blob: 1576e829e738fb245ca30adfcced840c925fc902 [file] [log] [blame]
// Ceres Solver - A fast non-linear least squares minimizer
// Copyright 2019 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.
//
// Author: keir@google.com (Keir Mierle)
// sameeragarwal@google.com (Sameer Agarwal)
#ifndef CERES_PUBLIC_LOCAL_PARAMETERIZATION_H_
#define CERES_PUBLIC_LOCAL_PARAMETERIZATION_H_
#include <array>
#include <memory>
#include <vector>
#include "ceres/internal/disable_warnings.h"
#include "ceres/internal/port.h"
namespace ceres {
// Purpose: Sometimes parameter blocks x can overparameterize a problem
//
// min f(x)
// x
//
// In that case it is desirable to choose a parameterization for the
// block itself to remove the null directions of the cost. More
// generally, if x lies on a manifold of a smaller dimension than the
// ambient space that it is embedded in, then it is numerically and
// computationally more effective to optimize it using a
// parameterization that lives in the tangent space of that manifold
// at each point.
//
// For example, a sphere in three dimensions is a 2 dimensional
// manifold, embedded in a three dimensional space. At each point on
// the sphere, the plane tangent to it defines a two dimensional
// tangent space. For a cost function defined on this sphere, given a
// point x, moving in the direction normal to the sphere at that point
// is not useful. Thus a better way to do a local optimization is to
// optimize over two dimensional vector delta in the tangent space at
// that point and then "move" to the point x + delta, where the move
// operation involves projecting back onto the sphere. Doing so
// removes a redundant dimension from the optimization, making it
// numerically more robust and efficient.
//
// More generally we can define a function
//
// x_plus_delta = Plus(x, delta),
//
// where x_plus_delta has the same size as x, and delta is of size
// less than or equal to x. The function Plus, generalizes the
// definition of vector addition. Thus it satisfies the identify
//
// Plus(x, 0) = x, for all x.
//
// A trivial version of Plus is when delta is of the same size as x
// and
//
// Plus(x, delta) = x + delta
//
// A more interesting case if x is two dimensional vector, and the
// user wishes to hold the first coordinate constant. Then, delta is a
// scalar and Plus is defined as
//
// Plus(x, delta) = x + [0] * delta
// [1]
//
// An example that occurs commonly in Structure from Motion problems
// is when camera rotations are parameterized using Quaternion. There,
// it is useful only make updates orthogonal to that 4-vector defining
// the quaternion. One way to do this is to let delta be a 3
// dimensional vector and define Plus to be
//
// Plus(x, delta) = [cos(|delta|), sin(|delta|) delta / |delta|] * x
//
// The multiplication between the two 4-vectors on the RHS is the
// standard quaternion product.
//
// Given g and a point x, optimizing f can now be restated as
//
// min f(Plus(x, delta))
// delta
//
// Given a solution delta to this problem, the optimal value is then
// given by
//
// x* = Plus(x, delta)
//
// The class LocalParameterization defines the function Plus and its
// Jacobian which is needed to compute the Jacobian of f w.r.t delta.
class CERES_EXPORT LocalParameterization {
public:
virtual ~LocalParameterization();
// Generalization of the addition operation,
//
// x_plus_delta = Plus(x, delta)
//
// with the condition that Plus(x, 0) = x.
virtual bool Plus(const double* x,
const double* delta,
double* x_plus_delta) const = 0;
// The jacobian of Plus(x, delta) w.r.t delta at delta = 0.
//
// jacobian is a row-major GlobalSize() x LocalSize() matrix.
virtual bool ComputeJacobian(const double* x, double* jacobian) const = 0;
// local_matrix = global_matrix * jacobian
//
// global_matrix is a num_rows x GlobalSize row major matrix.
// local_matrix is a num_rows x LocalSize row major matrix.
// jacobian(x) is the matrix returned by ComputeJacobian at x.
