Globally optimal and region scalable CV model on automatic target segmentation

CORC > 沈阳自动化研究所 > 中国科学院沈阳自动化研究所 > 机器人学研究室

	Globally optimal and region scalable CV model on automatic target segmentation
	Fan HJ(范慧杰); Dong W(董威); Tang YD(唐延东)
	2010
会议名称	3rd International Conference on Intelligent Robotics and Applications, ICIRA 2010
会议日期	November 10-12, 2010
会议地点	Shanghai, China
关键词	Data handling Drop breakup Level measurement Numerical methods Optimization Robotics
页码	600-608
中文摘要	This paper proposes a new Globally Optimal and Region Scalable Chan-Vese model (GRCV model), which combines the advantages of both local and global intensity information. Intensity information in local regions is emphasized by adding a kernel function in the data fitting term, which thereby enables the proposed model to extract the fine structure in local region. The global information is also considered by adding one data fitting term in Piecewise Constant (PC) model to guarantee that proposed model can successfully segment complete target regions from intensity inhomogeneous images. We give the selection rules of combining the different terms under different target and background intensities: when the target is intensity inhomogeneous while the background is homogeneous, we choose the data fitting term which effects on the outside region of the evolution contour; conversely, we choose the other term. Experiments have been done on different images to compare the effectiveness of our methods with that of the classical PC model, Li's Local Binary Fitting (LBF) model, and Wang's Local and Global Intensity Fitting (LGIF) model. Comparison results show that our model obtains more satisfactory segmentation results. Moreover, it is robust to the curve initialization and noise in images. © 2010 Springer-Verlag.
收录类别	EI ; CPCI（ISTP）
产权排序	1
会议录	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
会议录出版者	Springer Verlag
会议录出版地	Heidelberg, Germany
语种	英语
ISSN号	0302-9743
ISBN号	978-3-642-16583-2
WOS记录号	WOS:000324576300058
内容类型	会议论文
源URL	[http://ir.sia.cn/handle/173321/8724]
专题	沈阳自动化研究所_机器人学研究室
推荐引用方式 GB/T 7714	Fan HJ,Dong W,Tang YD. Globally optimal and region scalable CV model on automatic target segmentation[C]. 见:3rd International Conference on Intelligent Robotics and Applications, ICIRA 2010. Shanghai, China. November 10-12, 2010.