摘要
通过将梯度向量流(Gradient vector flow,GVF)场与种子区域生长(Seeded region growing,SRG)法相结合,提出了一种快速的自动图像分割方法.首先基于梯度向量流场构建一个流向标量场,然后提出一种新型的快速种子区域生长分割法—快速扫掠法(Fast scanning method,FSM)对标量场进行初始分割,最后采用区域邻接图对初始分割结果进行区域合并得到最终结果.本方法分割速度快,如采用一个快速的梯度向量流算法,则可以用于实时应用.实验结果证实了本方法的高效与鲁棒.
A fast automatic image segmentation method is presented through the combination of gradient vector flow （GVF） field and seeded region growing （SRG）. First, a flow direction scalar （FDS） field is derived from GVF. Then, a new SRG method called fast scanning method （FSM） is proposed. Finally, region adjacency graph （RAG） based region merging is used to get the final result. This method is very efficient that it is possible to fit real-time applications if a fast GVF computation method is adopted. Experiments demonstrate its effectiveness and robustness.
出处
《自动化学报》
EI
CSCD
北大核心
2008年第8期993-996,共4页
Acta Automatica Sinica
关键词
梯度向量流
种子区域生长
流向标量场
快速扫掠法
Gradient vector flow （GVF）, seeded region growing （SRG）, flow direction scalar field, fast scanning method （FSM）
