An improved watershed algorithm for image segmentation

Wenhong Wu¹, Hengmao Niu^{1, 2}

¹Inner Mongolia University of Technology, Inner Mongolia, 010051, China
²Inner Mongolia Technical College of Construction, Inner Mongolia, 010070, China

Watershed transform is a key operator in image segmentation algorithms. However, the computation load of watershed transform is too large for real-time applications. Previously published watershed segmentation algorithms required at least three global synchronization points: minima detection, labelling and flooding. This paper presented an algorithm of watershed transformation based on opening-closing operation and distance transform. It improved the classical watershed segmentation algorithm based on distance transform, overcoming over segmentation. The experiment result demonstrated that this method for segmentation inherits the advantage of watershed algorithm based on distance transform that it successfully segment out each dowel in the image bringing convenience to computer vision and auto-counting of dowels. It also overcame over-segmentation existed in traditional watershed segmentation preserving the original edges of each dowel in the image completely. This algorithm can be combined with any image segmentation algorithm to give more precise segmentation results. An example is also shown by combining a background registration and change-detection-based segmentation algorithm with Watershed. This new video segmentation algorithm can give accurate object masks with acceptable computation complexity.