I can't readily tell how to do in Scilab, but I think you miss a problem (or maybe it's me...).
The thing is, your example shows a point cloud, not a mesh: you can't directly get a surface from a point cloud...
How would Scilab or any other software know how to link the correct set of points between themselves?
However, if your point cloud is one that would correspond to a "relatively flat" shell mesh (in the sense it's not a compact ball of points everywhere for instance...), you can use Delaunay3D meshing to have a triangular mesh corresponding to a surface.
I did a similar thing with ParaView from 3D scans. I wrote the coordinates in a file understandable by ParaView (csv I think to remember), then I used the Delaunay3D filter available. It worked rather well.