Reliable real-time foreground detection for video surveillance applications

Abstract

Foreground segmentation is usually needed as an initial step in video surveillance applications. Background subtraction is typically used to segment moving regions by comparing each new frame to a model of the scene background. We present a segmentation algorithm that works in real-time and efficiently extracts foreground objects from indoor and outdoor scenes that may contain small environment motions. The model adapts quickly to changes in the video which enables very sensitive detection of moving targets. The evaluation performed shows that this approach reliably extracts the foreground with very low false alarms and false misses.