(Reverse) Distance Transformation on Periodic Spaces

[rolanddenis]

PR Description

Adding per-dimension periodicity specification to VoronoiMap (+DistanceTransformation) and PowerMap (+ReverseDistanceTransformation), following D. Coeurjolly, 2008, Distance Transformation, Reverse Distance Transformation and Discrete Medial Axis on Toric Spaces 😲

Checklist

• Adding per-dimension periodicity specification
  • VoronoiMap & DistanceTransformation
  • PowerMap & ReverseDistanceTransformation
• Adding methods to get bounded coordinates of nearest site.
• Unit-test of your feature with Catch.
  • VoronoiMap & DistanceTransformation(partial)
  • PowerMap & ReverseDistanceTransformation(partial) & ReducedMedialAxis
• Doxygen documentation of the code completed (classes, methods, types, members...)
  • VoronoiMap & DistanceTransformation
  • PowerMap & ReverseDistanceTransformation
• Documentation module page added or updated.
  • nD Volumetric Analysis using Separable Processes (moduleVolumetric)
  • [ ] Tutorial "Image -> Region -> Distance Transformation" (tutoImageRegionDT)
• New entry in the ChangeLog.md added.
• No warning raised in Debug cmake mode (otherwise, Travis C.I. will fail).
• All continuous integration tests pass (Travis & appveyor)

[dcoeurjo]

sorry for the delay @rolanddenis .

What's wrong with the break index in power map?

[dcoeurjo]

request changes
(still playing;))

[dcoeurjo]

lowest

(just playing with new Github review system)

[rolanddenis]

Hum ... and now ? Do you have to approve the changes or did I miss something ?

[dcoeurjo]

not clear..

[dcoeurjo]

(your HDF5 fix does not work neither on my mac, I've no idea what the problem is :()

[rolanddenis]

😰 I was trying the solution from http://stackoverflow.com/questions/34050155/symbol-not-found-linking-to-hdf-library ... Even if it doesn't solve the problem, it seems that requiring the CXX component of HDF5 is not a bad idea (but I will revert the commit, it is not the right PR for this).

So you also have the error on your computer ? It is not just a Travis specific configuration problem...

[rolanddenis]

I think the problem is related to that : http://stackoverflow.com/questions/16352833/linking-with-clang-on-os-x-generates-lots-of-symbol-not-found-errors

It may come from the fact that hdf5 has been compiled with a C++ library different to the one used to compile DGtal. I also had the same problem when linking a program compiled with gcc >=5 with boost compiled with gcc < 5 : the std::basic_string symbol changed with gcc 5 (like the error in Travis).

Another link on this problem: http://stackoverflow.com/questions/8454329/why-cant-clang-with-libc-in-c0x-mode-link-this-boostprogram-options-examp

[dcoeurjo]

By default, brew use clang compiler (same one as DGTal). I think the problem is more related to hdf5 upstream issue rather that something from our side

[dcoeurjo]

(And I agree, CXX flag for hdf5 was missing)

[rolanddenis]

Yes, I agree...

[dcoeurjo]

thanks a lot @rolanddenis for your edits. I'm reviewing them.

[dcoeurjo]

all bots are fine now. @rolanddenis can you please add something in the doc about the "(partial)" thing?
thx

[rolanddenis]

I added a small paragraph on partial periodic specification (even if your remark was not about that) ;)

[dcoeurjo]

thanks.

[dcoeurjo]

Great merging. thanks again