Fast Image Segmentation Using Watershed Transform
Tao Lei, Asoke Kumar Nandi · 2022
This chapter focuses on the watershed transform and power watershed, but they have a serious over-segmentation phenomenon. It presents a novel adaptive morphological reconstruction (AMR) operation. Morphological reconstruction (MR) is a powerful operation in mathematical morphology. MR is an image transformation that requires two input images, a marker image, and a mask image. It presents an AMR that is able to filter useless regional minima and maintain meaningful ones generated by salient objects. AMR is useful for improving seeded image segmentation because it employs multiscale structuring elements to obtain a convergent seed image without pre-setting many parameters. The property of monotonic increase helps AMR to achieve a hierarchical segmentation. The property of convergence is able to alleviate the drawback of MR by filtering out useless regional minima in a gradient image and guarantees a convergent result.