Use EdgeKey logic

release/4.3a0
Frank Dellaert 2024-10-24 14:50:31 -07:00
parent e6bfcada40
commit ade1207334
2 changed files with 91 additions and 69 deletions

View File

@ -17,56 +17,83 @@
#include <gtsam/base/numericalDerivative.h>
#include <gtsam/geometry/FundamentalMatrix.h>
#include <gtsam/inference/EdgeKey.h>
#include <gtsam/nonlinear/NonlinearFactor.h>
namespace gtsam {
template <typename F>
struct TripletError {
Point2 p0, p1, p2;
/// vector of errors returns 6D vector
Vector evaluateError(const F& F01, const F& F12, const F& F20, //
Matrix* H01, Matrix* H12, Matrix* H20) const {
std::function<Vector6(F, F, F)> fn = [&](const F& F01, const F& F12,
const F& F20) {
Vector6 error;
error << //
F::transfer(F01.matrix(), p1, F20.matrix().transpose(), p2) - p0,
F::transfer(F01.matrix().transpose(), p0, F12.matrix(), p2) - p1,
F::transfer(F20.matrix(), p0, F12.matrix().transpose(), p1) - p2;
return error;
};
if (H01) *H01 = numericalDerivative31<Vector6, F, F, F>(fn, F01, F12, F20);
if (H12) *H12 = numericalDerivative32<Vector6, F, F, F>(fn, F01, F12, F20);
if (H20) *H20 = numericalDerivative33<Vector6, F, F, F>(fn, F01, F12, F20);
return fn(F01, F12, F20);
}
};
/**
* Binary factor in the context of Structure from Motion (SfM).
* It is used to transfer points between different views based on the
* fundamental matrices between these views. The factor computes the error
* between the transferred points `pi` and `pj`, and the actual point `pk` in
* the target view. Jacobians are done using numerical differentiation.
*/
template <typename F>
class TransferFactor : public NoiseModelFactorN<F, F> {
Point2 p0, p1, p2;
EdgeKey key1_, key2_; ///< the two EdgeKeys
Point2 pi, pj, pk; ///< The points in the three views
public:
// Constructor
TransferFactor(Key key1, Key key2, const Point2& p0, const Point2& p1,
const Point2& p2, const SharedNoiseModel& model = nullptr)
: NoiseModelFactorN<F, F>(model, key1, key2), p0(p0), p1(p1), p2(p2) {}
/**
* @brief Constructor for the TransferFactor class.
*
* Uses EdgeKeys to determine how to use the two fundamental matrix unknowns
* F1 and F2, to transfer points pi and pj to the third view, and minimize the
* difference with pk.
*
* The edge keys must represent valid edges for the transfer operation,
* specifically one of the following configurations:
* - (i, k) and (j, k)
* - (i, k) and (k, j)
* - (k, i) and (j, k)
* - (k, i) and (k, j)
*
* @param key1 First EdgeKey specifying F1: (i, k) or (k, i).
* @param key2 Second EdgeKey specifying F2: (j, k) or (k, j).
* @param pi The point in the first view (i).
* @param pj The point in the second view (j).
* @param pk The point in the third (and transfer target) view (k).
* @param model An optional SharedNoiseModel that defines the noise model
* for this factor. Defaults to nullptr.
*/
TransferFactor(EdgeKey key1, EdgeKey key2, const Point2& pi, const Point2& pj,
const Point2& pk, const SharedNoiseModel& model = nullptr)
: NoiseModelFactorN<F, F>(model, key1, key2),
key1_(key1),
key2_(key2),
pi(pi),
pj(pj),
pk(pk) {}
// Create Matrix3 objects based on EdgeKey configurations
std::pair<Matrix3, Matrix3> getMatrices(const F& F1, const F& F2) const {
// Fill Fki and Fkj based on EdgeKey configurations
if (key1_.i() == key2_.i()) {
return {F1.matrix(), F2.matrix()};
} else if (key1_.i() == key2_.j()) {
return {F1.matrix(), F2.matrix().transpose()};
} else if (key1_.j() == key2_.i()) {
return {F1.matrix().transpose(), F2.matrix()};
} else if (key1_.j() == key2_.j()) {
return {F1.matrix().transpose(), F2.matrix().transpose()};
} else {
throw std::runtime_error(
"TransferFactor: invalid EdgeKey configuration.");
}
}
/// vector of errors returns 2D vector
Vector evaluateError(const F& F12, const F& F20, //
OptionalMatrixType H12 = nullptr,
OptionalMatrixType H20 = nullptr) const override {
std::function<Vector2(F, F)> fn = [&](const F& F12, const F& F20) {
Vector2 error;
error << //
F::transfer(F20.matrix(), p0, F12.matrix().transpose(), p1) - p2;
return error;
Vector evaluateError(const F& F1, const F& F2,
OptionalMatrixType H1 = nullptr,
OptionalMatrixType H2 = nullptr) const override {
std::function<Point2(F, F)> transfer = [&](const F& F1, const F& F2) {
auto [Fki, Fkj] = getMatrices(F1, F2);
return F::transfer(Fki, pi, Fkj, pj);
};
if (H12) *H12 = numericalDerivative21<Vector2, F, F>(fn, F12, F20);
if (H20) *H20 = numericalDerivative22<Vector2, F, F>(fn, F12, F20);
return fn(F12, F20);
if (H1) *H1 = numericalDerivative21<Point2, F, F>(transfer, F1, F2);
if (H2) *H2 = numericalDerivative22<Point2, F, F>(transfer, F1, F2);
return transfer(F1, F2) - pk;
}
};

