Complex scenes, simple neurons, and complex applications

@InProceedings\{TTP2002,
  author       = "Tom Troscianko and D.J. Tolhurst and C. Alejandro Parraga",
  title        = "Complex scenes, simple neurons, and complex applications",
  booktitle    = "Perception, European Conference on Visual Perception (ECVP\´02)",
  volume       = "31",
  pages        = "383b",
  year         = "2002",
  abstract     = "A great deal is known about the behaviour of the human visual system from both psychophysical and physiological studies using simple stimuli such as gratings. However, the visual environment consists of complex scenes and often elicits complex actions. Can we use information gained about the behaviour of simple units in the visual pathway to say something about how we perceive complex scenes? If so, what novel applications exist that can make use of this knowledge?
We have been considering the issue of power spectra of natural scenes. We have shown (Parraga, Troscianko, Tolhurst,  Current Biology, 10, 35-38) that subtle shape discrimination tasks are optimised when image power spectra are \“natural\”. Furthermore, a simple model of contrast discrimination by cortical neurons predicts discrimination performance quite well. This model has been tested more extensively at a variety of eccentricities, and found to apply there as well.
Such a model could form the basis for predicting the visibility of differences between pairs of images. This has applications in computer graphics, where considerations of power spectra are relatively novel. There is evidence that images with power spectra corresponding to those of real images are judged more \“natural\” than others \– this has implications for generation of fractal terrains and texture mapping.
Finally, we have recently shown that the spatial transfer characteristics of colour vision (which are probably mediated by single-opponent cells in the mammalian cortex) provide a close match to the properties of a subset of natural scenes \– fruit on foliage \– so the notion of optimisation to natural scene statistics may well apply in the chromatic domain as well. This may provide a metric for assessing the utility of colour information in various kinds of display.",
  ifactor      = "1.314",
  area         = "PSYCHOLOGY",
  url          = "http://cic.uab.cat/Public/Publications/2002/TTP2002"
}