Purchase this article with an account.
James O'Shea, Maneesh Agrawala, Martin Banks; The influence of shape cues on detecting lighting inconsistencies. Journal of Vision 2010;10(7):73. doi: https://doi.org/10.1167/10.7.73.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Shape-from-shading is an ill-posed problem that can only be solved if observers assume or accurately estimate other scene properties such as the lighting. Recent psychophysical work demonstrates that observers often fail to detect lighting inconsistencies in an image, suggesting that the human visual system is often perceptually unaware of the direction of illumination in a scene (Ostrovsky 2005). We present a study that examines how shape cues influence this sensitivity to the direction of illumination. We conducted a visual search task in which we asked subjects to identify the location of an inconsistently illuminated target object among an array of distractor objects. We used a set of smooth, irregular 3D shapes as the objects, and we presented the stimuli for 1.0 second in each trial. We measured the detection threshold angle between the target light direction and the distractor light direction. To examine the influence of shape information, we varied the shape cues used to render the objects. In the cue-rich condition, we rendered the objects using binocular disparity, texture gradients, and shading. In the cue-impoverished condition, we provided only shading information. Five subjects completed the experiment, and we consistently found lower detection thresholds in the cue-rich condition compared to the cue-impoverished condition (average cue-rich threshold: 49.2deg +/− 3.2deg; average cue-impoverished threshold: 57.6deg +/− 6.8deg). These results demonstrate that an observer's sensitivity to lighting inconsistencies depends in part on the shape information available in the image.
This PDF is available to Subscribers Only