View File

@ -6,20 +6,15 @@
*/
#include <CppUnitLite/TestHarness.h>
#include <gtsam/geometry/Cal3_S2.h>
#include <gtsam/geometry/EssentialMatrix.h>
#include <gtsam/geometry/FundamentalMatrix.h>
#include <gtsam/geometry/Point2.h>
#include <gtsam/geometry/Point3.h>
#include <gtsam/geometry/Rot3.h>
#include <gtsam/geometry/SimpleCamera.h>
#include <gtsam/nonlinear/LevenbergMarquardtOptimizer.h>
#include <gtsam/nonlinear/NonlinearFactorGraph.h>
#include <gtsam/nonlinear/factorTesting.h>
#include <gtsam/sfm/TransferFactor.h>
using namespace gtsam;
double focalLength = 1000;
Point2 principalPoint(640 / 2, 480 / 2);
//*************************************************************************
/// Generate three cameras on a circle, looking in
std::array<Pose3, 3> generateCameraPoses() {
@ -34,6 +29,7 @@ std::array<Pose3, 3> generateCameraPoses() {
return cameraPoses;
}
//*************************************************************************
// Function to generate a TripleF from camera poses
TripleF<SimpleFundamentalMatrix> generateTripleF(
const std::array<Pose3, 3>& cameraPoses) {
@ -48,9 +44,7 @@ TripleF<SimpleFundamentalMatrix> generateTripleF(
return {F[0], F[1], F[2]}; // Return a TripleF instance
}
double focalLength = 1000;
Point2 principalPoint(640 / 2, 480 / 2);
//*************************************************************************
// Test for TransferFactor
TEST(TransferFactor, Jacobians) {
// Generate cameras on a circle
@ -70,28 +64,29 @@ TEST(TransferFactor, Jacobians) {
}
// Create a TransferFactor
TripletError<SimpleFundamentalMatrix> error{p[0], p[1], p[2]};
Matrix H01, H12, H20;
Vector e = error.evaluateError(triplet.F01, triplet.F12, triplet.F20, &H01,
&H12, &H20);
std::cout << "Error: " << e << std::endl;
std::cout << H01 << std::endl << std::endl;
std::cout << H12 << std::endl << std::endl;
std::cout << H20 << std::endl;
EdgeKey key01(0, 1), key12(1, 2), key20(2, 0);
TransferFactor<SimpleFundamentalMatrix> //
factor0{key01, key20, p[1], p[2], p[0]},
factor1{key12, key01, p[2], p[0], p[1]},
factor2{key20, key12, p[0], p[1], p[2]};
// Create a TransferFactor
TransferFactor<SimpleFundamentalMatrix> factor{0, 1, p[0], p[1], p[2]};
Matrix H0, H1;
Vector e2 = factor.evaluateError(triplet.F12, triplet.F20, &H0, &H1);
std::cout << "Error: " << e2 << std::endl;
std::cout << H0 << std::endl << std::endl;
std::cout << H1 << std::endl << std::endl;
// Check that getMatrices is correct
auto [Fki, Fkj] = factor2.getMatrices(triplet.Fca, triplet.Fbc);
EXPECT(assert_equal<Matrix3>(triplet.Fca.matrix(), Fki));
EXPECT(assert_equal<Matrix3>(triplet.Fbc.matrix().transpose(), Fkj));
// Check Jacobians
// Create Values with edge keys
Values values;
values.insert(0, triplet.F12);
values.insert(1, triplet.F20);
EXPECT_CORRECT_FACTOR_JACOBIANS(factor, values, 1e-5, 1e-7);
values.insert(key01, triplet.Fab);
values.insert(key12, triplet.Fbc);
values.insert(key20, triplet.Fca);
// Check error and Jacobians
for (auto&& f : {factor0, factor1, factor2}) {
Vector error = f.unwhitenedError(values);
EXPECT(assert_equal<Vector>(Z_2x1, error));
EXPECT_CORRECT_FACTOR_JACOBIANS(f, values, 1e-5, 1e-7);
}
}
//*************************************************************************