Most animals use optical systems to form images but a substantial number rely on optics-less cutaneous vision or skin vision. And while computer scientists have spent a good deal of time and effort trying to reproduce the former, how many will even have heard of skin vision?
So a systematic investigation of this kind of imaging is long overdue, argue Leonid Yaroslavsky and mates from Tel Aviv University in Israel on the arXiv today.
What exactly are we talking about here? Yaroslavsky and pals give a list of examples from the natural world including:
The ability of some plants to orient their leaves or flowers
towards the sun
Cutaneous photoreception in reptiles
“Pit organs” of vipers located near their normal eyes – these organs are sensitive
to infra-red radiation and do not contain any optics
They even mention one or two anecdotal examples in humans to which I might add the common sense of knowing where the sun is with your eyes closed, by feeling its heat alone.
Skin vision has a number of advantages over conventional optics. Since it requires no lenses skin vision can be adapted to virtually any type of radiation at any wavelength. It can work on almost any surface and its resolution is determined by number of sensors and not by diffraction limits.
Equally, there are significant disadvantages, chief among them being that a lens creates an image using no processing power at all whereas skin vision needs significant post-processing to produce an image.
In fact, the trade off between lenses and optics-less vision systems seems to be between simplicity of design and and computational complexity. As Yaroslavskyputs it:
“What a lens does in parallel and at the speed of light, optics-less vision must replace by computations in neural machinery, which is slower and requires high energy (food) consumption.“
But where does the crossover occur that makes one type of vision better than another? This looks to be a field that is too important and potentially useful to be overlooked by biologists and computer scientists alike.
Ref: arxiv.org/abs/0808.1225: Optics-less Smart Sensors and a Possible Mechanism of Cutaneous Vision in Nature
Nice post.
However, the “Pit organs” of pit vipers are pinhole cameras with a rather large pinhole. To me this is clearly a form of optics.
The human eye also uses “significant post-processing to produce an image.” While the lens of an eye is very similar to the lens of a camera, the retina has evolved clusters of photoreceptors that work together to perceive edges, lines, and other patterns. The trade-off between “simplicity of design and and computational complexity” is far from being an all-or-none situation.