Globally optimal and region scalable CV model on automatic target segmentation

Authors:
Huijie Fan;Wei Dong;Yandong Tang
Affiliations:
State Key Laboratory of Robotics, Shenyang Institute of Automation, Chinese Academy of Science, Shenyang, China and Graduate University of the Chinese Academy of Science, Beijing, China;State Key Laboratory of Robotics, Shenyang Institute of Automation, Chinese Academy of Science, Shenyang, China;State Key Laboratory of Robotics, Shenyang Institute of Automation, Chinese Academy of Science, Shenyang, China
Venue:
ICIRA'10 Proceedings of the Third international conference on Intelligent robotics and applications - Volume Part I
Year:
2010

Citing 8
Cited 0

Fronts propagating with curvature-dependent speed: algorithms based on Hamilton-Jacobi formulations

Journal of Computational Physics
Scale-Space and Edge Detection Using Anisotropic Diffusion

IEEE Transactions on Pattern Analysis and Machine Intelligence
Region-based strategies for active contour models

International Journal of Computer Vision
A Multiphase Level Set Framework for Image Segmentation Using the Mumford and Shah Model

International Journal of Computer Vision
Brain MR Image Segmentation Using Local and Global Intensity Fitting Active Contours/Surfaces

MICCAI '08 Proceedings of the 11th international conference on Medical Image Computing and Computer-Assisted Intervention - Part I
Snakes, shapes, and gradient vector flow

IEEE Transactions on Image Processing
Active contours without edges

IEEE Transactions on Image Processing
Minimization of Region-Scalable Fitting Energy for Image Segmentation

IEEE Transactions on Image Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

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.