Vision
- We simulate the retina as a bunch of on-off and off-on receptors.
- There are various resolutions with the current system having
3x3, 6x6, and 9x9 fields.
- This is passed along to the simulated primary visual cortex.
- This has different receptors for different orientations of lines
(4) and angles (4). Edges are currently being implemented.
- Primary visual cortex passes information along to the secondary
visual cortex.
- In V2 two types of triangles are recognised (pyramids and stalagtites)
- These have position invariant and position variant representations.
- We got a paper on this published at a Cybernetics conference in
September.
- It's not exact but it pretty much follows the way the brain does
it.
- (We skip the LGN for example).
