Proper scale for modeling visual data

摘要

We propose a method for determining the proper scale for modeling visual data. An efficient architecture for selective image modeling is discussed which selects models according to the task, the nature of the scene and the computational constraints. We give an example in which models of different scales are recovered in parallel and show that this redundant representation can effectively be pruned using the criterion of Minimal Description Length. Models that are selected in the final description indicate the appropriate scale of observation.