|Source:||Watkins, W.R.Clement, D.Reynolds, W.R., Target and Backgrounds VIII: Characterization and Representation, 1 April 2002 through 3 April 2002, Orlando,FL, 4718, 83 - 94|
|Proceedings of SPIE - The International Society for Optical Engineering|
Vision · Target detection · Conspicuity · Detection probability · Detection time · Field of regard · Field of view · Modeling · Scan path · Search performance · Target · Zooming · Cameras · Focusing · Image processing · Image quality · Mathematical models · Optoelectronic devices · Probability · Speed · Conspicuity · Detection probability · Detection time · Field of regard · Field of view · Scan path · Zooming · Scanning
In order to find an optimal scanning strategy we determined the relationship between search performance and several scanning- and scene parameters in human observer experiments. The observers searched with a limited field of view (FOV) through a large search sector for a target (a camouflaged person) on a heath. From trial to trial the target appeared at a different location. Predefined (horizontal) scan paths were used with constant speed. The subjects hit a button as soon as the target was spotted. We determined the effect of scanning speed, zoom factor, FOV width and target location on search performance (detection time and -probability). We also obtained estimates of target conspicuity for all targets and determined the relationship between conspicuity and search performance. We found that search performance was largely determined by the angular scanning speed of the simulated camera, and largely independent of zoom factor. The results show that conspicuity can be used to predict search performance in search with a limited FOV (this was previously shown for unrestricted search). Target conspicuity decreased when the targets appeared closer to the horizon (and where therefore more distant). Such a systematic dependency between target conspicuity and target location can be (and is probably) used by the observers to optimize search performance. The results further show that a reduction in the FOV width affects search performance when the time that a target is visible falls below a certain value (in our case about 1 sec). These findings can be used in future models of search performance with a limited FOV.