MATLAB Isotropic Remesher

This is an Isoptropic Remesher based on OpenMesh (http://www.openmesh.org/) and OpenFlipper (http://www.openflipper.org/). It requires OpenMesh as a dependency.

Make sure to compile OpenMesh with the CMAKE Flag -DCMAKE_BUILD_TYPE=Release, otherwise dynamic casting is used which is not well supported by Matlab and hence can lead to segmentation faults. Otherwise set the flag

$ #define OM_FORCE_STATIC_CAST 1

in order to tell OpenMesh to use static casting.

I've done some testing, but please use at your own risk. Also I haven't tested under Windows.

Features

• Standalone except OpenMesh library
• Capable of handling feature edges, specified by feature vertices
• Easy Setting of target Edgelength
• Possibility of setting the number of iterations

Installation

With the project folder as Matlab source, you can compile the file directly in Matlab via

$ mex -I./ -I/usr/local/include -L/usr/local/lib -lOpenMeshTools -lOpenMeshCore LDFLAGS='\$LDFLAGS -Wl,-rpath,/usr/local/lib' remeshing.c ./src/BSP.cpp ./src/BSPTraits.cpp ./src/IsotropicRemesher.cpp

Make sure to point to the correct location of OpenMesh headers and libraries. The above command also sets the runtime path so libraries can be found once the mex function is loaded. On MacOSX, you can install OpenMesh via

$ brew install open-mesh --HEAD

which will install the latest version of OpenMesh. If Homebrew doesn't install the Release version, then you can quickly add the

$ -DCMAKE_BUILD_TYPE=Release

flag to the homebrew formula

$ /usr/local/Library/Formula/open-mesh.rb

in the line

$ system "cmake", "..", *std_cmake_args, "-DCMAKE_BUILD_TYPE=Release"

This is a quick-and-dirty fix, so it's much better if you compile OpenMesh yourself and set the flags manually.

Usage

In Matlab you can call the function via

$ [facetsOut, pointsOut] = remeshing(facets, points, features, targetedgelength, iterations);

facets is supposed to be a nx3 matrix of type int32 (you can simply convert a double matrix via int32(yourmatrix)) containing the IDs of the your points within the points matrix. IDs should, in Matlab notation, start with 1 and end with the size of the points matrix.

points is a nx3 matrix of type double, with x,y and z values in each column.

features is a nx1 matrix with the IDs of points considered as features vertices.

targetedgelength is your desired target edge length and should be a double value, while iterations specifies the number of iterations and should be of type int32.

In return, you'll receive similar structures as the input, with facetsOut being a nx3 matrix with the triangles specified by points IDs, and pointsOut a nx3 double matrix with the x,y and z values of the new points.

Example

$ points=randi([1 50],3,3);
$ facets=int32([1 2 3]);
$ [facetsRemeshed, pointsRemeshed] = remeshing(facets, points, int32([]), 25, int32(10));