Automatic Peak Number Detection in Image Symmetry Analysis

Abstract

In repeated pattern analysis, peak number detection in autocorrelation is of key importance, which subsequently determines the correctness of the constructed lattice. Previous work inevitably needs users to select peak number manually, which limits its generalization to applications in large image database. The main contribution of this paper is to propose an optimization-based approach for automatic peak number detection, i.e., we first formulate it as an optimization problem by a straightforward yet effective criterion function, and then resort to Simulated Annealing to optimize it. Based on this approach, we design a new feature to depict image symmetry property which can be automatically extracted for repeated pattern retrieval. Experimental results demonstrate the effectiveness of the optimization approach and the superiority of symmetry feature over wavelet feature in discriminating similar repeated patterns.

This work was performed at Microsoft Research Asia.