Motion segmentation is one of the first steps in image sequence applications. The problem has been extensively studied and approached in several ways but it remains a difficult task even for sequences acquired by stationary cameras. In this paper we propose a novel algorithm that is based on Mathematical Morphology. The algorithm is simple to be implemented, less subject to lighting changes and more robust to false positives. It has been tested in security cameras? image sequences showing promising results. Some algorithms from literature have been implemented and compared to the one described here.