摘要： 给出了图像层次分割的新架构，阐述了基于重构形态学操作的尺度空间产生方法，提出了基于形态学尺度空间和形态学分水岭的多尺度分割策略，其目的是生成图像的分割堆栈，并通过链接技术获得有意义的对象分割。该文给出了对各种图像的分割的实验结果，验证了分割策略的有效性

关键词: 图像分割；形态学尺度空间；形态学分水岭；形态学操作

Abstract: This paper presents a new framework for the hierarchical segmentation of various images. The scale-space is based on morphologicalanisotropic diffusion that uses reconstruction morphological operators. The proposed scheme for multiscale segmentation comprises a morphologicalscale-space and morphological gradient watersheds. Its aim is to produce steak of image segmentation and obtain a meaningful hierarchy among theobjects in the image by linking technique. The experimental results of the proposed scheme are obtained with a wide range of image includingnatural scenes. The result shows that the scheme is effective.

Key words: Image segmentation; Morphological scale space Morphological watersheds; Morphological operation