Cosegmentation Revisited: Models and Optimization

Abstract

The problem of cosegmentation consists of segmenting the same object (or objects of the same class) in two or more distinct images. Recently a number of different models have been proposed for this problem. However, no comparison of such models and corresponding optimization techniques has been done so far. We analyze three existing models: the L1 norm model of Rother et al. [1], the L2 norm model of Mukherjee et al. [2] and the “reward” model of Hochbaum and Singh [3]. We also study a new model, which is a straightforward extension of the Boykov-Jolly model for single image segmentation [4]. In terms of optimization, we use a Dual Decomposition (DD) technique in addition to optimization methods in [1,2]. Experiments show a significant improvement of DD over published methods. Our main conclusion, however, is that the new model is the best overall because it: (i) has fewest parameters; (ii) is most robust in practice, and (iii) can be optimized well with an efficient EM-style procedure.