结合均值聚类的窄带水平集曲线演化

水平集方法是目前常用的一种图像分割方法,但它在构造速度函数时仅使用了图像的梯度信息,对于MRI这类含有强噪音、弱边界等现象的图像很难取得理想的分割结果.针对这一问题,将图像的区域信息和梯度信息相结合,构造新的基于K-均值聚类的水平集速度函数,该速度函数有较强的抗噪性能,并且能够处理含有弱边界、低对比度的图像.对左心室MR图像的分割实验表明该方法具有良好的分割效果.

CURVE EVOLVING OF NARROW BAND LEVEL SET METHOD BASED ON MEANS CLUSTERING

Level Set method is commonly used in the field of image segmentation recently.However the classic Level Set method only uses gradient information of image when constructing its velocity function.It is hard to get good segmentation result for Magnetic Resonance Imaging(MRI)which has strong noses and weak edges.To solve these problems,this paper integrates region information of image with gradient information,and constructs a new velocity function based on K-means clustering.This new velocity function has better antinose capability and can deal with images which have loud noises,weak edges and low contrast.The experiments on the tagged left ventricle Magnetic Resonance Images have shown the effectiveness of this method.