Local Region Descriptors for Active Contours Evolution

被引:56
作者
Darolti, Cristina [1 ]
Mertins, Alfred [1 ]
Bodensteiner, Christoph [2 ]
Hofmann, Ulrich G. [1 ]
机构
[1] Med Univ Lubeck, Inst Signal Proc, D-23538 Lubeck, Germany
[2] Med Univ Lubeck, Inst Robot, D-23538 Lubeck, Germany
关键词
Active contours; image segmentation; local statistics;
D O I
10.1109/TIP.2008.2006443
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Edge-based and region-based active contours are frequently used in image segmentation. While edges characterize small neighborhoods of pixels, region descriptors characterize entire image regions that may have overlapping probability densities. In this paper, we propose to characterize image regions locally by defining Local Region Descriptors (LRDs). These are essentially feature statistics from pixels located within windows centered on the evolving contour, and they may reduce the overlap between distributions. LRDs are used to define general-form energies based on level sets. In general, a particular energy is associated with an active contour by means of the logarithm of the probability density of features conditioned on the region. In order to reduce the number of local minima of such energies, we introduce two novel functions for constructing the energy functional which are both based on the assumption that local densities are approximately Gaussian. The first uses a similarity measure between features of pixels that involves confidence intervals. The second employs a local Markov Random Field (MRF) model. By minimizing the associated energies, we obtain active contours that can segment objects that have largely overlapping global probability densities. Our experiments show that the proposed method can accurately segment natural large images in very short time when using a fast level-set implementation.
引用
收藏
页码:2275 / 2288
页数:14
相关论文
共 45 条
[31]  
PARAGIOS N, 1999, IEEE COMP SOC C COMP
[32]  
PARAGIOS N, 1999, P IEEE COMP SOC C CO, V2, P2422
[33]  
Piovano J, 2007, LECT NOTES COMPUT SC, V4485, P709
[34]   Higher order active contours [J].
Rochery, Marie ;
Jermyn, Ian H. ;
Zerubia, Josiane .
INTERNATIONAL JOURNAL OF COMPUTER VISION, 2006, 69 (01) :27-42
[35]   REGION-BASED STRATEGIES FOR ACTIVE CONTOUR MODELS [J].
RONFARD, R .
INTERNATIONAL JOURNAL OF COMPUTER VISION, 1994, 13 (02) :229-251
[36]  
ROUSSON M, 2003, IEEE C COMP VIS PATT
[37]   Segmentation of a vector field:: Dominant parameter and shape optimization [J].
Roy, T ;
Debreuve, É ;
Barlaud, M ;
Aubert, G .
JOURNAL OF MATHEMATICAL IMAGING AND VISION, 2006, 24 (02) :259-276
[38]  
ROY T, 2003, WORKSH VAR GEOM LEV
[39]  
SAMSON C, 1999, P INT C SCAL SPAC TH, P306
[40]  
SETHIAN J. A., 1999, LEVEL SET METHODS FA