A rough set-based approach to handling spatial uncertainty in binary images

In this paper we consider the problem of detecting binary objects using rough sets. We present a method for constructing a gray-scaled (or, fuzzy) template for use in correlation-based matching of Boolean images. We assume a cause for spatial uncertainty that is quite common in machine vision applications and present a methodology for modeling it indirectly in the construction of the template. Our technique is computationally efficient and is superior to correlation-based techniques, which can be easily fooled and automates the hand-selection of structuring elements for the hit-or-miss transform technique, both of which are usually used to solve this problem.