Abstract: To tackle the over-segmentation problem in watershed algorithm, a method based on dominant gray levels and constraint marker watershed was proposed to extract cingulum region. In this algorithm, original image was reconstructed by dominant gray levels, eliminating local minima and noise disturber. Then markers of regional minima were extracted from gradient image by using auto-threshold, and imposed by shape, area and centroid constraints. The watershed transform of the maker-modified gradient image was performed to achieve the cingulum region segmentation. Simulation results show that the new scheduling algorithm has the benefits of less over-segmentation areas, the more accurate segmentation result and higher flexibility compared to the traditional algorithms.

Key words: Harmful algae, Microscopic image segmentation, Watershed transform