Using optimal circular path method to match piecewise iris templates

Majority of methods of iris recognition use templates containing localized texture features. Thus perfect alignment of templates plays crucial role in the work of recognition system. Known state-of-the-art alignment techniques are quite intricate. We present simpler and still efficient method. It is proposed to split the normalized iris image into several segments, calculate distances between possibly matching segments, represent these distances in form of a raster and search for the optimal path. The coordinates of this path's elements give displacements of individual segments. The method was tested with public domain images of ICE2005 and CASIA databases. Its performance is comparable or superior to existing algorithms.