Computation of Texture and Stereoscopic Depth in Humans
MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
Pagination or Media Count:
The computation of texture and of stereoscopic depth is limited by the eyes and by the subsequent stage of visual system in humans, and by the quality of the optical front end as well as by the computational hard- and software in machines. The quality of the optics and the resolution of the opto- electronic transducer e.g. the retina limit spatial resolution, and , consequently, the discrimination of textures. In stereoscopic depth, thresholds far below the grain of the input-device in humans the photoreceptor diameter can be attained. This extreme accuracy in locating a stimulus, called hyperacuity, is due to interpolation between the positions of the input elements, such as the photoreceptors in humans. Interpolation is most likely a feat achieved by the visual cortex, depending on a good signal-to-noise ratio of the stimulus representation. Again, resolution and contrast modulation are critical factors.