Utilization of Pyramid Structure in Image Simulation

摘要

Multiresolution approach increases the speed of an image analysis algorithm by Reducing the volume of input data. Representation and processing of an image at multiple resolutions also provides convenient and efficient structures for the transition between the local and global features. In this research, a utilization of pyramid structure in MRF modeling had been explored and applied to the simulation of stochastic scenes. It would support efficient algorithms that initiate the simulation at a coarse resolution and then progressively refine the image to finer resolutions. Non-Markovian fields resulting from resolution transformation were approximated by Markov fields in the way that the parameters corresponding to the MRF approximation at coarse resolutions are related to the parameters at the fine resolution such that the interaction' between distant pixels is decimated. This can provide the useful methodology to generate stochastic scenes of large size.