examples/libmv_homography.cc

// Ceres Solver - A fast non-linear least squares minimizer
// Copyright 2015 Google Inc. All rights reserved.
// http://ceres-solver.org/
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are met:
//
// * Redistributions of source code must retain the above copyright notice,
//   this list of conditions and the following disclaimer.
// * Redistributions in binary form must reproduce the above copyright notice,
//   this list of conditions and the following disclaimer in the documentation
//   and/or other materials provided with the distribution.
// * Neither the name of Google Inc. nor the names of its contributors may be
//   used to endorse or promote products derived from this software without
//   specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
// ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
// LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
// CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
// SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
// INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
// CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
// ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
// POSSIBILITY OF SUCH DAMAGE.
//
// Copyright (c) 2014 libmv authors.
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to
// deal in the Software without restriction, including without limitation the
// rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
// sell copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
// FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
// IN THE SOFTWARE.
//
// Author: sergey.vfx@gmail.com (Sergey Sharybin)
//
// This file demonstrates solving for a homography between two sets of points.
// A homography describes a transformation between a sets of points on a plane,
// perspectively projected into two images. The first step is to solve a
// homogeneous system of equations via singular value decomposition, giving an
// algebraic solution for the homography, then solving for a final solution by
// minimizing the symmetric transfer error in image space with Ceres (called the
// Gold Standard Solution in "Multiple View Geometry"). The routines are based
// on the routines from the Libmv library.
//
// This example demonstrates custom exit criterion by having a callback check
// for image-space error.

#include <utility>

#include "ceres/ceres.h"
#include "glog/logging.h"

using EigenDouble = Eigen::NumTraits<double>;

using Mat = Eigen::MatrixXd;
using Vec = Eigen::VectorXd;
using Mat3 = Eigen::Matrix<double, 3, 3>;
using Vec2 = Eigen::Matrix<double, 2, 1>;
using MatX8 = Eigen::Matrix<double, Eigen::Dynamic, 8>;
using Vec3 = Eigen::Vector3d;

namespace {

// This structure contains options that controls how the homography
// estimation operates.
//
// Defaults should be suitable for a wide range of use cases, but
// better performance and accuracy might require tweaking.
struct EstimateHomographyOptions {
  // Default settings for homography estimation which should be suitable
  // for a wide range of use cases.
  EstimateHomographyOptions() = default;

  // Maximal number of iterations for the refinement step.
  int max_num_iterations{50};

  // Expected average of symmetric geometric distance between
  // actual destination points and original ones transformed by
  // estimated homography matrix.
  //
  // Refinement will finish as soon as average of symmetric
  // geometric distance is less or equal to this value.
  //
  // This distance is measured in the same units as input points are.
  double expected_average_symmetric_distance{1e-16};
};

// Calculate symmetric geometric cost terms:
//
// forward_error = D(H * x1, x2)
// backward_error = D(H^-1 * x2, x1)
//
// Templated to be used with autodifferentiation.
template <typename T>
void SymmetricGeometricDistanceTerms(const Eigen::Matrix<T, 3, 3>& H,
                                     const Eigen::Matrix<T, 2, 1>& x1,
                                     const Eigen::Matrix<T, 2, 1>& x2,
                                     T forward_error[2],
                                     T backward_error[2]) {
  using Vec3 = Eigen::Matrix<T, 3, 1>;
  Vec3 x(x1(0), x1(1), T(1.0));
  Vec3 y(x2(0), x2(1), T(1.0));

  Vec3 H_x = H * x;
  Vec3 Hinv_y = H.inverse() * y;

  H_x /= H_x(2);
  Hinv_y /= Hinv_y(2);

  forward_error[0] = H_x(0) - y(0);
  forward_error[1] = H_x(1) - y(1);
  backward_error[0] = Hinv_y(0) - x(0);
  backward_error[1] = Hinv_y(1) - x(1);
}

// Calculate symmetric geometric cost:
//
//   D(H * x1, x2)^2 + D(H^-1 * x2, x1)^2
//
double SymmetricGeometricDistance(const Mat3& H,
                                  const Vec2& x1,
                                  const Vec2& x2) {
  Vec2 forward_error, backward_error;
  SymmetricGeometricDistanceTerms<double>(
      H, x1, x2, forward_error.data(), backward_error.data());
  return forward_error.squaredNorm() + backward_error.squaredNorm();
}

// A parameterization of the 2D homography matrix that uses 8 parameters so
// that the matrix is normalized (H(2,2) == 1).
// The homography matrix H is built from a list of 8 parameters (a, b,...g, h)
// as follows
//
//         |a b c|
//     H = |d e f|
//         |g h 1|
//
template <typename T = double>
class Homography2DNormalizedParameterization {
 public:
  using Parameters = Eigen::Matrix<T, 8, 1>;     // a, b, ... g, h
  using Parameterized = Eigen::Matrix<T, 3, 3>;  // H

