16 #include <Eigen/Dense> 18 #include <Eigen/StdVector> 19 #include <unsupported/Eigen/Polynomials> 35 for (
int i = 0; i <
n; i++)
36 Q.col(i) =
Q.col(i) /
Q.col(i).norm();
46 double lmin = Q(0, i1)*Q(0, i2) + Q(1, i1)*Q(1, i2) + Q(2, i1)*Q(2, i2);
48 Eigen::MatrixXi rij (
n,2);
50 R_=Eigen::MatrixXd::Identity(3,3);
51 t_=Eigen::Vector3d::Zero();
53 for (
int i = 0; i <
n; i++)
54 for (
int j = 0; j < 2; j++)
55 rij(i, j) = rand() %
n;
57 for (
int ii = 0; ii <
n; ii++)
59 int i = rij(ii, 0), j = rij(ii,1);
64 double l = Q(0, i)*Q(0, j) + Q(1, i)*Q(1, j) + Q(2, i)*Q(2, j);
75 Eigen::Vector3d p1, p2, p0,
x,
y,
z, dum_vec;
81 x = p2 - p0;
x /=
x.norm();
83 if (std::abs(
x(1)) < std::abs(
x(2)) )
86 z =
x.cross(dum_vec);
z /=
z.norm();
87 y =
z.cross(
x);
y /=
y.norm();
92 y = dum_vec.cross(
x);
y /=
y.norm();
93 z =
x.cross(
y);
x /=
x.norm();
98 R0.col(0) =
x; R0.col(1) =
y; R0.col(2) =
z;
100 for (
int i = 0; i <
n; i++)
101 P.col(i) = R0.transpose() * (P.col(i) - p0);
106 Eigen::Vector3d
v1 = Q.col(i1),
v2 = Q.col(i2);
107 double cg1 =
v1.dot(
v2);
108 double sg1 = sqrt(1 - cg1*cg1);
109 double D1 = (P.col(i1) - P.col(i2)).
norm();
110 Eigen::MatrixXd D4(
n - 2, 5);
114 Eigen::VectorXd rowvec(5);
115 for (
int i = 0; i <
n; i++)
117 if (i == i1 || i == i2)
121 double cg2 =
v1.dot(vi);
122 double cg3 =
v2.dot(vi);
123 double sg2 = sqrt(1 - cg2*cg2);
124 double D2 = (P.col(i1) - P.col(i)).
norm();
125 double D3 = (P.col(i) - P.col(i2)).
norm();
129 rowvec = getp3p(cg1, cg2, cg3, sg1, sg2, D1, D2, D3);
137 Eigen::VectorXd D7(8), dumvec(8), dumvec1(5);
140 for (
int i = 0; i <
n-2; i++)
143 dumvec= getpoly7(dumvec1);
147 Eigen::PolynomialSolver<double, 7> psolve(D7.reverse());
148 Eigen::VectorXcd comp_roots = psolve.roots().transpose();
149 Eigen::VectorXd real_comp, imag_comp;
150 real_comp = comp_roots.real();
151 imag_comp = comp_roots.imag();
153 Eigen::VectorXd::Index max_index;
155 double max_real= real_comp.cwiseAbs().maxCoeff(&max_index);
157 std::vector<double> act_roots_;
161 for (
int i=0; i<imag_comp.size(); i++ )
163 if(std::abs(imag_comp(i))/max_real<0.001)
165 act_roots_.push_back(real_comp(i));
170 double* ptr = &act_roots_[0];
171 Eigen::Map<Eigen::VectorXd> act_roots(ptr, cnt);
178 Eigen::VectorXd act_roots1(cnt);
179 act_roots1 << act_roots.segment(0,cnt);
181 std::vector<Eigen::Matrix3d> R_cum(cnt);
182 std::vector<Eigen::Vector3d> t_cum(cnt);
183 std::vector<double> err_cum(cnt);
185 for(
int i=0; i<cnt; i++)
187 double root = act_roots(i);
