Pyroelectric sensors and classification algorithms for border/perimeter security
Abstract
It has been shown that useful classifications can be made with a sensor that detects the shape of moving objects. This type of sensor has been referred to as a profiling sensor. In this research, two configurations of pyroelectric detectors are considered for use in a profiling sensor, a linear array and a circular array. The linear array produces crude images representing the shape of objects moving through the field of view. The circular array produces a temporal motion vector. A simulation of the output of each detector configuration is created and used to generate simulated profiles. The simulation is performed by convolving the pyroelectric detector response with images derived from calibrated thermal infrared video sequences. Profiles derived from these simulations are then used to train and test classification algorithms. Classification algorithms examined in this study include a naive Bayesian (NB) classifier and Linear discriminant analysis (LDA). Each classification algorithm assumes a three class problem where profiles are classified as either human, animal, or vehicle. Simulation results indicate that these systems can reliably classify outputs from these types of sensors. These types of sensors can be used in applications involving border or perimeter security © 2009 SPIE.
Publication Title
Proceedings of SPIE - The International Society for Optical Engineering
Recommended Citation
Jacobs, E., Chari, S., Halford, C., & Mcclellan, H. (2009). Pyroelectric sensors and classification algorithms for border/perimeter security. Proceedings of SPIE - The International Society for Optical Engineering https://doi.org/10.1117/12.830578
