Progressive image coding for visual surveillance applications based on statistical morphological skeleton

This paper presents a new shape representation method for progressive image coding at very-low bit rate. A real application in a railway surveillance system for unattended level-crossings is considered. First, semantic information, e.g., classification of possible obstacles provided by a recognition subsystem, is sent to a remote control center, then, binary shape information is transmitted, in order to allow the remote operator to validate the alarm situation. Pictorial information can be required as a further step by the operator of the control center.