Video motion processing for event detection and other applications
Richard Evans · 2003
This paper describes a widely applicable approach to extracting behavioural information from CCTV data with a generic video motion processor which tracks localised image features. One use of this is in VMAD, an anomaly-based video motion detector which has potentially several advantages over brightness-based activity detectors. A brief description of VMAD is presented together with a discussion of other potential applications of the video motion processor.