Automatic Monitoring of Archaeological Sites

—This paper deals with the problem of the automatic monitoring of archaeological sites in order to prevent illegal acts such as theft. In particular, the paper presents a new methodology to understand human behavior through the temporal modeling of human posture. The proposed approach consists of three steps: the segmentation of a moving object by adaptive background subtraction, the estimation of posture of the detected moving people, and, finally, the temporal analysis of the detected postures by discrete hidden Markov models. Experimental results on image sequences acquired at a real archaeological site where actors perform legal and illegal actions will be shown.