Purchase this article with an account.
S Chaudhry, GA Geri, BJ Pierce, MD Winterbottom; Visual Cues to Altitude in Simulated Flight Over Complex Terrain . Invest. Ophthalmol. Vis. Sci. 2002;43(13):4793.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: Accurate maintenance of altitude during simulated flight is known to be dependent on both terrain texture density and texture density gradients related to terrain elevation. Although commonly identified cues such as optic flow and familiar object size have been shown to be sufficient for maintaining altitude, the necessary cues have not been adequately determined. Methods: Observers attempted to actively maintain their initial altitude during simulated flight over terrains that had been textured with each of three densities of full gray-scale, random-dot stimuli. The densities of the input textures corresponded to 2.4 10-4, 1, and 4096 texture elements/m2. As displayed, the spectral context of the textures approximated that of natural terrain. The terrains were constructed from height maps representing a series of hills that extended about 18000 meters. Flight at 232 m/sec and an initial altitude of 30 m was simulated using a high-performance PC-based image generator (MetaVR Inc.). The terrain imagery was displayed across three-channels of a wide-field, rear-projection display (180º H 63º V). Results: An input texture density change of more than a factor of 4096 was required to significantly alter the RMS error for altitude maintenance. The RMS error varied from about 2.0 m to 4.0 m from the lowest to the highest terrain density. The RMS error was also consistent with the mean altitude deviation from the initial 30 m altitude. This deviation was also greater for the highest texture density. Thus, the highest density was consistently associated with the worst performance. Conclusion: Altitude maintenance in complex simulated environments can be accurately performed using relatively sparse visual cues that are not directly related to optic flow, edge rate, or familiar size. Although we have not completely isolated and identified these cues, they appear to be related to linear perspective and its interaction with terrain height and terrain texture gradients. A practical implication of the finding of lowered performance with the highest texture density is that visual cues may be compromised even by high-fidelity, wide-field display systems whose spatial and temporal resolution may actually decline when texture density is increased.
This PDF is available to Subscribers Only