On the simulation and development of massive parallel digital architectures for Markov random fields [image processing applications]

Markov random field modeling is a powerful parallel processing paradigm which can appropriately deal with the huge amount of data in the domain of low-level image processing problems. This paper describes a novel combined simulation and semiconductor-technology independent VLSI design environment for Markov random field based processing models and systems. The concepts of this novel combined simulation- and VLSI design-environment are experimentally demonstrated and proved by simulation results and detailed chip-layouts of a special Markov random field, which simultaneously solves the image processing problem of noise removing, intensity-level preserving and intensity histogram based segmentation.