191 double d2 = cg1 + root;
193 Eigen::Vector3d unitx, unity, unitz;
199 if (std::abs(unity.dot(
x)) < std::abs(unitz.dot(
x)))
201 z =
x.cross(unity);
z/=
z.norm();
202 y=
z.cross(
x);
y/
y.norm();
206 y=unitz.cross(
x);
y/=
y.norm();
207 z =
x.cross(
y);
z/=
z.norm();
215 Eigen::MatrixXd D(2 *
n, 6);
219 Eigen::VectorXd
r(Eigen::Map<Eigen::VectorXd>(R0.data(), R0.cols()*R0.rows()));
221 for (
int j = 0; j<
n; j++)
223 double ui = img_pts(j, 0), vi = img_pts(j, 1), xi = P(0, j), yi = P(1, j), zi = P(2, j);
224 D.row(2 * j) << -
r(1)*yi + ui*(
r(7)*yi +
r(8)*zi) -
r(2)*zi,
225 -
r(2)*yi + ui*(
r(8)*yi -
r(7)*zi) +
r(1)*zi,
229 ui*
r(6)*xi -
r(0)*xi;
231 D.row(2 * j + 1) << -
r(4)*yi + vi*(
r(7)*yi +
r(8)*zi) -
r(5)*zi,
232 -
r(5)*yi + vi*(
r(8)*yi -
r(7)*zi) +
r(4)*zi,
236 vi*
r(6)*xi -
r(3)*xi;
239 Eigen::MatrixXd DTD = D.transpose()*D;
241 Eigen::EigenSolver<Eigen::MatrixXd> es(DTD);
243 Eigen::VectorXd Diag = es.pseudoEigenvalueMatrix().diagonal();
245 Eigen::MatrixXd V_mat = es.pseudoEigenvectors();
247 Eigen::MatrixXd::Index min_index;
249 Diag.minCoeff(&min_index);
251 Eigen::VectorXd V = V_mat.col(min_index);
255 double c = V(0),
s = V(1);
256 t << V(2), V(3), V(4);
259 Eigen::VectorXd xi, yi, zi;
264 Eigen::MatrixXd XXcs(3,
n), XXc(3,
n);
267 XXcs.row(0) =
r(0)*xi + (
r(1)*
c +
r(2)*
s)*yi + (-
r(1)*
s +
r(2)*
c)*zi +
t(0)*Eigen::VectorXd::Ones(
n);
268 XXcs.row(1) =
r(3)*xi + (
r(4)*
c +
r(5)*
s)*yi + (-
r(4)*
s +
r(5)*
c)*zi +
t(1)*Eigen::VectorXd::Ones(
n);
269 XXcs.row(2) =
r(6)*xi + (
r(7)*
c +
r(8)*
s)*yi + (-
r(7)*
s +
r(8)*
c)*zi +
t(2)*Eigen::VectorXd::Ones(
n);
271 for (
int ii = 0; ii <
n; ii++)
272 XXc.col(ii) = Q.col(ii)*XXcs.col(ii).norm();
277 Eigen::MatrixXd XXw = obj_pts.transpose();
279 calcampose(XXc, XXw, R2, t2);
284 for (
int k = 0; k <
n; k++)
285 XXc.col(k) = R2 * XXw.col(k) + t2;
287 Eigen::MatrixXd xxc(2,
n);
289 xxc.row(0) = XXc.row(0).array() / XXc.row(2).array();
290 xxc.row(1) = XXc.row(1).array() / XXc.row(2).array();
292 double res = ((xxc.row(0) - img_pts.col(0).transpose()).
norm() + (xxc.row(1) - img_pts.col(1).transpose()).
norm()) / 2;
298 int pos_cum = std::min_element(err_cum.begin(), err_cum.end()) - err_cum.begin();
308 Eigen::MatrixXd X = XXc;
309 Eigen::MatrixXd Y = XXw;
310 Eigen::MatrixXd K = Eigen::MatrixXd::Identity(
n,
n) - Eigen::MatrixXd::Ones(
n,
n) * 1 /
n;
311 Eigen::VectorXd ux, uy;
312 uy = X.rowwise().mean();
313 ux = Y.rowwise().mean();
316 double sigmax2 = (((X*K).array() * (X*K).array()).colwise().sum()).
