摘要
提出了一种序列超声图像边缘检测与跟踪的离散动态模型.在基于动态规划的Snake模型中,引入目标边缘的形状信息改进了内部能量约束形式;充分利用序列图像相邻层间的灰度相关信息弥补了传统Snake模型中图像梯度能量对噪声敏感的不足.实验结果表明,改进的Snake模型仅需少量的用户交互就能快速、准确地从超声序列图像中检测、跟踪出目标物体的边缘,在医学图像分析中具有一定的实用价值.
A new discrete dynamic model for detecting and tracking boundaries in sequential 2D ultrasound images was presented. In the objective, a Snake is optimized by a dynamic programming algorithm, in which a new internal constraint was proposed and some correlative information between successive image slices were introduced into the external energy. The experimental results show that the new algorithm can obtain the boundary of the target object from a sequential ultrasound images quickly and reliably with only little user intervention. It has practical value in the medical image analysis.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2003年第z1期106-108,117,共4页
Journal of Shanghai Jiaotong University
关键词
SNAKE模型
动态规划
超声图像
边缘检测
相关性信息
Snake model
dynamic programming
ultrasound image
boundary detection
correlative information