  // Convert from the 8 parameters to a H matrix.
  static void To(const Parameters& p, Parameterized* h) {
    // clang-format off
    *h << p(0), p(1), p(2),
          p(3), p(4), p(5),
          p(6), p(7), 1.0;
    // clang-format on
  }

  // Convert from a H matrix to the 8 parameters.
  static void From(const Parameterized& h, Parameters* p) {
    // clang-format off
    *p << h(0, 0), h(0, 1), h(0, 2),
          h(1, 0), h(1, 1), h(1, 2),
          h(2, 0), h(2, 1);
    // clang-format on
  }
};

// 2D Homography transformation estimation in the case that points are in
// euclidean coordinates.
//
//   x = H y
//
// x and y vector must have the same direction, we could write
//
//   crossproduct(|x|, * H * |y| ) = |0|
//
//   | 0 -1  x2|   |a b c|   |y1|    |0|
//   | 1  0 -x1| * |d e f| * |y2| =  |0|
//   |-x2  x1 0|   |g h 1|   |1 |    |0|
//
// That gives:
//
//   (-d+x2*g)*y1    + (-e+x2*h)*y2 + -f+x2          |0|
//   (a-x1*g)*y1     + (b-x1*h)*y2  + c-x1         = |0|
//   (-x2*a+x1*d)*y1 + (-x2*b+x1*e)*y2 + -x2*c+x1*f  |0|
//
bool Homography2DFromCorrespondencesLinearEuc(const Mat& x1,
                                              const Mat& x2,
                                              Mat3* H,
                                              double expected_precision) {
  assert(2 == x1.rows());
  assert(4 <= x1.cols());
  assert(x1.rows() == x2.rows());
  assert(x1.cols() == x2.cols());

  int n = x1.cols();
  MatX8 L = Mat::Zero(n * 3, 8);
  Mat b = Mat::Zero(n * 3, 1);
  for (int i = 0; i < n; ++i) {
    int j = 3 * i;
    L(j, 0) = x1(0, i);              // a
    L(j, 1) = x1(1, i);              // b
    L(j, 2) = 1.0;                   // c
    L(j, 6) = -x2(0, i) * x1(0, i);  // g
    L(j, 7) = -x2(0, i) * x1(1, i);  // h
    b(j, 0) = x2(0, i);              // i

    ++j;
    L(j, 3) = x1(0, i);              // d
    L(j, 4) = x1(1, i);              // e
    L(j, 5) = 1.0;                   // f
    L(j, 6) = -x2(1, i) * x1(0, i);  // g
    L(j, 7) = -x2(1, i) * x1(1, i);  // h
    b(j, 0) = x2(1, i);              // i

    // This ensures better stability
    // TODO(julien) make a lite version without this 3rd set
    ++j;
    L(j, 0) = x2(1, i) * x1(0, i);   // a
    L(j, 1) = x2(1, i) * x1(1, i);   // b
    L(j, 2) = x2(1, i);              // c
    L(j, 3) = -x2(0, i) * x1(0, i);  // d
    L(j, 4) = -x2(0, i) * x1(1, i);  // e
    L(j, 5) = -x2(0, i);             // f
  }
  // Solve Lx=B
  const Vec h = L.fullPivLu().solve(b);
  Homography2DNormalizedParameterization<double>::To(h, H);
  return (L * h).isApprox(b, expected_precision);
}

// Cost functor which computes symmetric geometric distance
// used for homography matrix refinement.
class HomographySymmetricGeometricCostFunctor {
 public:
  HomographySymmetricGeometricCostFunctor(Vec2 x, Vec2 y)
      : x_(std::move(x)), y_(std::move(y)) {}

  template <typename T>
  bool operator()(const T* homography_parameters, T* residuals) const {
    using Mat3 = Eigen::Matrix<T, 3, 3>;
    using Vec2 = Eigen::Matrix<T, 2, 1>;

    Mat3 H(homography_parameters);
    Vec2 x(T(x_(0)), T(x_(1)));
    Vec2 y(T(y_(0)), T(y_(1)));

    SymmetricGeometricDistanceTerms<T>(H, x, y, &residuals[0], &residuals[2]);
    return true;
  }

  const Vec2 x_;
  const Vec2 y_;
};

