基于直观分水岭定义的图像分割算法

由Vincent与Soille提出的基于沉浸模拟的分水岭算法(以下简写为V-S算法)已经成为图像分割领域中最主要的数学形态学方法。虽然V-S算法被认为是目前最快的分水岭算法,但它仍然不能满足一些实时应用对分割速度的要求。为此,在V-S算法的基础上提出了一种新的图像分割算法。新算法采用了一种新的泛洪方法,它利用了二维图像中各像素点间的规则空间关系信息计算满足基本直观定义的分水岭。实验结果表明:新算法的执行速度大大快于V-S算法,而分割效果与V-S算法相当。对新算法、V-S算法和两个分水岭定义之间的关系也作了讨论。

Image Segmentation Algorithm Based on Intuitive Watershed

The watershed algorithm by Vincent and Soille has emerged as the primary method of mathematical morphology for image segmentation.However,the computation load of this algorithm is still too large for some real-time applications.In this paper,a new fast algorithm is proposed based on Vincent and Soille's algorithm.By utilizing the regular spatial information existing in the image grid and a whole new flooding step,new algorithm computes a watershed in accord with the basic intuitive definition in a speed about ten times faster than that of the Vincent and Soille's algorithm.Experimental results show that this algorithm preserves most important information useful for segmentation in such a high speed.The relationship between new algorithm and the watershed definition is also discussed.