This tool consists of a few functions, like, im2Bounds (image to polygonal boundaries), getCtrlPnts (mark intersecting vertex between polygons, serving as fixed point for polygon simplification and meshing), simplifyBounds (simplify polygon), poly2mesh (polygon to triangular meshes), getInterf (search nodes at interface), printInp (export as Inp file), and printBdf (export as Bdf file).
- Exactly reserve the contact detail between different phases.
- Able to avoid sharp corners when simplifying polyline.
- The exported Inp file can be imported into software Abaqus. The exported Bdf file can be imported into software Nastran and Comsol.
Note: To make this code run (especially for poly2mesh), you need to download MESH2D from https://www.mathworks.com/matlabcentral/fileexchange/25555-mesh2d-delaunay-based-unstructured-mesh-generation, and add the folder (mesh2d-master) to your path. After that, you can start with demo.m in im2mesh, and try 4 examples listed in demo.m. Check the images in the folder (im2mesh\images\...).
If you have any questions or suggestions, feel free to send me an email.
email@example.com, Jan 2020.
Great thanks Dr. Yang Lu providing valuable advice to this tool.
Jiexian Ma (2020). Im2mesh (2D image to triangular meshes) (https://www.mathworks.com/matlabcentral/fileexchange/71772-im2mesh-2d-image-to-triangular-meshes), MATLAB Central File Exchange. Retrieved .
@chunfeng: Yes, that's the purpose of this tool. However, for large images (>1000*1000), it takes some time to get the meshing result. Check the latest version. I add some examples. About boundary conditions, check the comments that I wrote in printInp_multiPart.m. Boundary node numbers of each material is extracted. Node set and node-based surface at boundary of each material are defined in the exported inp file. Since the node-based surface is not supported in Abaqus CAE, you might need to read the documentation of Abaqus about how to use inp file to define boundary conditions and constraints.
Hello, I want to know if this method can accurately define the mesh of boundary intersection of different materials to make the boundary smoother. In addition, does the number and node generated by this method satisfy the finite element calculation? The boundary node numbers of different materials can be extracted separately and used to impose boundary conditions in the finite element?
@Jiexian: Got it! This is helpful - thanks. FYI: I have tweaked it to just select which specific gray levels we want to mesh (and which to ignore).
Nice work and helpful code. Thanks!
@Seth, yep, I know your meaning.
You can replace line 19 to 21 in im2Bounds.m with following code. It should work.
nlevels = 1;
bounds = cell( 1, nlevels );
intensity = max(max(im));
Is there a simple way to provide the mesh of just a single object in the image instead of meshing the full image space? For example, you provide an example image titled "a1.tif". Is it easy to just mesh the white object instead of both the white and black space? (Say I wanted to build a solid model in Abaqus of this white object, but the black color is just air and is not usually modeled).
This may also be helpful in the cases of very large images that require a long time to mesh. In that case, I don't want to waste resources on meshing the background of the image.
I tried the code by running the "demo.m" file. There is an Abaqus input file, "test.inp", generated. However, when I import the inp file to ABAQUS CAE or Hypermesh, both of them complain about errors. Would you please double check the formatting of the inp file?
revise demo(), add examples
Improve time efficiency of getCtrlPnts(). Faster. Add new parameter - tf_avoid_sharp_corner. Able to avoid non-convergence when meshing.
new image for cover
revise getCtrlPnts() for time efficiency