Completely Convex Formulation of the Chan-Vese Image Segmentation Model
The active contours without edges model of Chan and Vese (IEEE Transactions on Image Processing 10(2):266–277, 2001) is a popular method for computing the segmentation of an image into two phases, based on the piecewise constant Mumford-Shah model. The minimization problem is non-convex even when the optimal region constants are known a priori. In (SIAM Journal of Applied Mathematics 66(5):1632–1648, 2006), Chan, Esedoḡlu, and Nikolova provided a method to compute global minimizers by showing that solutions could be obtained from a convex relaxation. In this paper, we propose a convex relaxation approach to solve the case in which both the segmentation and the optimal constants are unknown for two phases and multiple phases. In other words, we propose a convex relaxation of the popular K-means algorithm. Our approach is based on the vector-valued relaxation technique developed by Goldstein et al. (UCLA CAM Report 09-77, 2009) and Brown et al. (UCLA CAM Report 10-43, 2010). The idea is to consider the optimal constants as functions subject to a constraint on their gradient. Although the proposed relaxation technique is not guaranteed to find exact global minimizers of the original problem, our experiments show that our method computes tight approximations of the optimal solutions. Particularly, we provide numerical examples in which our method finds better solutions than the method proposed by Chan et al. (SIAM Journal of Applied Mathematics 66(5):1632–1648, 2006), whose quality of solutions depends on the choice of the initial condition.
KeywordsImage segmentation Chan-Vese model Convex relaxation Level set method Vector-valued functional lifting K-means
Unable to display preview. Download preview PDF.
- Arrow, K., Hurwicz, L., & Uzawa, H. (1958). Stanford mathematical studies in the social sciences: Vol. II. Studies in linear and non-linear programming. With contributions by H. B. Chenery, S. M. Johnson, S. Karlin, T. Marschak, R. M. Solow. Stanford: Stanford University Press. Google Scholar
- Bae, E., Yuan, J., Tai, X.-C., & Boykov, Y. (2010). A study on continuous max-flow and min-cut approaches. Part II: multiple linearly ordered labels (UCLA CAM Report 10-62). Google Scholar
- Berkeley Dataset. http://www.eecs.berkeley.edu/CS/vision.
- Brown, E. S., Chan, T. F., & Bresson, X. (2009). Convex formulations for piecewise constant Mumford-Shah image segmentation (UCLA CAM Report 09-66). Google Scholar
- Brown, E. S., Chan, T. F., & Bresson, X. (2010). A convex relaxation method for a class of vector-valued minimization problems with applications to Mumford-Shah segmentation (UCLA CAM Report 10-43). Google Scholar
- Chambolle, A., Cremers, D., & Pock, T. (2008). A convex approach for computing minimal partitions (Technical report TR-2008-05). Bonn: Dept. of Computer Science, University of Bonn. Google Scholar
- Chambolle, A., & Pock, T. (2010). A first-order primal-dual algorithm for convex problems with applications to imaging (R.I. 685). CMAP, Ecole Polytechnique. Google Scholar
- El-Zehiry, N., Xu, S., Sahoo, P., & Elmaghraby, A. (2007). Graph cut optimization for the Mumford-Shah model. In IASTED international conference on visualization, imaging and image processing (pp. 182–187). Google Scholar
- Evans, L. C., & Gariepy, R. F. (2000). Measure theory and fine properties of functions. Boca Raton: CRC Press. Google Scholar
- Goldluecke, S., & Cremers, D. (2010). Convex relaxation for multilabel problems with product label spaces. In European conference on computer vision (pp. 225–238). Google Scholar
- Goldstein, T., Bresson, X., & Osher, S. (2009). Global minimization of Markov random fields with applications to optical flow (UCLA CAM Report 09-77). Google Scholar
- Lellmann, J., & Schnörr, C. (2010). Continuous multiclass labeling approaches and algorithms (Tech. Rep.). Heidelberg: University of Heidelberg. Google Scholar
- MacQueen, J. (1967). Some methods for classification and analysis of multivariate observations. In Proceedings of the fifth Berkeley symposium on mathematical statistics and probability (pp. 281–297). Google Scholar
- Pock, T., Chambolle, A., Bischof, H., & Cremers, D. (2009). An algorithm for minimizing the Mumford-Shah functional. In IEEE conference on computer vision (ICCV). Google Scholar
- Pock, T., Schoenemann, T., Graber, G., Bischof, H., & Cremers, D. (2008). A convex formulation of continuous multi-label problems. In European conference on computer vision (ECCV) (pp. 792–805). Google Scholar
- Weizmann Dataset. http://www.weizmann.ac.il/~vision.
- Yuan, J., Bae, E., Tai, X.-C., & Boykov, Y. (2010). A study on continuous max-flow and min-cut approaches. Part I: binary labeling (UCLA CAM Report 10-61). Google Scholar
- Zach, C., Gallup, D., Frahm, J. M., & Niethammer, M. (2008). Fast global labeling for real-time stereo using multiple plane sweeps. In Vision, modeling, and visualization (pp. 243–252). Google Scholar