Precise Foreground Detection Algorithm Using Motion Estimation, Minima and Maxima Inside the Foreground Object

In this paper the precise foreground mask is obtained
in a complex environment by applying simple and effective
methods on a video sequence consisting of multi-colour
and multiple foreground object environment. To detect moving
objects we use a simple algorithm based on block-based motion
estimation, which requires less computational time. To obtain a
full and improved mask of the moving object, we use an openingand-
closing-by-reconstruction mechanism to identify the minima
and maxima inside the foreground object by applying a set of
morphological operations. This further enhances the outlines
of foreground objects at various stages of image processing.
Therefore, the algorithm does not require the knowledge of the
background image. That is why it can be used in real world video
sequences to detect the foreground in cases where we do not have
a background model in advance. The comparative performance
results demonstrate the effectiveness of the proposed algorithm.