Abstract
Purpose. Perception of motion-defined form from apparent motion depends on the ability to detect and segregate regions of coherent motion. We investigated the effect of superimposed luminance noise on the ability to detect motion-defined form.
Methods. Stimuli consisted of randomly textured squares that subtended 8.6 degrees of visual angle. The image sequences depicted either a motion-defined square subtending 4.3 degrees (the ‘target’) or only the moving background. If the difference in the speed between the foreground dots and the background dots exceeded a certain threshold, the form of the foreground was visible. The images were rendered in Open GL at 100 Hz and displayed at 80% contrast.
Observers viewed the displays from 1.2 m with their head stabilized on a chin rest. For each trial, subjects were shown a pair of image sequences and required to indicate which sequence contained the target stimulus in a two-interval forced-choice procedure.
Poisson distributed spatiotemporal image noise was added to both the background and foreground of the display. At each of a variety of stimulus speeds (20.1, 50.4 100.7, 201.4, and 302.2 arc min/second), we measured detection threshold as a function of stimulus signal to noise ratio.
Results and discussion. All subjects could easily detect the motion-defined forms in the absence of any superimposed noise. As the power of spatiotemporal noise was increased, subjects had increased difficulty detecting the target. The influence of added noise was most pronounced at the lowest and highest image speeds. These results will be discussed in terms of models of motion processing and in the context of the usability of enhanced vision displays under noisy conditions.
The support of Government of Ontario (CRESTech) is gratefully acknowledged.