基于力场分析的多目标图像分割算法研究  

Multi-objective Image Segmentation Based on Field Analysis

作　　者：于磊[1] 范延滨[1] 

机构地区：[1]青岛大学信息工程学院,山东青岛266071

出　　处：《计算机仿真》2010年第3期206-210,共5页Computer Simulation

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.

关 键 词：扩散梯度矢量流 多目标 图像分割 梯度矢量流 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]