A Multiresolution Independent Component Analysis for textile images
Daniela Colţuc, Thierry Fournel, J.-M. Becker, Michel Jourlin · Journal of Physics Conference Series · 2007
This paper aims to provide an efficient tool for pattern recognition in the fight against counterfeiting in textile design. As fabrics patterns to be protected can present numerous and various characteristics related to intensity or color feature but also to texture and relative scales features, we introduce a tool able to separate image independent components at different resolutions. The suggested `Multiresolution ICA' combines the properties from both wavelet transform and Independent Component Analysis.