26 April 2018 Validating models of target acquisition performance in the dismounted soldier context
Author Affiliations +
Abstract
The problem of predicting real-world operator performance with digital imaging devices is of great interest within the military and commercial domains. There are several approaches to this problem, including: field trials with imaging devices, laboratory experiments using imagery captured from these devices, and models that predict human performance based on imaging device parameters. The modeling approach is desirable, as both field trials and laboratory experiments are costly and time-consuming. However, the data from these experiments is required for model validation. Here we considered this problem in the context of dismounted soldiering, for which detection and identification of human targets are essential tasks. Human performance data were obtained for two-alternative detection and identification decisions in a laboratory experiment in which photographs of human targets were presented on a computer monitor and the images were digitally magnified to simulate range-to-target. We then compared the predictions of different performance models within the NV-IPM software package: Targeting Task Performance (TTP) metric model and the Johnson model. We also introduced a modification to the TTP metric computation that incorporates an additional correction for target angular size. We examined model predictions using NV-IPM default values for a critical model constant, V50, and we also considered predictions when this value was optimized to fit the behavioral data. When using default values, certain model versions produced a reasonably close fit to the human performance data in the detection task, while for the identification task all models substantially overestimated performance. When using fitted V50 values the models produced improved predictions, though the slopes of the performance functions were still shallow compared to the behavioral data. These findings are discussed in relation to the models’ designs and parameters, and the characteristics of the behavioral paradigm.
© (2018) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Mackenzie G. Glaholt, Mackenzie G. Glaholt, Rachel K. Wong, Rachel K. Wong, Justin G. Hollands, Justin G. Hollands, } "Validating models of target acquisition performance in the dismounted soldier context", Proc. SPIE 10625, Infrared Imaging Systems: Design, Analysis, Modeling, and Testing XXIX, 106250R (26 April 2018); doi: 10.1117/12.2304607; https://doi.org/10.1117/12.2304607
PROCEEDINGS
14 PAGES


SHARE
Back to Top