Completely Convex Formulation of the Chan-Vese Image Segmentation Model

被引:134
作者
Brown, Ethan S. [2 ]
Chan, Tony F. [3 ]
Bresson, Xavier [1 ]
机构
[1] City Univ Hong Kong, Dept Comp Sci, Hong Kong, Hong Kong, Peoples R China
[2] Univ Calif Los Angeles, Dept Math, Los Angeles, CA 90024 USA
[3] Hong Kong Univ Sci & Technol, Hong Kong, Hong Kong, Peoples R China
关键词
Image segmentation; Chan-Vese model; Convex relaxation; Level set method; Vector-valued functional lifting; K-means; OPTIMIZATION; ALGORITHMS;
D O I
10.1007/s11263-011-0499-y
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
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, Esedoalu, 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.
引用
收藏
页码:103 / 121
页数:19
相关论文
共 43 条
[1]   The calibration method for the Mumford-Shah functional [J].
Alberti, G ;
Bouchitté, G ;
Dal Maso, G .
COMPTES RENDUS DE L ACADEMIE DES SCIENCES SERIE I-MATHEMATIQUE, 1999, 329 (03) :249-254
[2]  
Ambrosio L., 2000, OX MATH M, pxviii, DOI 10.1017/S0024609301309281
[3]  
[Anonymous], 2008, Vision Modeling and Visualization
[4]  
[Anonymous], 1958, Stanford Mathematical Studies in the Social Sciences
[5]  
[Anonymous], 1960, Arch. Math. (Basel)
[6]  
[Anonymous], INT J COMPUT VIS
[7]  
[Anonymous], 1 ORDER PRIMAL DUAL
[8]  
[Anonymous], 1980, Mat. Zametki
[9]  
Bae E., 2010, UCLA CAM Report 10-62
[10]  
Bae E, 2009, LECT NOTES COMPUT SC, V5567, P1, DOI 10.1007/978-3-642-02256-2_1