A Fuzzy-Statistical Contour Model for MRI Segmentation and Target Tracking

摘要

In this paper, a new formulation for the parametric active contour model is presented. The new formulation is based on statistical pattern recognition theory. A hybrid of kernel density estimation and fuzzy logic is used to show that active contours can be thought of as a pattern recognition problem. The proposed approach is used in two different application domains, with different performance requirements, to demonstrate its effectiveness. First, the proposed approach is used for a magnetic resonance image segmentation problem to demonstrate the segmentation accuracy. Second, the contour is used in a target tracking experiment to show its tracking capabilities.