Image segmentation: Difference between revisions

Content deleted Content added
No edit summary
Line 9:
Region Growing: a recursive alghorithm, starting from a seed point, each neighbour pixel (or voxel, for 3D-data) is investigated and if it meets a homogenity-criteria, it is added to a list, then the neighbours of the next pixel on the list are investigated, etc.
 
Watershed segmentation is an example of a region growing algorithm, which has its origin in mathematical morphology. One of its draw-backs is oversegmentation. In any case, regions usually need merging to produce a desired outcome. A region-adjacency graph (RAG) with weighted links can be used to merge regions.
Region Merging: quad-tree/oct-tree: a recursive alghorithm: the picture is divided into 4/8 parts, if the resulting subpicture doesn´t meet a homogenity criteria, it is further divided, etc. The resulting data structure is a [[Quadtree]]/[[Octree]], from which the merging process can be performed.
 
Region Merging: quad-tree/oct-tree: a recursive alghorithm: the picture is divided into 4/8 parts, if the resulting subpicture doesn´t meet a homogenity criteria, it is further divided, etc. The resulting data structure is a [[Quadtree]]/[[Octree]], from which the merging process can be performed.
 
== Model based Segmentation ==