一种基于主动轮廓模型的图像分割新方法  

New Image Segmentation Method Based on Active Contour Model

作　　者：张弛 唐克伦 章华钎 潘书浩 ZHANG Chi;TANG Kelun;ZHANG Huaqian;PAN Shuhao(School of Mechanical Engineering,Sichuan University of Science and Engineering,Yibin 643002,China;School of Civil Engineering,Architecture and Environment Science,Xihua University,Chengdu 611730,China)

机构地区：[1]四川轻化工大学机械工程学院,四川宜宾643002 [2]西华大学土木建筑与环境科学学院,四川成都611730

出　　处：《成都大学学报（自然科学版）》2021年第1期48-51,共4页Journal of Chengdu University（Natural Science Edition）

基　　金：国家自然科学基金项目(52008340);西华大学校人才引进项目(Z201130)。

摘　　要：针对迭代收敛时主动轮廓模型在图像深凹处难以完全收敛且在较高的图像分辨率下收敛速度缓慢的缺陷,提出了一种基于主动轮廓模型的图像分割新方法.本方法结合了多分辨率与梯度向量流概念,首先在原始图像的低分辨率图像上进行轮廓提取,并将提取到的轮廓作为初始轮廓再在高分辨率图像上进行轮廓提取,最终得到原始图像的轮廓.实验结果表明,本方法在图像深凹处收敛效果良好,且极大提升了收敛速度.Aiming at the defect that active contour model is difficult to converge completely in the deep concave of image when it converges iteratively,and that the convergence speed is slow when the image resolution is high,a new image segmentation method based on active contour model is proposed.The method combines the idea of multi-resolution with the concept of GVF(Gradient Vector Flow).Firstly,the contour of the original image is extracted from the low resolution image,and then the extracted contour is used as the initial contour to extract the contour from the high-resolution image.Finally,the contour of the original image is obtained.The experimental results show that the method has good convergence effect on the image deep concave,greatly improves the convergence speed,and has high robustness to image noise.

关 键 词：主动轮廓模型 多分辨率 梯度向量流 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]