一种结合图割与双水平集的图像分割方法

针对水平集方法在图像分割中需要多次迭代,且计算量大的问题,提出一种基于图割与双水平集的图像分割方法。首先在目标边界内外部各设置一条初始轮廓线和一个阈值,通过双水平集方法对轮廓线进行演化。当轮廓线的能量变化率小于给定阈值时,终止水平集演化。将得到的两条轮廓线化为源点和汇点,通过图割方法得到最终目标边界。该方法有效减少了水平集迭代次数,提高了分割效率,而且给出了一种终止水平集迭代的方式。实验表明该方法具有较好的分割效果和较高的分割效率。

Image segmentation approach based on graph cuts and dual level set method

Taking into account the problem that level set method for image segmentation requires more iterations and needs more computing time, a new method based on graph cuts and dual level set has been introduced. First of all, inside and outside of the boundary of the object set an initial contour separately and a threshold, and evolve the two contours by level set method. When the change rate of the contour is less than the given threshold, terminate the evolution. Then the two contours will be as source and sink, and obtained the final edge by graph cuts. This approach reduces the level set iterations, thus improves the efficiency for segmentation, and provides a way to end the level set iteration. Experiments show that the method has better effect and higher efficiency for segmentation.