mean();
318 Eigen::MatrixXd SXY = Y*K*(X.transpose()) /
n;
320 Eigen::JacobiSVD<Eigen::MatrixXd> svd(SXY, Eigen::ComputeThinU | Eigen::ComputeThinV);
322 Eigen::Matrix3d S = Eigen::MatrixXd::Identity(3, 3);
323 if (SXY.determinant() <0)
326 R2 = svd.matrixV()*S*svd.matrixU().transpose();
328 double c2 = (svd.singularValues().asDiagonal()*S).trace() / sigmax2;
331 Eigen::Vector3d
x,
y,
z;
336 if ((
x.cross(
y) -
z).
norm()>0.02)
337 R2.col(2) = -R2.col(2);
342 Eigen::VectorXd vout(8);
343 vout << 4 * pow(vin(0), 2),
345 6 * vin(2)*vin(0) + 3 * pow(vin(1), 2),
346 5 * vin(3)*vin(0) + 5 * vin(2)*vin(1),
347 4 * vin(4)*vin(0) + 4 * vin(3)*vin(1) + 2 * pow(vin(2), 2),
348 3 * vin(4)*vin(1) + 3 * vin(3)*vin(2),
349 2 * vin(4)*vin(2) + pow(vin(3), 2),
356 double A1 = (D2 / D1)*(D2 / D1);
357 double A2 =
A1*pow(C1, 2) - pow(C2, 2);
358 double A3 = l2*A5 - l1;
359 double A4 = l1*A5 - l2;
360 double A6 = (pow(D3, 2) - pow(D1, 2) - pow(D2, 2)) / (2 * pow(D1, 2));
361 double A7 = 1 - pow(l1, 2) - pow(l2, 2) + l1*l2*A5 + A6*pow(C1, 2);
363 Eigen::VectorXd vec(5);
365 vec << pow(A6, 2) -
A1*pow(A5, 2),
366 2 * (A3*A6 -
A1*A4*A5),
367 pow(A3, 2) + 2 * A6*A7 -
A1*pow(A4, 2) - A2*pow(A5, 2),
368 2 * (A3*A7 - A2*A4*A5),
369 pow(A7, 2) - A2*pow(A4, 2);
Eigen::MatrixXd P
Camera Intrinsic Matrix.
void calcampose(Eigen::MatrixXd &XXc, Eigen::MatrixXd &XXw, Eigen::Matrix3d &R2, Eigen::Vector3d &t2)
Function to calculate final pose.
Eigen::MatrixXd cam_intrinsic
Image Points (n X 3) in Camera Co-ordinate system.
rpnp(Eigen::MatrixXd obj_pts_, Eigen::MatrixXd img_pts_, Eigen::MatrixXd cam_, int n0)
Number of 2D/3D correspondences.
Eigen::VectorXd getp3p(double l1, double l2, double A5, double C1, double C2, double D1, double D2, double D3)
Function to compute pose using P3P.
Eigen::MatrixXd img_pts
Object Points (n X 3) in Camera Co-ordinate system.
Robust - PnP class definition for computing pose.
bool compute_pose(Eigen::Ref< Eigen::Matrix3d > R_, Eigen::Ref< Eigen::Vector3d > t_)
Function to compute pose.
double A1
UTC constant and 1st order terms.
GLdouble GLdouble GLdouble r
Eigen::MatrixXd Q
Transposed Object Points (3 X n) for computations.
Eigen::Matrix3d R
Transposed Image Points (3 X n) for computations.
Eigen::VectorXd getpoly7(const Eigen::VectorXd &vin)
Get Polynomial from input vector.
GLfloat GLfloat GLfloat v2
EIGEN_STRONG_INLINE double mean() const
Computes the mean of the entire matrix.
CONTAINER::Scalar norm(const CONTAINER &v)