基于各向异性扩散的图像分割算法

提出了一种基于各向异性扩散的图像分割算法.对现有的各向异性扩散的正则化方法进行了分析.根据微分几何中共形映射的有关理论,把原扩散方程分解为关于表面曲率的二阶方程,给出了分解式的正则化条件,保证了解的稳定性.通过对扩散系数的调节,提高了对各向异性扩散过程的控制能力.在形态学分割的基础上,通过能量函数最小化实现非线性尺度空间中的区域合并,消除了分水岭算法造成的严重过分割现象.实验结果表明,该算法的分割结果可为后续识别和理解提供较理想的方式.

Algorithm for Image Segmentation Based on Anisotropic Diffusion

A new algorithm for image segmentation based on anisotropic diffusion is proposed. The existing regularity method about anisotropic diffusion is introduced. The diffusion equation is decomposed as a second order equation about surface differential curvature according to the theory of conformal mapping in differential geometry. The regularity condition of the decomposed equation is discussed to guarantee a stable solution. Then, the diffusion coefficients in the decomposed equation are analyzed to get a further control on the anisotropic diffusion. An initial segmentation is generated with a watershed transform based on the mathematical morphology. The final segmentation is reached by eliminating the over-segmentation of initial regions. An energy minimum criterion is applied to guide the region merging in nonlinear scale space. The experiment shows that the proposed algorithm provides recognition and understanding processing with improved input.