Abstract
There is convincing evidence that the visual system acts as an optimal integrator when combining texture with disparity or motion to interpret 3D surface shape. Unfortunately, similar experiments with motion and disparity have proven challenging due to the presence of additional shape cues. Our experiments address these stimulus issues and evaluate how stereo and motion cues are combined to resolve 3D form. Three-dimensional cylinders were covered with a random-element greyscale texture. A black occluder with randomly positioned, 1.5 degree circular holes was placed in front of the display to limit observers' ability to track local features, or extract shape from texture. Using an implicit standard technique, with the method of constants, we assessed the accuracy and precision of observers' curvature discrimination judgments for a range of implicit reference curvatures (radii of 15, 16 and 17.50 deg). We did this first for motion and disparity alone, and in combination (equivalent and conflicting). In the combined conditions the relative strength of the disparity and motion cues was determined by selecting the 70% correct point from each individual psychometric function obtained in the single cue condition. Combined-equivalent results showed a marked increase in the slope of the psychometric function for all test curvatures. In the conflict conditions there were considerable individual differences in the weighting of the two cues. However, in all cases, there is support for cue integration rather than a vetoing process. These results and analyses will be discussed in the context of current models of cue integration.
This work was been supported by an NSERC grant to LMW