Title :
Structured patterns from random fields
Author :
Picard, Rosalind W.
Author_Institution :
MIT Media Lab., Cambridge, MA, USA
Abstract :
The problem of modeling structured textures and image data with random field models is discussed. Random field models have recently been applied to a variety of problems such as image enhancement, segmentation, and coding. The models have been shown to successfully reproduce a variety of stochastic patterns, but have not been successful for capturing structure in patterns. Two possible solutions to the structured pattern problem are examined. The first solution is based on the structure present in the initial state of the random process, and the second is based on a relationship between internal random field interactions and coupling to a set of external constraints
Keywords :
image coding; image segmentation; image texture; coding; external constraints; image data; image enhancement; internal random field interactions; random field models; random process; segmentation; stochastic patterns; structured textures; Discrete cosine transforms; Higher order statistics; Image coding; Image enhancement; Image reconstruction; Laboratories; Predictive models; Random media; Random processes; Stochastic processes;
Conference_Titel :
Signals, Systems and Computers, 1992. 1992 Conference Record of The Twenty-Sixth Asilomar Conference on
Conference_Location :
Pacific Grove, CA
Print_ISBN :
0-8186-3160-0
DOI :
10.1109/ACSSC.1992.269068
