摘要
提出基于力场分析的图像分割方法,为解决参数轮廓模型处理多对象的图像分割问题。首先分析梯度矢量流力场生成机理与力场的特点,建立真实轮廓处力场(零力场)充分条件,依据零力场充分条件生成离真实轮廓处3-5像素距离的初始轮廓线,逼近得到多对象的目标轮廓线;再引入扩散梯度矢量流力场,简化零力场识别与连接过程,降低时间复杂度。试验结果表明,方法对多目标图像分割可得到精确的结果,可以提高轮廓精确度与抗噪性,降低时间复杂度,用户可自主选择图像中感兴趣的对象,在医学图像处理、遥感测绘领域有一定应用价值。
The paper proposes an image segmentation method based on field analysis to solve the problem of multi - objective image segmentation. First of all, the generating mechanism and characteristic of gradient vector flow field are analyzed and the sufficient conditions of GVF in real contour position (zero -field) are constructed. Initial contours which keep 3 -5 pixels away from real contour are obtained automatically due to sufficient conditions and approach target contour. Then extended gradient vector flow is brought to simplify zero - field identification and connection process and to reduce the time complexity. The simulation results show good performance of multi - objective image segmentation. The suggested method improves the accuracy of extraction and the insensitiveness to noise, reduces the time complexity and allows users to choose the interesting objects autonomously and has application value in medical images process and remote sensing images fields.
出处
《计算机仿真》
CSCD
北大核心
2010年第3期206-210,共5页
Computer Simulation
关键词
扩散梯度矢量流
多目标
图像分割
梯度矢量流
Extended gradient vector flow
Multi -objective
Image segmentation
Gradient vector flow