102 lines
3.7 KiB
C++
102 lines
3.7 KiB
C++
/*
|
|
* Copyright (C) 2014 Open Source Robotics Foundation
|
|
*
|
|
* Licensed under the Apache License, Version 2.0 (the "License");
|
|
* you may not use this file except in compliance with the License.
|
|
* You may obtain a copy of the License at
|
|
*
|
|
* http://www.apache.org/licenses/LICENSE-2.0
|
|
*
|
|
* Unless required by applicable law or agreed to in writing, software
|
|
* distributed under the License is distributed on an "AS IS" BASIS,
|
|
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
* See the License for the specific language governing permissions and
|
|
* limitations under the License.
|
|
*
|
|
*/
|
|
#ifndef _GAZEBO_KMEANS_HH_
|
|
#define _GAZEBO_KMEANS_HH_
|
|
|
|
#include <vector>
|
|
#include "gazebo/math/Vector3.hh"
|
|
#include "gazebo/util/system.hh"
|
|
|
|
namespace gazebo
|
|
{
|
|
namespace math
|
|
{
|
|
/// \addtogroup gazebo_math
|
|
/// \{
|
|
|
|
/// \class Kmeans Kmeans.hh math/gzmath.hh
|
|
/// \brief K-Means clustering algorithm. Given a set of observations,
|
|
/// k-means partitions the observations into k sets so as to minimize the
|
|
/// within-cluster sum of squares.
|
|
/// Description based on http://en.wikipedia.org/wiki/K-means_clustering.
|
|
class GZ_MATH_VISIBLE Kmeans
|
|
{
|
|
/// \brief constructor
|
|
/// \param[in] _obs Set of observations to cluster.
|
|
public: Kmeans(const std::vector<Vector3> &_obs);
|
|
|
|
/// \brief Destructor
|
|
public: virtual ~Kmeans();
|
|
|
|
/// \brief Get the observations to cluster.
|
|
/// \return The vector of observations.
|
|
public: std::vector<Vector3> Observations() const;
|
|
|
|
/// \brief Set the observations to cluster.
|
|
/// \param[in] _obs The new vector of observations.
|
|
/// \return True if the vector is not empty or false otherwise.
|
|
public: bool Observations(const std::vector<Vector3> &_obs);
|
|
|
|
/// \brief Add observations to the cluster.
|
|
/// \param[in] _obs Vector of observations.
|
|
/// \return True if the _obs vector is not empty or false otherwise.
|
|
public: bool AppendObservations(const std::vector<Vector3> &_obs);
|
|
|
|
/// \brief Executes the k-means algorithm.
|
|
/// \param[in] _k Number of partitions to cluster.
|
|
/// \param[out] _centroids Vector of centroids. Each element contains the
|
|
/// centroid of one cluster.
|
|
/// \param[out] _labels Vector of labels. The size of this vector is
|
|
/// equals to the number of observations. Each element represents the
|
|
/// cluster to which observation belongs.
|
|
/// \return True when the operation succeed or false otherwise. The
|
|
/// operation will fail if the number of observations is not positive,
|
|
/// if the number of clusters is non positive, or if the number of
|
|
/// clusters if greater than the number of observations.
|
|
public: bool Cluster(int _k,
|
|
std::vector<Vector3> &_centroids,
|
|
std::vector<unsigned int> &_labels);
|
|
|
|
/// \brief Given an observation, it returns the closest centroid to it.
|
|
/// \param[in] _p Point to check.
|
|
/// \return The index of the closest centroid to the point _p.
|
|
private: unsigned int ClosestCentroid(const Vector3 &_p) const;
|
|
|
|
/// \brief Observations.
|
|
private: std::vector<Vector3> obs;
|
|
|
|
/// \brief Centroids.
|
|
private: std::vector<Vector3> centroids;
|
|
|
|
/// \brief Each element stores the cluster to which observation i belongs.
|
|
private: std::vector<unsigned int> labels;
|
|
|
|
/// \brief Used to calculate the centroid of each partition.
|
|
private: std::vector<Vector3> sums;
|
|
|
|
/// \brief Counts the number of observations contained in each partition.
|
|
private: std::vector<unsigned int> counters;
|
|
};
|
|
/// \}
|
|
}
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|