基于形态学和距离变换的根系图像交叉的分离

现今图像分割已然成为数字图像处理中必不可少的步骤之一，但是图像来源和目标形态的多样化使得至今还未有一种图像分割方法普遍适用于所有图像的分割。针对根系图像存在分支、根系交叉或重叠的现象，在现有算法的基础上，提出了一种基于形态学和距离变换相结合的分离方法。将形态学用于根系骨架交点膨胀，对根段图像利用改进的距离变换求其影响区，得到的影响区边界作为根系的分离线。实验证明和数据显示，该方法有效地解决了根系交叉和分支的分离问题，效果明显，效率较高。

Roots image intersection separation based on morphology and distance transfor- mation

Image segmentation has now become one of the indispensable steps in digital image processing. Due to various?image sources and target forms, no one method is suitable for all images. In order to separate the branches and crossing and overlapping roots, this paper puts forward a method based on the combination of morphology and distance transformation. Morphology is used to dilate intersections of roots skeleton and get root segments. Improved distance transformation is used to get influence areas and the boundary is considered as the separation lines. Experimental results demonstrate that proposed approach is able to resolve the separation of crossed roots and branches of root with obvious separation results and reasonable computational complexity.