Image Segmentation by Region Based and Watershed Algorithms

Allan Hanbury · Wiley Encyclopedia of Computer Science and Engineering · 2009

Abstract The segmentation of an image is defined as its partition into regions, in which the regions satisfy some specified criteria. First, two region‐based image segmentation techniques are described: region growing and split and merge. Second, the watershed transform, which combines aspects of both edge‐based and region‐based segmentation approaches, is described. In general, the basic watershed transform leads to an over‐segmentation of an image. Methods to reduce this over‐segmentation are discussed: preprocessing and filtering the input image, the marker‐based watershed algorithm, and hierarchical watershed algorithms.

Read the paper · More papers on PaperTik