LAUSR

In modern days, the importance of developing intelligent surveillance tools for activities of daily living is rising day by day. Identifying intruder(s) among the people who have access to a secured environment is crucial to achieve a flawless surveillance system in residential and office environments. This present work shows how a novel dictionary learning (DL) algorithm-based approach can be combined with four pyroelectric infrared (PIR) sensor-based indigenously developed hardware modules to design an intruder detection system. A novel DL approach is proposed combining the concepts of label consistency (LC) with a modified consistent adaptive sequential DL approach, named here as the LC-based modified consistent adaptive sequential DL (LC-MCAS-DL) algorithm. The conventional objective function in DL has been reformulated here by introducing the LC constraints along with reconstruction and classification errors. Then, the solution of this objective function is obtained by using a modified version of the consistent adaptive sequential DL algorithm (MCAS-DL). Extensive experiments have been performed to establish the suitability of our proposed approach for the problem under consideration.