I can easily mentally continuously transform the shape of a raindrop into that of a mountain, a cloud, a lake or a dragon
Physically, I can't transform a raindrop into much, it will just pop... So without a lot more specification, this approach to image understanding doesn't seem to say very much.... If you want to say each category of images is characterized by some class of transformations under which it tends to remain invariant (so that applying a transformation in that class to a raindrop will tend to yield another raindrop), that's fine, but then you have to say which are the particular transformations corresponding to each image category. It's going to vary a lot. Also, transformations of shapes are easy to describe using mathematics.... Identifying which transformations tend to preserve a given category (e.g. raindrops) is an AI problem not a math problem, but I don't see why that is out of the grasp of algorithmic methods. ... ben On Thu, Jun 7, 2012 at 5:25 PM, Bob Mottram <[email protected]> wrote: > On 07.06.2012 22:03, Mike Tintner wrote: >> >> It starts with *any* drop shape - and it theorises about the >> *principles of transformation/transfiguration* which underlie changes >> of drop forms. They will depend on directly playing with the drops, >> or having played with similar bodies. > > > > Raindrops don't just come in any arbitrary shape. They might be pushed into > arbitrary shapes momentarily, but surface tension will cause them to adopt a > particular kind of shape in general. > > In the example photo photons are passing through the droplets as they move > towards the sensor. Since the droplets have a continuous convex shape, like > a lens, above the surface on which they're photographed this results in > gradient effects which would be a detectable feature of raindrops. > > Another feature of raindrops would be their blob-like appearance. Detecting > blobs alone isn't sufficient for identification specifically as raindrops, > but it's a contributing piece of evidence. Many weak classifiers can make a > strong one. Similarly cells in microscope images may be detected with blob > operators, even though their shapes may be highly variant. > > > > ------------------------------------------- > AGI > Archives: https://www.listbox.com/member/archive/303/=now > RSS Feed: https://www.listbox.com/member/archive/rss/303/212726-11ac2389 > Modify Your Subscription: > https://www.listbox.com/member/?&; > Powered by Listbox: http://www.listbox.com -- Ben Goertzel, PhD http://goertzel.org "My humanity is a constant self-overcoming" -- Friedrich Nietzsche ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968 Powered by Listbox: http://www.listbox.com
