| | |
Summary: Texture Defect Detection
Using Invariant Textural Features
Marc Schael
Institute for Pattern Recognition and Image Processing,
Computer Science Department,
Albert-Ludwigs-Universitat Freiburg, Germany
marc.schael@informatik.uni-freiburg.de
http://lmb.informatik.uni-freiburg.de
Abstract. In this paper we propose a novel method for the construction
of invariant textural features for grey scale images. The textural features
are based on an averaging over the 2D Euclidean transformation group
with relational kernels. They are invariant against 2D Euclidean motion
and strictly increasing grey scale transformations. Beside other elds of
texture analysis applications we consider texture defect detection here.
We provide a systematic method how to apply these grey scale features
to this task. This will include the localization and classi cation of the
defects. First experiments with real textile texture images taken from
the TILDA database show promising results. They are presented in this
paper.
1 Introduction
|
