A framework on fuzzy intrusion detection

Automated video surveillance systems have been a very popular exploration area since the last two decades when the demand of such systems becomes incredibly high. Real-time video capturing system normally comes along with uncertainties, vagueness, ambiguous, and ill-defined data. Difficult situations arise not only by the change of illumination conditions, but also by the large cast shadows of surrounding structures. In this paper, we proposed to use two cues: standard deviation and dissimilarity of histogram in a fuzzy inference system to distinguish between intrusion events and changes of illumination. Experiment results using four datasets and a comparison with the state-of-the-art solutions using distance measure have shown the effectiveness of our proposed framework.