manolas
/
vcglib
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Removed also from this file the deprecated dependencies from linalg. NOW EVERY PIECE OF THE VCG relies on eigen for linalgebra.

This commit is contained in:
Paolo Cignoni 2013-03-20 08:32:53 +00:00
parent 72d67f4a11
commit 12543d68a2
1 changed files with 85 additions and 73 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

76
vcg/complex/algorithms/create/extended_marching_cubes.h
View File

@ -30,8 +30,6 @@
#include <assert.h>
#include <vector>
#include <vcg/math/base.h>
#include <vcg/math/matrix.h>
#include <vcg/math/lin_algebra.h>
#include <vcg/simplex/face/topology.h>
#include <vcg/complex/algorithms/update/normal.h>
#include <vcg/complex/algorithms/update/topology.h>
@ -39,6 +37,7 @@
#include <vcg/complex/algorithms/clean.h>
#include <vcg/space/point3.h>
#include "emc_lookup_table.h"
#include <eigenlib/Eigen/SVD>
namespace vcg
{
@ -336,45 +335,58 @@ namespace vcg
rank = (c > cos(_featureAngle) ? 2 : 3);
// setup linear system (find intersection of tangent planes)
vcg::ndim::Matrix<double> A((unsigned int) vertices_num, 3);
double *b = new double[ vertices_num ];
//--vcg::ndim::Matrix<double> A((unsigned int) vertices_num, 3);
Eigen::MatrixXd A(vertices_num,3);
//--double *b = new double[ vertices_num ];
Eigen::MatrixXd b(vertices_num,1);
for (i=0; i<vertices_num; ++i)
{
A[i][0] = normals[i][0];
A[i][1] = normals[i][1];
A[i][2] = normals[i][2];
b[i] = (points[i] * normals[i]);
//--A[i][0] = normals[i][0];
//--A[i][1] = normals[i][1];
//--A[i][2] = normals[i][2];
//--b[i] = (points[i] * normals[i]);
A(i,0) = normals[i][0];
A(i,0) = normals[i][1];
A(i,0) = normals[i][2];
b(i) = (points[i] * normals[i]);
}
// SVD of matrix A
vcg::ndim::Matrix<double> V(3, 3);
double *w = new double[vertices_num];
vcg::SingularValueDecomposition< typename vcg::ndim::Matrix<double> > (A, w, V, LeaveUnsorted, 100);
Eigen::JacobiSVD<Eigen::MatrixXd> svd(A);
Eigen::MatrixXd sol(3,1);
sol=svd.solve(b);
// vcg::ndim::Matrix<double> V(3, 3);
// double *w = new double[vertices_num];
// vcg::SingularValueDecomposition< typename vcg::ndim::Matrix<double> > (A, w, V, LeaveUnsorted, 100);
// rank == 2 -> suppress smallest singular value
if (rank == 2)
{
double smin = DBL_MAX; // the max value, as defined in <float.h>
unsigned int sminid = 0;
unsigned int srank = std::min< unsigned int >(vertices_num, 3u);
// if (rank == 2)
// {
// double smin = DBL_MAX; // the max value, as defined in <float.h>
// unsigned int sminid = 0;
// unsigned int srank = std::min< unsigned int >(vertices_num, 3u);
for (i=0; i<srank; ++i)
{
if (w[i] < smin)
{
smin = w[i];
sminid = i;
}
}
w[sminid] = 0.0;
}
// SVD backsubstitution -> least squares, least norm solution x
double *x = new double[3];
vcg::SingularValueBacksubstitution< vcg::ndim::Matrix<double> >(A, w, V, x, b);
// for (i=0; i<srank; ++i)
// {
// if (w[i] < smin)
// {
// smin = w[i];
// sminid = i;
// }
// }
// w[sminid] = 0.0;
// }
//
// // SVD backsubstitution -> least squares, least norm solution x
// double *x = new double[3];
// vcg::SingularValueBacksubstitution< vcg::ndim::Matrix<double> >(A, w, V, x, b);
// transform x to world coords
CoordType point((ScalarType) x[0], (ScalarType) x[1], (ScalarType) x[2]);
//--CoordType point((ScalarType) x[0], (ScalarType) x[1], (ScalarType) x[2]);
CoordType point((ScalarType) sol[0], (ScalarType) sol[1], (ScalarType) sol[2]);
point += center;
// Safety check if the feature point found by svd is
@ -386,7 +398,7 @@ namespace vcg
mean_point->SetUserBit(_featureFlag);
mean_point->P() = point;
mean_point->N().SetZero();
delete []x;
// delete []x;
delete []points;
delete []normals;
return mean_point;