作者: Stefan Glock , Eugen Gillich , Johannes Schaede , Volker Lohweg
DOI: 10.1007/978-3-642-03798-6_43
关键词:
摘要: Segmentation and feature extraction algorithms based on Wavelet Transform or Packet are established in pattern recognition. Especially the field of texture analysis they known to be practical. One difficulty was past characterization different printing processes. In this paper we present a new algorithmic concept textures, printed by techniques, without necessity previous teaching phase. The typical characters distinct textures extracted first order statistical moments wavelet coefficients. algorithm uses 2D incomplete shift invariant Transform, resulting fast execution time O(N log2 (N )). Since exclusively defined for 1D-signals, it has been modified research. application describes detection security textures.