Euclid
Geometry Processing and Shape Analysis in C++
|
Wave kernel signature. More...
#include <WKS.h>
Public Types | |
using | FT = FT_t< Mesh > |
using | Vec = Eigen::Matrix< FT, Eigen::Dynamic, 1 > |
using | Mat = Eigen::Matrix< FT, Eigen::Dynamic, Eigen::Dynamic > |
Public Member Functions | |
void | build (const Mesh &mesh, unsigned k=300) |
Build up the necessary computational components. More... | |
void | build (const Mesh &mesh, const Vec *eigenvalues, const Mat *eigenfunctions) |
Build up the necessary computational components. More... | |
template<typename Derived > | |
void | compute (Eigen::ArrayBase< Derived > &wks, unsigned escales=100, float emin=0.0f, float emax=-1.0f, float sigma=-1.0f) |
Compute wks for all vertices. More... | |
WKS is a intrinsic, multiscale, local shape descriptor.
Reference
Aubry, M., Schlickewei U., Cremers D.. The wave kernel signature: A quantum mechanical approach to shape analysis. The IEEE International Conference on Computer Vision (ICCV), 2011.
void Euclid::WKS< Mesh >::build | ( | const Mesh & | mesh, |
unsigned | k = 300 |
||
) |
Compute eigen decomposition of the mesh Laplacian. Note that the result number of eigenstructures might be smaller than what is requested in the because of numerical issues and some may not converge.
mesh | The target mesh. |
k | Number of eigenvalues/eigenvectors to use. |
void Euclid::WKS< Mesh >::build | ( | const Mesh & | mesh, |
const Vec * | eigenvalues, | ||
const Mat * | eigenfunctions | ||
) |
From precomputed eigen decomposition.
mesh | The target mesh. |
eigenvalues | Precomputed eigenvalues. |
eigenfunctions | Precomputed eigenfunctions. |
void Euclid::WKS< Mesh >::compute | ( | Eigen::ArrayBase< Derived > & | wks, |
unsigned | escales = 100 , |
||
float | emin = 0.0f , |
||
float | emax = -1.0f , |
||
float | sigma = -1.0f |
||
) |
wks | Output wave kernel signatures |
escales | Number of energy scales to use. |
emin | The minimum energy scale. Setting emin >= emax will use the parameters described in the paper. |
emax | The maximum energy scale. Setting emin >= emax will use the parameters described in the paper. |
sigma | The variance of the log normal distribution. Setting sigma <= 0 will use the parameters described in the paper. |