python - Vectorizing Photos: Finding an Adapted Algorithm -
as little project, i've decided want write small raster vector converter. lots , lots of resources available online, many fewer actual implementations can give me kind of starting point. haven't decided language i'm going in, python seems pretty adapted.
the first issue papers directed @ vectorizing either logos or grayscale images, neither of i'm interested in. potrace, algorithm described here, 1 of libraries. same techniques applicable photo bitmaps ?
i discouraged findings until stumbled upon vector magic. results astonishing ! however, don't provide information concerning algorithm. method produces similar quality results described here: http://eprints.gla.ac.uk/47879/1/id47879.pdf. strategy remove contours , process them, before drawing them vectors.
but there no mention of how this. can extract contours opencv, ? have trouble understanding creation of vector, or happens after contour extraction.
so, conclude, here questions: language or library recommend project ? , have tips on implementing algorithm can produce satisfactory results ?
here couple of approaches:
ardeco: http://alice.loria.fr/index.php/publications.html?paper=egsr_ardeco@2006
casts problem energy minimization problem, solved via voronoi-like approachhttp://www.cs.northwestern.edu/~sco590/npar2011.pdf
more similar in spirit portrace mentionedusing gradient meshes: https://www.cct.lsu.edu/~fharhad/ganbatte/siggraph2007/cd2/content/papers/011-sun.pdf
represent gradients in image meshusing diffusion curves:
http://hal.archives-ouvertes.fr/docs/00/27/47/68/pdf/diffusion_curves.pdf
used create art drawing gradients, propose inverse problem of vectorizing image primitives.
Comments
Post a Comment