BIOLOGICALLY INSPIRED ATTENTIVE MOTION ANALYSIS FOR VIDEO SURVEILLANCE

Recently proposed algorithms in the field of vision-based vi deo surveillance are build upon directionally consistent flow (Wixson and Hansen, 1999; Tian and Hampapur, 200 5), or statistics of foreground and background (Ren et al., 2003; Zhang et al., 2007). Here, we presen t a novel approach which utilizes an attention mechanism to focus processing on (highly) suspicious image re ions. The attention signal is generated through temporal integration of localized image features from mono cular image sequences. This approach incorporates biologically inspired mechanisms, for feature extra ction and spatio-temporal grouping. We compare our approach with an existing method for the task of video survei llance (Tian and Hampapur, 2005) with a receiver operator characteristic (ROC) analysis. In conclus ion our model is shown to yield results which are comparable with existing approaches.