An Information Theory framework for two-stage binary image operator design
Pattern Recognition Letters
Hi-index | 0.00 |
We address the problem of binary image operator design over large windows by breaking it into two phases. Firstly, we design several operators over small sub-windows of the main window. The outputs of these first level operators are then combined into a global operator. We devise a heuristic, motivated by Information Theory, for breaking the main window into sub-windows. Preliminary results show that the proposed scheme improves operator performance over single-level operators.