ADAPT: A Parameter Optimised Segmentation Algorithm for Estimating Seeds

Seeded watersheds are very powerful for image segmentation for object detection in cell images to quantify single-cell features. Applied to cells, the results can be used for cell counting which is very important in medical diagnosis and treatment, and biological research. While a plethora of segmen- tation methods exists, accurate segmentation is challenging and usually requires problem-specific tailoring of algorithms. In this paper, we propose a supervised segmentation algorithm entitled ‘ADAPT’ having only two input parameters for estimating seeds. It is found that, the proposed method shows faster and accurate results compared to some state-art algorithms which are ’SMASEG’, ’SMOOTH’, ’GRADM’ and ’DUBALL’.