Neurophysiology and Human Vision



next up previous
Next: Image Processing and Up: Introduction Previous: On Painting and

Neurophysiology and Human Vision

In the early part of this century a series of psychophysical experiments to understand the nature of stereopsis, texture and the identification of objects began. At the same time beginning with the work of Ramon y Cajal there was a detailed neuro-physiological study of the circuitry associated with human vision starting from the rods and cones on the retina to the optic chiasm to the lateral geniculate body to the primary visual cortex. Hubel, Wiesel and later Marr postulated that there was processing at each level with low level vision consisting of activities such as image segmentation, edge enhancement happening at the retinal level, multiple filtering operations such as those required for edge detection and orientation at different scales possibly in the striate cortex (also known as V-1), mid level vision such as stereopsis, line drawing interpretation, motion flow, stereopsis happening in areas of the sensory motor cortex such as V-2, V-3 and high level vision such as object recognition, generation of commands for the movement of the eyeball (the so-called oculomotor system), and finally to send signals for the planning of movement of the body to the other areas of the sensory motor cortex in other areas of the visual cortex. In fact it has been argued by some that the primary evolutionary function of the vision system was to give us (and other mammals) the ability to flee our predators. Thus, is not a surprise that organisms which are incapable of motion do not have visual systems.

It is of interest to understand the nature of information processing that goes on in the human visual systems. For instance, the difference between the rods and cones as receptors. Rods which are far more numerous than cones are responsible for our vision in dim light. Cones do not respond to dim light, but are responsible for finer detail and for color vision. Also, it is surprising that there are approximately 125 million rods and cones on the retina but only about 1 million neurons (more properly called retinal ganglion cells) feeding the optic nerve. There is image compression performed by some intermediate cells between the rods and cones and the ganglion cells called bipolar cells. The compression is greater than 125:1 further away from the fovea or ``sweet spot'' (a rod free area) in the eye, and less nearer the fovea. Finally, it is amazing that the retina is organized with its three regions in opposite sequence, with the ganglion cells in front of the bipolar cells in front of the rods and cones. It seems reasonable that there are a 1000 million neurons in V-1 (expanded from 1 million in the optic nerve), but it seems odd that there are 10 million neurons headed down from V-1 to the lateral geniculate nucleus.



next up previous
Next: Image Processing and Up: Introduction Previous: On Painting and



S Sastry
Sun May 4 11:53:52 PDT 1997