Share Email Print

Optical Engineering

IVF3: exploiting intensity variation function for high-performance pedestrian tracking in forward-looking infrared imagery
Author(s): Fabrizio Lamberti; Andrea Sanna; Gianluca Paravati; Luca Belluccini
Format Member Price Non-Member Price
PDF $20.00 $25.00

Paper Abstract

Tracking pedestrian targets in forward-looking infrared video sequences is a crucial component of a growing number of applications. At the same time, it is particularly challenging, since image resolution and signal-to-noise ratio are generally very low, while the nonrigidity of the human body produces highly variable target shapes. Moreover, motion can be quite chaotic with frequent target-to-target and target-to-scene occlusions. Hence, the trend is to design ever more sophisticated techniques, able to ensure rather accurate tracking results at the cost of a generally higher complexity. However, many of such techniques might not be suitable for real-time tracking in limited-resource environments. This work presents a technique that extends an extremely computationally efficient tracking method based on target intensity variation and template matching originally designed for targets with a marked and stable hot spot by adapting it to deal with much more complex thermal signatures and by removing the native dependency on configuration choices. Experimental tests demonstrated that, by working on multiple hot spots, the designed technique is able to achieve the robustness of other common approaches by limiting drifts and preserving the low-computational footprint of the reference method.

Paper Details

Date Published: 27 February 2014
PDF: 16 pages
Opt. Eng. 53(2) 023105 doi: 10.1117/1.OE.53.2.023105
Published in: Optical Engineering Volume 53, Issue 2
Show Author Affiliations
Fabrizio Lamberti, Politecnico di Torino (Italy)
Andrea Sanna, Politecnico di Torino (Italy)
Gianluca Paravati, Politecnico di Torino (Italy)
Luca Belluccini, Politecnico di Torino (Italy)

© SPIE. Terms of Use
Back to Top