Improved marker-controlled watershed segmentation with local boundary priors

Separating touching or overlapping objects in an image is one of the most challenging image processing operations. The marker-controlled watershed segmentation is often applied to resolve this problem and shows effectiveness and practicability. But the boundaries of the objects can sometimes be imprecise or completely wrong, which will significantly affect the subsequent image quantification. This paper describes a new technique that incorporates local boundaries of the touching or overlapping objects to improve the segmentation of marker-controlled watershed segmentation algorithm. Experimental results show that the proposed technique can significantly improve the separation of touching or overlapping objects and produce precise boundaries.