一种隐式曲面多相图像分割的变分水平集模型

隐式曲面上的图像处理,与曲面的性状和特征息息相关,运用多个函数标记不同区域来进行图像分割计算量大。针对上述问题首先借助遥感图像提出了一种隐式曲面构建方式,利用图像中的高程数据来构建山体曲面模型,进而计算山体区域的曲面面积。其次将基于一个水平集函数的多相图像分割的模型推广到隐式曲面上,并设计了相应的交替方向乘子法,通过求解一个函数的极值实现对图像多个区域的分割,最后多个数值实验对该方法和模型的高效性和鲁棒性进行了验证。

A Variational Level Set Model for Implicit Surface Multiphase Image Segmentation

The image processing on the implicit surface is closely related to the traits and features of the surface.The use of multiple functions to mark different regions for image segmentation is computationally intensive.Aiming at the above problems,an implicit surface construction method is proposed by means of remote sensing image.The elevation data in the image were used to construct the mountain surface model,and then the surface area of the mountain region was calculated.Secondly,the model of multi-phase image segmentation based on a level set function was extended to the implicit surface,and the corresponding alternating direction multiplier method was designed.By solving the extremum of a function,the segmentation of multiple regions of the image was realized.Finally,the model was used.Numerical experiments verify the efficiency and robustness of the proposed method and model.