Detecting Salient Objects in an Image using Fast Watershed Method

Abstract

The Fast watershed method is an innovative approach that detects salient objects in an image. This transformation is entirely different from traditional watershed as it doesn’t depend on mathematical morphology. The flooding of water in the image is efficiently simulated using the FIFO queue of pixels. It starts with sorting image pixel according to their intensity levels and stores them in their corresponding FIFO structure. This can be implemented using chain code algorithm. In addition it is faster than any other watershed algorithm. Integrating this fast watershed with energy based segmentation leads to a new segmentation method called fast water snakes.