Gabriel Peyré’s Fast Marching Toolbox

Hello,

I am working with Gabriel Peyré’s Fast Marching Toolbox in MATLAB. The isotropic version of the toolbox compiles and runs fine after adjusting for mwSize compatibility (using -compatibleArrayDims). I can successfully run 2D, 3D, and mesh-based isotropic fast marching (perform_front_propagation_2d, perform_front_propagation_3d, perform_fast_marching_mesh).

However, when I try to use anisotropic fast marching, I get the following error:

Unrecognized function or variable 'perform_front_propagation_anisotropic'.

When I attempt to compile perform_front_propagation_anisotropic.cpp, MATLAB cannot find the required header:

fatal error: anisotropic-fm/AnisotropicTensorDistance.h: No such file or directory

So it seems my version of the toolbox is missing key anisotropic files (e.g., AnisotropicTensorDistance.h).

Answer 1

[dpb]

It appears that if you go to the Version History tab and use the download link for the BSD license V 1.1.0.0 instead there are some additional files including that missing header file in the anisotropic folder.

I haven't done any further comparison other than I noticed that download is 6.1MB instead of 5.8 that the main page link and the V 1.2.0.0 version links are.

You'll have to figure out whether anything else is different or not, but that is at least a file of that name...

[dpb]

ADDENDUM

If this works and you can straighten things out, you might see about following through to clean up the submission page. It appears there is no active support; the only links I found to Gabriel Peyre web site all 404'ed and nothing showed up on GitHub.

[Mehmoona]

Even after downloading the version I still get the error, although I have added the path correctly and then it gives error like check for incorrect data type or missing argument in call to function 'getoptions'.

[dpb]

The above is a missing m-file it appears.

If it isn't in either of the download links, then it would appear your only recourse will be to continue to search the internet to see if you can find any support for the package or create a replacement yourself.

That's the problem with old, unsupported code--when the author loses interest and moves on, there's little recourse other than investing the time and effort to do the work oneself or find another source.

[Torsten]

The advice here

https://jvgomez.github.io/pages/fast-marching-toolbox-installation-tutorial.html

is to comment the lines

% mex mex/anisotropic-fm//perform_front_propagation_anisotropic.cpp
% mex mex/anisotropic-fm-feth/fm2dAniso.cpp

in the compile_mex.m file because otherwise, an error will occur during compilation.

I guess that if you did this, you cannot expect "perform_front_propagation_anisotropic" to be found during execution.

[Mehmoona]

" I guess that if you did this, you cannot expect "perform_front_propagation_anisotropic" to be found during execution."

Yeah, that's true I should better start looking at some other methods.

[dpb]

I suppose you could uncomment the anisotropic lines and see what is then missing and decide whether those are something you could create replacements for. From the looks of it, there appears to be a lot of effort in building what is there so it's probably a nontrivial task to complete.

[Mehmoona]

I have uncommented it and resolve all the errors in the .cpp and .h files and it is working for anisotropic.

[dpb]

Kewl. I'd suggest then if you are convinced it is correct to update the FEX submission. I don't know if it's possible to do so at the existing submission or not without Mathworks support.