基于Snake模型的图像分割新算法

发布时间：2018-01-28 23:15

本文关键词： 图像分割 Snake模型梯度向量流边缘保护弱边缘　出处：《计算机应用》2017年12期 　论文类型：期刊论文

【摘要】：针对目前基于Snake模型的图像分割算法普遍存在噪声鲁棒性差、适用范围受限、易发生弱边缘泄露以及轮廓曲线难以收敛到细小深凹边界的缺陷,提出了一种基于Snake模型的图像分割新算法。首先,选取新的扩散项代替具有各向同性光滑作用的拉普拉斯算子;其次,引入p-拉普拉斯泛函到平滑能量项中强化法线方向外力;最后,利用边缘保护项使外力场方向与边缘方向一致,以防止弱边缘泄漏并促使轮廓线收敛到细小深凹边界。实验结果表明,所提模型不仅克服了现有基于Snake模型的图像分割算法的缺陷,具有更好的分割效果,明显提高了抗噪性能和角点定位精度,而且耗时更少,适用于噪声图像、医学图像以及含有很多弱边缘的自然图像分割。
[Abstract]:At present, image segmentation algorithms based on Snake model have the disadvantages of poor noise robustness, limited application range, weak edge leakage and the difficulty of contour curve converging to small deep concave boundaries. A new image segmentation algorithm based on Snake model is proposed. Firstly, a new diffusion term is selected to replace the Laplace operator with isotropic smoothness. Secondly, pLaplacian Functionals are introduced to the smooth energy term to strengthen the normal forces. Finally, the edge protection term is used to make the direction of the external force field consistent with the direction of the edge, so as to prevent the leakage of the weak edge and promote the contour to converge to the small deep concave boundary. The proposed model not only overcomes the defects of the existing image segmentation algorithms based on Snake model, but also has better segmentation effect, improves the anti-noise performance and the accuracy of corner location obviously, and takes less time. It is suitable for noise image, medical image and natural image segmentation with many weak edges.
【作者单位】：重庆邮电大学通信与信息工程学院;重庆邮电大学系统理论与应用研究中心;
【基金】：国家自然科学基金资助项目(61571017)~~
【分类号】：TP391.41
【正文快照】： 0引言图像分割是图像分析到图像理解的关键步骤[1]。活动轮廓模型是获取目标边界的重要工具之一,具有良好的封闭性与平滑性,被广泛应用于图像分割领域中。参数活动轮廓模型又称为Snake模型,自Kass等[2]提出以来,已广泛应用于数字图像分析和计算机视觉领域[3]。其基本原理是将

【相似文献】