基于扩展邻域的动态方向GVF计算方法

采用基于动态方向梯度矢量流(GVF)分割边界不清晰、灰度分布不均图像时,GVF外力场中通过拉普拉斯算子的扩散作用,利用4邻域进行掩膜运算;但未充分利用图像信息形成光滑的矢量场,同时迭代次数较高。提出一种基于扩展邻域的动态方向GVF计算方法。新方法从掩膜的角度分析了梯度矢量流模型,通过采用更大的邻域的掩膜来代替原来的基于4邻域的掩膜,得到了基于扩展邻域的梯度矢量流的计算方法。新的计算方法仅仅需要较少的迭代次数就能得到较好的效果。

Dynamic Directional Gradient Vector Flow based on Extended Neighbors

Gradient vector flow(GVF) is an effective external force for active contours which has a large capture range, when it is used in segmentation of image , especially focuses on boundary is not clear, gray uneven distribution of images. The diffusion of GVF is based on the Laplacian operator which is only based on 4 neighbors. In this paper, gradient vector flow based on extended neighbors is proposed by using a large mask to replace the original one which is only based on 4 neighbors. The new method would get a better result and only needs less iteration.