Lucas
Reputation
959
Top tag
Next privilege 1,000 Rep.
Create tags
 Oct6 awarded Yearling Jul2 awarded Curious May15 comment Axis of symmetry of a binary image I understand that. It seems like a sensible optimisation when there is a single convex shape (and some cases which are not). I was more trying say that even if you skipped the efficiencies it would be faster. I just read what I wrote in the message above, I don't know why I said "compatible to", I definitely meant to say "faster than". That said, the down-/sub- sampling method has the advantage of "softening" the optimisation landscape, and will perform better in some cases - but it really depends on the images, and what kind of result is needed. May15 comment Axis of symmetry of a binary image I would expect, even if you didn't use the binary search, and just brute-forced every possibility, it would still be comparable to a particle based search. i.e. (1000 particles)x(1000 iterations) is probably larger than (180deg at 3deg itervals = 60angles)x(image diagonal in pixels). May15 comment Axis of symmetry of a binary image Indeed, but it's a vanishingly unlikely case. (technically you're not going to do any better than $C_4$ symmetry in an image. - assuming > 1 pixel is different) May15 comment Axis of symmetry of a binary image @String a line of symmetry that passes through the centre of mass would probably be a good starting point for an algorithm. May15 revised Axis of symmetry of a binary image added 95 characters in body May15 answered Axis of symmetry of a binary image May15 comment Axis of symmetry of a binary image OK, I'll outline what I mean in an answer. May15 comment Axis of symmetry of a binary image I'd say do it as an optimisation. But you could optimise your optimisation proceedure by doing things like taking a initial approximations using a low resolution image, or by sparsely representing the image in terms of edges, or a randomly sampled subset of your image. It'll still be a bit slow. There's loads of algorithms for these things if you look at the machine vision literature, but if you want something quick and dirty that works, a few optimisations on a brute force search would be the way forward IMO May15 comment Find the minimal value of a function May3 comment Matlab iteration @user142176 The answer is always zero. So function j=iter(p,q,m,e); return 0; end. Also, your code would work if line 3 was j=0, but I think there might be a mistake somewhere in the specification of your problem. May3 answered Matlab iteration May1 comment Whats the intergral of xe^-y dy dx with 0 and 1 as limits Often a second symbol is omitted, though it is a bit ambiguous in this case. I'd read it with [0,1] for both integrals. But reading it as an indefinite integral over x might be valid too. Apr24 comment Does relative positive definiteness have a good name? @PavelJiranek This little exchange has helped me figure out what I need to write. Thanks :) Apr24 comment Does relative positive definiteness have a good name? @PavelJiranek Indeed. The thing I'm trying to articulate in the thing I'm writing is just a sense of "bigger", that's about all that matters for my purposes. This kind of technicality (and quarrel) poses a real danger of distracting the reader from what should be a really simple and intuitive point, as well as confusing and alienating them. Apr24 revised Does relative positive definiteness have a good name? not *semi* definite Apr24 comment Does relative positive definiteness have a good name? @PavelJiranek Oops, I'll edit it. The point of avoiding it is that most people will be unfamiliar with the idea of partial orderings. I'm already asking my audience to grapple with a number of unfamiliar topics. I don't want to make it worse if I can help it. Apr24 comment Does relative positive definiteness have a good name? That's an excellent suggestion. Apr24 asked Does relative positive definiteness have a good name?