On Improving the Performance of the Wallpaper Symmetry Group Classification Algorithm

Abstract

The wallpaper group classification algorithm published in the paper "A Computational Model for Periodic Pattern Perception Based on Frieze and Wallpaper Groups" (Y. Liu, R.T. Collins and Y. Tsin, IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 26, no. 3, pp. 354-371, March 2004) is the first computer algorithm for automatic frieze and wallpaper group classification for patterns in real images. In this report, we describe our effort on identifying the limitations of this wallpaper symmetry group classification algorithm. We document the improvements incorporated in the algorithm and code. This report also covers preliminary work done towards ideas for future research to further explore the space of symmetry or near-symmetry 2D patterns. This work is completed during the CRA Distributed Mentor Project summer research internship, June-August 2005, at the Robotics Institute, Carnegie Mellon University. Sravana Reddy is an undergraduate student from the computer science department of Brandeis University and the mentor Professor Yanxi Liu is an associate research professor of the Robotics Institute in Carnegie Mellon University.