融合变分模型与快速算法分割噪声图像

A Novel Variational Model and Its Fast Algorithm for Noisy Image Segmentation

针对现有的变分水平集方法对噪声图像分割不理想和计算效率较低的情况,提出一种改进的可有效分割噪声图像的变分模型。首先改进了Chan-Vese模型的能量泛函,并引入辅助变量耦合某些拟合能量项,接着用凸松弛方法将其转化为凸优化问题。该优化问题可转化为几个子问题,在求解时结合快速的Split-Bregman算法和AOS算法以提高速度。对噪声图像作分割实验,并与不引入辅助变量的水平集方法作比较。结果表明,本文提出的变分模型对带某些类型噪声的图像分割不仅提高了计算效率,还能较好地分割目标。

A new variational model for segmenting the images corrupted by various noise is proposed. Firstly, the energy function of the presented model based on Chan-Vese model is modified, and a new auxiliary variable is introduced to integrate some fitting terms. Second- ly, it is extended to convex optimization by convex relaxation. And the solution is decom- posed into solving a few functional optimization sub-problems, which can be obtained by ap- plying Split-Bregman algorithm and additive operator splitting （AOS） numerical algorithm, the effective results are obtained. Comparing with the model in which a auxiliary variable is not introduced, experimental results verify that the proposed model for segmenting the noisy images reduce computational time and has better results.