Author Login Editor-in-Chief Peer Review Editor Work Office Work

Computer Engineering

Previous Articles     Next Articles

Interactive Object Segmentation Algorithm Based on Level Set Method

Pak Chun-hyok,ZHAO Hai,ZHU Hongbo,XU Jiuqiang   

  1. (College of Information Science and Engineering,Northeastern University,Shenyang 110819,China)
  • Received:2014-07-08 Online:2016-02-15 Published:2016-01-29

基于水平集方法的交互式目标分割算法

朴春赫,赵海,朱宏博,徐久强   

  1. (东北大学信息科学与工程学院,沈阳 110819)
  • 作者简介:朴春赫(1976-),男,博士研究生,主研方向为计算机视觉、图像处理;赵海,教授;朱宏博,博士研究生;徐久强,教授。
  • 基金资助:
    国家科技支撑计划基金资助项目(2012BAH82F04)。

Abstract: This paper focuses on an object segmentation algorithm based on the interactive information as the prior knowledge between user and system,and then proposes a level set object segmentation algorithm based on improved speed function.It improves speed function of Chan & Verse (CV) model by regional and contour information.The input contour is gradually close to the actual one by using gradient descent vector.Probability density model ensures all the pixels in the images approximate Gaussian probability distribution.To speed-up the convergence,it utilizes the boundary of convergence model.Experimental results show that this algorithm has the better segmentation performance and robustness than the segmentation algorithms based on CV model and Mean Square Error(MSE) level set for the images including noise and object edge blur.

Key words: image segmentation, interactive segmentation, level set, active contour model, curve evolution

摘要: 研究以用户与系统之间的交互信息为先验知识的目标分割算法,提出一种基于改进速度函数的水平集图像分割算法。利用待分割图像的区域信息与目标轮廓信息,改进传统CV模型中的速度函数。使用图像边缘的梯度下降向量逐步逼近目标边缘,采用概率密度函数使图像中的所有像素值近似满足高斯概率分布,并通过边界收敛模型保证速度函数快速收敛至目标边缘。实验结果表明,对于包含噪声和目标边缘模糊的图像,与基于CV模型的分割算法和基于均方差水平集方法的分割算法相比,该算法具有更好的分割性能和鲁棒性。

关键词: 图像分割, 交互式分割, 水平集, 主动轮廓模型, 曲线演化

CLC Number: