Data Science
Hierarchical Image Segmentation - Glossary

Glossary

  • Hierarchical Boundary Map – a map of region boundaries for a segmented image wherein the region boundaries that continue to persist for coarser segmentations in the segmentation hierarchy are given a darker shade than those regions boundaries that only exist in the more finely detailed segmentations in the segmentation hierarchy.
  • Image Segmentation – the partitioning or separating of an image into related sections or regions
  • Merging – the combining of two or more objects, pixels or regions, by which a larger single object is created with a new boundary
  • Multiresolution Analysis – an analysis of an image at several different levels of spatial resolution
  • Pixel - the smallest discrete component of an image or picture on a CRT screen represented by a colored dot; the greater the number of pixels per inch the greater the resolution
  • Region – a group of pixels with similar or identical characteristics. A "connected region" is a region in which all pixels are connected through a path consisting only of pixels in the region. A "disjoint region" contains some pixels that are not connected through a path consisting only of pixels in the region.
  • Region Boundary Image – derived from the original image data, the region boundary image is a two toned image map that outlines the definitive region boundaries of the imagery data
  • Region Growing – method by which neighboring pixels or regions are merged together if the pixels or regions satisfy a specified criterion.
  • Region Mean Image – derived from the original image data, the region mean image is a processed image by which the regions are filled with the average color value that distinguishes that particular region
  • Scan – the processing of an image or image data; usually left to right, top to bottom
  • Segmentation Hierarchy – a set of several image segmentations of the same image at different levels of detail in which the segmentations at coarser levels of detail can be produced from simple merges of regions at finer levels of detail.
  • Spatial Structure – the form and attributes of an image that defines what the image represents
  • Threshold – the numerical value by which a formula is stop running; either defined by the user or by the programmer depending upon the application of the value