Watershed segmentation is a region-based method that has its origins in mathematical morphology. In watershed segmentation an image is regarded as a topographic landscape with ridges and valleys. The elevation values of the landscape are typically defined by the gray values of the respective pixels or their gradient magnitude. Based on such a 3D representation the watershed transform decomposes an image into catchment basins. For each local minimum, a catchment basin comprises all points whose path of steepest descent terminates at this minimum. Watersheds separate basins from each other. The watershed transform decomposes an image completely and thus assigns each pixel either to a region or a watershed.
Discuss how low-pass filtering of an image impacts the resulting regions derived with watershed segmentation
New