医学图像轮廓跟踪的广义模糊粒子滤波方法

Active Contours Tracking of Medical Images Based on the Generalized Fuzzy Particle Filter

在医学图像运动跟踪领域 ,轮廓线跟踪是描绘边缘运动的有力手段 .为避免观测噪声的影响 ,增加轮廓的时空局部约束并利用粒子滤波 (PF)技术解决该类跟踪问题是非常有效的 .为更好地优化计算PF的重要比率 (IR)以提高粒子滤波器的性能 ,该文提出了广义模糊粒子滤波 (GFPF)方法 ,通过与当前较好的无迹粒子滤波 (UPF)相比较 ,GFPF显示了很好的效果 ;另外 ,在似然估计方面 ,GFPF提供了独特的似然轮廓估计算法 .理论和试验证明 ,GFPF不仅能够很好地解决动态轮廓跟踪问题 ,还为当前各种PF算法的IR计算提供了全新的解决途径 .

In the field of medical image visual tracking, Contour-based tracking methods have been proved to be a powerful tool for boundary delineation. During contour evolution, the particle filter (PF) is used to track the feature points by enforcing spatio-temporal local constraints to handle the observation noise. To improved the capability of PF and optimize its importance ratios (IR), the generalized fuzzy particle filter (GFPF) is presented in this paper. By comparing with the UPF which is a good method in object tracking, the GFPF shows great advantage. Another, with regard to likelihood estimation (LE), a special model of LE is constituted for contour estimation. By theoretic evaluation and sufficient contrast experiments, it is clear that the GFPF is a better measure for contours tracking and provide a novel resource for computing the IR of PF.