Watershed Algorithm
Also known as: Watershed Segmentation, Watershed Transform
An image segmentation technique inspired by geographical hydrology, where the gradient magnitude of an image is treated as a topographical surface. The algorithm simulates water flowing downhill from each pixel to local minima, forming catchment basins that define segmented regions, with watershed lines along the ridges between basins forming region boundaries. While effective at identifying detailed boundaries, watershed algorithms tend to produce oversegmentation (too many small regions) and typically require post-processing such as region merging to produce useful results. In accessibility contexts, watershed-based methods are used to simplify images for tactile conversion.
Category: Computer Vision · Image Processing · algorithms
Related: Image Segmentation · Tactile Imaging · Edge Detection