the transformation group of planar or Euclidean motion (translation and rota­
tion) one can show, that the integration can be split into two parts: The first is the
evaluation of a nonlinear local function for every pixel of the image, and the second
the summing of the results of these local computations. Instead of the second step
we calculate a histogram of the local computations which preserves the invariance
property and is more robust to real texture deviations than a single feature. Fur­
thermore in a multidimensional histogram approach the combination of different
features can be performed, thus increasing the discrimination power.

Source: Albert-Ludwigs-Universität Freiburg, Institut für Informatik,, Lehrstuhls für Mustererkennung und Bildverarbeitung

Collections: Computer Technologies and Information Sciences