prosac.hpp
Go to the documentation of this file.
00001 /*
00002  * Software License Agreement (BSD License)
00003  *
00004  *  Copyright (c) 2009, Willow Garage, Inc.
00005  *  All rights reserved.
00006  *
00007  *  Redistribution and use in source and binary forms, with or without
00008  *  modification, are permitted provided that the following conditions
00009  *  are met:
00010  *
00011  *   * Redistributions of source code must retain the above copyright
00012  *     notice, this list of conditions and the following disclaimer.
00013  *   * Redistributions in binary form must reproduce the above
00014  *     copyright notice, this list of conditions and the following
00015  *     disclaimer in the documentation and/or other materials provided
00016  *     with the distribution.
00017  *   * Neither the name of Willow Garage, Inc. nor the names of its
00018  *     contributors may be used to endorse or promote products derived
00019  *     from this software without specific prior written permission.
00020  *
00021  *  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
00022  *  "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
00023  *  LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
00024  *  FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
00025  *  COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
00026  *  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
00027  *  BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
00028  *  LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
00029  *  CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
00030  *  LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
00031  *  ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
00032  *  POSSIBILITY OF SUCH DAMAGE.
00033  *
00034  * $Id: prosac.hpp 5026 2012-03-12 02:51:44Z rusu $
00035  *
00036  */
00037 
00038 #ifndef PCL_SAMPLE_CONSENSUS_IMPL_PROSAC_H_
00039 #define PCL_SAMPLE_CONSENSUS_IMPL_PROSAC_H_
00040 
00041 #include <boost/math/distributions/binomial.hpp>
00042 #include <pcl/sample_consensus/prosac.h>
00043 
00045 // Variable naming uses capital letters to make the comparison with the original paper easier
00046 template<typename PointT> bool 
00047 pcl::ProgressiveSampleConsensus<PointT>::computeModel (int debug_verbosity_level)
00048 {
00049   // Warn and exit if no threshold was set
00050   if (threshold_ == DBL_MAX)
00051   {
00052     PCL_ERROR ("[pcl::ProgressiveSampleConsensus::computeModel] No threshold set!\n");
00053     return (false);
00054   }
00055 
00056   // Initialize some PROSAC constants
00057   const int T_N = 200000;
00058   const size_t N = sac_model_->indices_->size ();
00059   const size_t m = sac_model_->getSampleSize ();
00060   float T_n = static_cast<float> (T_N);
00061   for (unsigned int i = 0; i < m; ++i)
00062     T_n *= static_cast<float> (m - i) / static_cast<float> (N - i);
00063   float T_prime_n = 1.0f;
00064   size_t I_N_best = 0;
00065   float n = static_cast<float> (m);
00066 
00067   // Define the n_Start coefficients from Section 2.2
00068   float n_star = static_cast<float> (N);
00069   float epsilon_n_star = 0.0;
00070   size_t k_n_star = T_N;
00071 
00072   // Compute the I_n_star_min of Equation 8
00073   std::vector<unsigned int> I_n_star_min (N);
00074 
00075   // Initialize the usual RANSAC parameters
00076   iterations_ = 0;
00077 
00078   std::vector<int> inliers;
00079   std::vector<int> selection;
00080   Eigen::VectorXf model_coefficients;
00081 
00082   // We will increase the pool so the indices_ vector can only contain m elements at first
00083   std::vector<int> index_pool;
00084   index_pool.reserve (N);
00085   for (unsigned int i = 0; i < n; ++i)
00086     index_pool.push_back (sac_model_->indices_->operator[](i));
00087 
00088   // Iterate
00089   while (static_cast<unsigned int> (iterations_) < k_n_star)
00090   {
00091     // Choose the samples
00092 
00093     // Step 1
00094     // According to Equation 5 in the text text, not the algorithm
00095     if ((iterations_ == T_prime_n) && (n < n_star))
00096     {
00097       // Increase the pool
00098       ++n;
00099       if (n >= N)
00100         break;
00101       index_pool.push_back (sac_model_->indices_->at(static_cast<unsigned int> (n - 1)));
