作者: Jaap A. Beintema , Piet Bijl , Maarten A. Hogervorst , Judith Dijk
DOI: 10.1117/12.777561
关键词: Signal processing 、 Target acquisition 、 Range (statistics) 、 Computer vision 、 Computer science 、 Orientation (computer vision) 、 Dynamic imaging 、 Mathematical model 、 Magnitude (mathematics) 、 Contrast (vision) 、 Artificial intelligence
摘要: In an extensive Target Acquisition (TA) performance study, we recorded static and dynamic imagery of a set military civilian two-handheld objects at range distances aspect angles with under-sampled uncooled thermal imager. Next, applied signal processing techniques including DSR (Dynamic Super Resolution) LACE (Local Adaptive Contrast Enhancement) to the imagery. perception experiment, determined identification (ID) threat/non-threat discrimination as function target for variety conditions. The experiment was performed validate extend current TA models. addition, predictions were two models: TOD model NVThermIP. results study are: i) orientation has strong effect on performance, ii) is well predicted by models, iii) absolute close models using recommended criteria objects, iv) there no positive sensor motion this against expectations based earlier studies, v) benefit smaller than expected basis predictions, vi) similar image optimized manually, indicating that can be used optimize contrast automatically. relatively poor are probably due smear induced higher camera speed in studies. Camera magnitude not yet implemented