Self-driven and direct spatio-temporal mechanisms for the vision-based parking slot surveillance

A vision-based and straightforward parking space surveillance system is presented. In particular, the novelty of the approach lies in the use of self-driven and direct dissimilarity computing mechanisms based on image raw brightness. Notably, the parking occupancy is estimated in real-time by applying fast computed estimators that optimize a global dissimilarity measure over slot-centered parking images. Experimental results show the operational capabilities and the efficiency of the proposed system mechanisms.