基于弹性交互作用和非线性扩散的水平集方法

Level set approach based on elastic interaction and nonlinear diffusion

首先介绍了一种基于弹性交互作用和非线性扩散的水平集分割方法。引入非线性扩散获得尺度概念,代替通常水平集演化方程中计算耗时的曲率计算,由基于物体边缘和移动曲线的弹性交互作用生成移动曲线的速度场,扩展了图像力的作用范围,不必预先人为选定初始轮廓,而是由速度场得到初始轮廓,并用水平集方法进行迭代计算,自动控制曲线演化过程中拓扑变化,可以很快得到物体轮廓。基于弹性交互作用的速度场和非线性扩散代替计算耗时的曲率计算,能够明显缩短算法的计算时间,并使用快速傅立叶变换(FFT)计算速度场,进一步提高计算效率。实验结果表明该方法有较好的分割效果和较高的分割效率。

This paper proposes a new edge-based level set algorithm based on nonlinear diffusion and elastic interaction.A length scale is obtained by using nonlinear diffusion instead of solving for the computationally expensive motion by mean curvature,the velocity field for the moving curves generated by this elastic interaction enlarges the capture range significantly,the initial value of the level set function is generated by velocity field.Level set method is used for the moving curves so that the topological changes during the evolution are handled automatically.The velocity field is calculated using the Fast Fourier Transform （FFT） method effficiently.Experiments prove the effectiveness of this algorithm and better segmentation result.