Mathematical morphology : a case study of linear shift-invariant filter implementation

Graduation date: 1989Morphological filters have shown good performance in various image processing applications. Furthermore, their algorithms are easy to implement in special architectures, with the exception that linear filtering cannot be implemented in many of these cases. In this study, a new architecture and corresponding algorithm, capable of realizing linear shiftinvariant filters, as well as morphological filters, are presented, using the symmetrical nature of the most commonly used linear filtering masks. The architecture suggested in the study, along with the algorithm presented, will provide considerable improvement in the power of image processing systems, as well as improvements in the application fields of mathematical morphology