Access
To read this article in full you may need to log in, make a payment or gain access through a site license (see right).
News and Views
Nature Neuroscience 11, 1243–1244 (1 November 2008) | doi:10.1038/nn1108-1243
So many pixels, so little time
Abstract
We effortlessly classify and identify objects hidden in the silver grains of photographs and the discrete pixels that appear on our computer screens, but the apparent simplicity of visual recognition belies the underlying computational complexity that is inherent in high-level vision. Even for a tiny 10 |[times]| 10 pixel patch of a typical computer screen, the number of possible binary black and white images that can be displayed (>1030) is staggeringly large, making it virtually inconceivable that our fast and effortless ability to perform high-level image recognition is based on an explicit pixel- or photoreceptor-based representation.
To read this article in full you may need to log in, make a payment or gain access through a site license (see right).
