| The premise of this study is that accurate communication of spatial information via perspective imagery can only be accomplished by understanding and accounting for the relationships between perspective geometry and the interpretive biases of the observer. In the experiments reported here, computer graphics techniques were used to quantify the image and viewing geometries of stimulus images, which, when compared with performance data, allows a detailed quantitative understanding of these relationships.;Although use of pictorial spatial information instruments in aerospace applications has been of interest for over forty years, current instruments are very limited. Computer graphics offers the potential for a revolution in displays for situation awareness and flight path control, but understanding of perspective geometry and effective spatial information transfer is not evident in many proposed formats.;Correct construction and viewing of perspective imagery has been studied by artists since the Renaissance, and more recently by visual psychologists. The potential for perceptual distortions of pictorial space is widely recognized.;In the present study, subjects observed hundreds of computer generated perspective images in which the image and viewing geometries were systematically varied. They made spatial judgments concerning the direction of a target object relative to a reference object in the scene. It was found that subjects tend to make direction judgment errors whose magnitudes vary as functions of image perspective, viewing geometry, and target direction.;A model is proposed which incorporates the influence of the differences between the 3-dimensional scene and its 2-dimensional projection, the 2D Effect, as well as the consequences of the relative positions of the actual eyepoint and the geometric center of projection, the Virtual Space Effect. This model is shown to correctly predict the magnitude and direction of azimuth judgment errors.;My virtual space geometry differs from that of Farber and Rosinski. The results of flight experiments, in which perspective displays were viewed to land aircraft, lend support to my assertion that the virtual space effect has a lateral component, which contradicts Farber and Rosinski. Assumptions in the literature concerning the geometry of virtual space should be reconsidered. |