// Termination checking callback. This is needed to finish the
// optimization when an absolute error threshold is met, as opposed
// to Ceres's function_tolerance, which provides for finishing when
// successful steps reduce the cost function by a fractional amount.
// In this case, the callback checks for the absolute average reprojection
// error and terminates when it's below a threshold (for example all
// points < 0.5px error).
class TerminationCheckingCallback : public ceres::IterationCallback {
 public:
  TerminationCheckingCallback(const Mat& x1,
                              const Mat& x2,
                              const EstimateHomographyOptions& options,
                              Mat3* H)
      : options_(options), x1_(x1), x2_(x2), H_(H) {}

  ceres::CallbackReturnType operator()(
      const ceres::IterationSummary& summary) override {
    // If the step wasn't successful, there's nothing to do.
    if (!summary.step_is_successful) {
      return ceres::SOLVER_CONTINUE;
    }

    // Calculate average of symmetric geometric distance.
    double average_distance = 0.0;
    for (int i = 0; i < x1_.cols(); i++) {
      average_distance +=
          SymmetricGeometricDistance(*H_, x1_.col(i), x2_.col(i));
    }
    average_distance /= x1_.cols();

    if (average_distance <= options_.expected_average_symmetric_distance) {
      return ceres::SOLVER_TERMINATE_SUCCESSFULLY;
    }

    return ceres::SOLVER_CONTINUE;
  }

 private:
  const EstimateHomographyOptions& options_;
  const Mat& x1_;
  const Mat& x2_;
  Mat3* H_;
};

bool EstimateHomography2DFromCorrespondences(
    const Mat& x1,
    const Mat& x2,
    const EstimateHomographyOptions& options,
    Mat3* H) {
  assert(2 == x1.rows());
  assert(4 <= x1.cols());
  assert(x1.rows() == x2.rows());
  assert(x1.cols() == x2.cols());

  // Step 1: Algebraic homography estimation.
  // Assume algebraic estimation always succeeds.
  Homography2DFromCorrespondencesLinearEuc(
      x1, x2, H, EigenDouble::dummy_precision());

  LOG(INFO) << "Estimated matrix after algebraic estimation:\n" << *H;

  // Step 2: Refine matrix using Ceres minimizer.
  ceres::Problem problem;
  for (int i = 0; i < x1.cols(); i++) {
    auto* homography_symmetric_geometric_cost_function =
        new HomographySymmetricGeometricCostFunctor(x1.col(i), x2.col(i));

    problem.AddResidualBlock(
        new ceres::AutoDiffCostFunction<HomographySymmetricGeometricCostFunctor,
                                        4,  // num_residuals
                                        9>(
            homography_symmetric_geometric_cost_function),
        nullptr,
        H->data());
  }

  // Configure the solve.
  ceres::Solver::Options solver_options;
  solver_options.linear_solver_type = ceres::DENSE_QR;
  solver_options.max_num_iterations = options.max_num_iterations;
  solver_options.update_state_every_iteration = true;

  // Terminate if the average symmetric distance is good enough.
  TerminationCheckingCallback callback(x1, x2, options, H);
  solver_options.callbacks.push_back(&callback);

  // Run the solve.
  ceres::Solver::Summary summary;
  ceres::Solve(solver_options, &problem, &summary);

  LOG(INFO) << "Summary:\n" << summary.FullReport();
  LOG(INFO) << "Final refined matrix:\n" << *H;

  return summary.IsSolutionUsable();
}

}  // namespace

int main(int argc, char** argv) {
  google::InitGoogleLogging(argv[0]);

  Mat x1(2, 100);
  for (int i = 0; i < x1.cols(); ++i) {
    x1(0, i) = rand() % 1024;
    x1(1, i) = rand() % 1024;
  }

  Mat3 homography_matrix;
  // This matrix has been dumped from a Blender test file of plane tracking.
  // clang-format off
  homography_matrix << 1.243715, -0.461057, -111.964454,
                       0.0,       0.617589, -192.379252,
                       0.0,      -0.000983,    1.0;
  // clang-format on

  Mat x2 = x1;
  for (int i = 0; i < x2.cols(); ++i) {
    Vec3 homogenous_x1 = Vec3(x1(0, i), x1(1, i), 1.0);
    Vec3 homogenous_x2 = homography_matrix * homogenous_x1;
    x2(0, i) = homogenous_x2(0) / homogenous_x2(2);
    x2(1, i) = homogenous_x2(1) / homogenous_x2(2);

    // Apply some noise so algebraic estimation is not good enough.
    x2(0, i) += static_cast<double>(rand() % 1000) / 5000.0;
    x2(1, i) += static_cast<double>(rand() % 1000) / 5000.0;
  }

  Mat3 estimated_matrix;

  EstimateHomographyOptions options;
  options.expected_average_symmetric_distance = 0.02;
  EstimateHomography2DFromCorrespondences(x1, x2, options, &estimated_matrix);

  // Normalize the matrix for easier comparison.
  estimated_matrix /= estimated_matrix(2, 2);

  std::cout << "Original matrix:\n" << homography_matrix << "\n";
  std::cout << "Estimated matrix:\n" << estimated_matrix << "\n";

  return EXIT_SUCCESS;
}