//
// This is only used by GradientProblem. For most normal uses, it is
// okay to use the default implementation.
virtual bool MultiplyByJacobian(const double* x,
const int num_rows,
const double* global_matrix,
double* local_matrix) const;
// Size of x.
virtual int GlobalSize() const = 0;
// Size of delta.
virtual int LocalSize() const = 0;
};
// Some basic parameterizations
// Identity Parameterization: Plus(x, delta) = x + delta
class CERES_EXPORT IdentityParameterization : public LocalParameterization {
public:
explicit IdentityParameterization(int size);
virtual ~IdentityParameterization() {}
bool Plus(const double* x,
const double* delta,
double* x_plus_delta) const override;
bool ComputeJacobian(const double* x, double* jacobian) const override;
bool MultiplyByJacobian(const double* x,
const int num_cols,
const double* global_matrix,
double* local_matrix) const override;
int GlobalSize() const override { return size_; }
int LocalSize() const override { return size_; }
private:
const int size_;
};
// Hold a subset of the parameters inside a parameter block constant.
class CERES_EXPORT SubsetParameterization : public LocalParameterization {
public:
explicit SubsetParameterization(int size,
const std::vector<int>& constant_parameters);
virtual ~SubsetParameterization() {}
bool Plus(const double* x,
const double* delta,
double* x_plus_delta) const override;
bool ComputeJacobian(const double* x, double* jacobian) const override;
bool MultiplyByJacobian(const double* x,
const int num_cols,
const double* global_matrix,
double* local_matrix) const override;
int GlobalSize() const override {
return static_cast<int>(constancy_mask_.size());
}
int LocalSize() const override { return local_size_; }
private:
const int local_size_;
std::vector<char> constancy_mask_;
};
// Plus(x, delta) = [cos(|delta|), sin(|delta|) delta / |delta|] * x
// with * being the quaternion multiplication operator. Here we assume
// that the first element of the quaternion vector is the real (cos
// theta) part.
class CERES_EXPORT QuaternionParameterization : public LocalParameterization {
public:
virtual ~QuaternionParameterization() {}
bool Plus(const double* x,
const double* delta,
double* x_plus_delta) const override;
bool ComputeJacobian(const double* x, double* jacobian) const override;
int GlobalSize() const override { return 4; }
int LocalSize() const override { return 3; }
};
// Implements the quaternion local parameterization for Eigen's representation
// of the quaternion. Eigen uses a different internal memory layout for the
// elements of the quaternion than what is commonly used. Specifically, Eigen
// stores the elements in memory as [x, y, z, w] where the real part is last
// whereas it is typically stored first. Note, when creating an Eigen quaternion
// through the constructor the elements are accepted in w, x, y, z order. Since
// Ceres operates on parameter blocks which are raw double pointers this
// difference is important and requires a different parameterization.
//
// Plus(x, delta) = [sin(|delta|) delta / |delta|, cos(|delta|)] * x
// with * being the quaternion multiplication operator.
class CERES_EXPORT EigenQuaternionParameterization
: public ceres::LocalParameterization {
public:
virtual ~EigenQuaternionParameterization() {}
bool Plus(const double* x,
const double* delta,
double* x_plus_delta) const override;
bool ComputeJacobian(const double* x, double* jacobian) const override;
int GlobalSize() const override { return 4; }
int LocalSize() const override { return 3; }
};
// This provides a parameterization for homogeneous vectors which are commonly
// used in Structure for Motion problems. One example where they are used is
// in representing points whose triangulation is ill-conditioned. Here
// it is advantageous to use an over-parameterization since homogeneous vectors
// can represent points at infinity.
