Separating non-stationary from stationary scene components in a sequence of real world TV-images
Ramesh Chandra Jain, D. Militzer, Hans-Hellmut Nagel · 1977
Results are presented for a new method to identify images of moving objects in a sequence of scene images, e.g. from a TV-camera observing a street intersection. The reported approach exploits the assumption that systematic greyvalue differences-based on second order statistics- between consecutive frames are due to images of moving objects. No knowledge is assumed about size, shape, or texture for images of stationary or non-stationary scene components. 1.