IEEE Transactions on Pattern Analysis and Machine Intelligence

Perceptual Organization for Scene Segmentation and Description

View publication


A novel data-driven system for segmenting scenes into objects and their components is presented. This segmentation system generates hierarchies of features that correspond to structural elements such as boundaries and surfaces of objects. The technique is based on perceptual organization. In humans, perceptual organization is the ability to readily group elements in an image based on various realtionships between them. Here, perceptual organization is implemented as a mechanism to exploit geometrical regularities in the shapes of objects as projected onto images. Edges are recursively grouped on geometrical relationships into a description hierarchy ranging from edges to the visible surfaces of objects. These edge groupings, which are termed collated features, are abstract descriptors encoding structural information. The geometrical relationships employed are quasiinvariant over 2-D projections and are common to structures of most objects. Thus, collations have a high likelihood of corresponding to parts of objects. Collations serve as intermediate and high-level features for various visual processes. Applications of collations to stereo correspondence, object level semgentation, and shape description are illustrated. © 1992 IEEE