//
// The plus operator is defined as
// Plus(x, delta) =
// [sin(0.5 * |delta|) * delta / |delta|, cos(0.5 * |delta|)] * x
// with * defined as an operator which applies the update orthogonal to x to
// remain on the sphere. We assume that the last element of x is the scalar
// component. The size of the homogeneous vector is required to be greater than
// 1.
class CERES_EXPORT HomogeneousVectorParameterization
: public LocalParameterization {
public:
explicit HomogeneousVectorParameterization(int size);
virtual ~HomogeneousVectorParameterization() {}
bool Plus(const double* x,
const double* delta,
double* x_plus_delta) const override;
bool ComputeJacobian(const double* x, double* jacobian) const override;
int GlobalSize() const override { return size_; }
int LocalSize() const override { return size_ - 1; }
private:
const int size_;
};
// This provides a parameterization for lines, where the line is
// over-parameterized by an origin point and a direction vector. So the
// parameter vector size needs to be two times the ambient space dimension,
// where the first half is interpreted as the origin point and the second half
// as the direction.
//
// The plus operator for the line direction is the same as for the
// HomogeneousVectorParameterization. The update of the origin point is
// perpendicular to the line direction before the update.
//
// This local parameterization is a special case of the affine Grassmannian
// manifold (see https://en.wikipedia.org/wiki/Affine_Grassmannian_(manifold))
// for the case Graff_1(R^n).
template <int AmbientSpaceDimension>
class LineParameterization : public LocalParameterization {
public:
static_assert(AmbientSpaceDimension >= 2,
"The ambient space must be at least 2");
bool Plus(const double* x,
const double* delta,
double* x_plus_delta) const override;
bool ComputeJacobian(const double* x, double* jacobian) const override;
int GlobalSize() const override { return 2 * AmbientSpaceDimension; }
int LocalSize() const override { return 2 * (AmbientSpaceDimension - 1); }
};
// Construct a local parameterization by taking the Cartesian product
// of a number of other local parameterizations. This is useful, when
// a parameter block is the cartesian product of two or more
// manifolds. For example the parameters of a camera consist of a
// rotation and a translation, i.e., SO(3) x R^3.
//
// Example usage:
//
// ProductParameterization product_param(new QuaterionionParameterization(),
// new IdentityParameterization(3));
//
// is the local parameterization for a rigid transformation, where the
// rotation is represented using a quaternion.
class CERES_EXPORT ProductParameterization : public LocalParameterization {
public:
ProductParameterization(const ProductParameterization&) = delete;
ProductParameterization& operator=(const ProductParameterization&) = delete;
//
// NOTE: The constructor takes ownership of the input local
// parameterizations.
//
template <typename... LocalParams>
ProductParameterization(LocalParams*... local_params)
: local_params_(sizeof...(LocalParams)),
local_size_{0},
global_size_{0},
buffer_size_{0} {
constexpr int kNumLocalParams = sizeof...(LocalParams);
static_assert(kNumLocalParams >= 2,
"At least two local parameterizations must be specified.");
using LocalParameterizationPtr = std::unique_ptr<LocalParameterization>;
// Wrap all raw pointers into std::unique_ptr for exception safety.
std::array<LocalParameterizationPtr, kNumLocalParams> local_params_array{
LocalParameterizationPtr(local_params)...};
// Initialize internal state.
for (int i = 0; i < kNumLocalParams; ++i) {
LocalParameterizationPtr& param = local_params_[i];
param = std::move(local_params_array[i]);
buffer_size_ =
std::max(buffer_size_, param->LocalSize() * param->GlobalSize());
global_size_ += param->GlobalSize();
local_size_ += param->LocalSize();
}
}
bool Plus(const double* x,
const double* delta,
double* x_plus_delta) const override;
bool ComputeJacobian(const double* x, double* jacobian) const override;
int GlobalSize() const override { return global_size_; }
int LocalSize() const override { return local_size_; }
private:
std::vector<std::unique_ptr<LocalParameterization>> local_params_;
int local_size_;
int global_size_;
int buffer_size_;
};
} // namespace ceres
#include "ceres/internal/reenable_warnings.h"
#include "ceres/internal/line_parameterization.h"
#endif // CERES_PUBLIC_LOCAL_PARAMETERIZATION_H_