00102       // Update other variables
00103       float T_n_minus_1 = T_n;
00104       T_n *= (static_cast<float>(n) + 1.0f) / (static_cast<float>(n) + 1.0f - static_cast<float>(m));
00105       T_prime_n += ceilf (T_n - T_n_minus_1);
00106     }
00107 
00108     // Step 2
00109     sac_model_->indices_->swap (index_pool);
00110     selection.clear ();
00111     sac_model_->getSamples (iterations_, selection);
00112     if (T_prime_n < iterations_)
00113     {
00114       selection.pop_back ();
00115       selection.push_back (sac_model_->indices_->at(static_cast<unsigned int> (n - 1)));
00116     }
00117 
00118     // Make sure we use the right indices for testing
00119     sac_model_->indices_->swap (index_pool);
00120 
00121     if (selection.empty ())
00122     {
00123       PCL_ERROR ("[pcl::ProgressiveSampleConsensus::computeModel] No samples could be selected!\n");
00124       break;
00125     }
00126 
00127     // Search for inliers in the point cloud for the current model
00128     if (!sac_model_->computeModelCoefficients (selection, model_coefficients))
00129     {
00130       ++iterations_;
00131       continue;
00132     }
00133 
00134     // Select the inliers that are within threshold_ from the model
00135     inliers.clear ();
00136     sac_model_->selectWithinDistance (model_coefficients, threshold_, inliers);
00137 
00138     size_t I_N = inliers.size ();
00139 
00140     // If we find more inliers than before
00141     if (I_N > I_N_best)
00142     {
00143       I_N_best = I_N;
00144 
00145       // Save the current model/inlier/coefficients selection as being the best so far
00146       inliers_ = inliers;
00147       model_ = selection;
00148       model_coefficients_ = model_coefficients;
00149 
00150       // We estimate I_n_star for different possible values of n_star by using the inliers
00151       std::sort (inliers.begin (), inliers.end ());
00152 
00153       // Try to find a better n_star
00154       // We minimize k_n_star and therefore maximize epsilon_n_star = I_n_star / n_star
00155       size_t possible_n_star_best = N, I_possible_n_star_best = I_N;
00156       float epsilon_possible_n_star_best = static_cast<float>(I_possible_n_star_best) / static_cast<float>(possible_n_star_best);
00157 
00158       // We only need to compute possible better epsilon_n_star for when _n is just about to be removed an inlier
00159       size_t I_possible_n_star = I_N;
00160       for (std::vector<int>::const_reverse_iterator last_inlier = inliers.rbegin (), 
00161                                                     inliers_end = inliers.rend (); 
00162            last_inlier != inliers_end; 
00163            ++last_inlier, --I_possible_n_star)
00164       {
00165         // The best possible_n_star for a given I_possible_n_star is the index of the last inlier
00166         unsigned int possible_n_star = (*last_inlier) + 1;
00167         if (possible_n_star <= m)
00168           break;
00169 
00170         // If we find a better epsilon_n_star
00171         float epsilon_possible_n_star = static_cast<float>(I_possible_n_star) / static_cast<float>(possible_n_star);
00172         // Make sure we have a better epsilon_possible_n_star
00173         if ((epsilon_possible_n_star > epsilon_n_star) && (epsilon_possible_n_star > epsilon_possible_n_star_best))
00174         {
00175           using namespace boost::math;
00176           // Typo in Equation 7, not (n-m choose i-m) but (n choose i-m)
00177           size_t I_possible_n_star_min = m
00178                            + static_cast<size_t> (ceil (quantile (complement (binomial_distribution<float>(static_cast<float> (possible_n_star), 0.1f), 0.05))));
00179           // If Equation 9 is not verified, exit
00180           if (I_possible_n_star < I_possible_n_star_min)
00181             break;
00182 
00183           possible_n_star_best = possible_n_star;
00184           I_possible_n_star_best = I_possible_n_star;
00185           epsilon_possible_n_star_best = epsilon_possible_n_star;
00186         }
00187       }
00188 
00189       // Check if we get a better epsilon
00190       if (epsilon_possible_n_star_best > epsilon_n_star)
00191       {
00192         // update the best value
00193         epsilon_n_star = epsilon_possible_n_star_best;
00194 
00195         // Compute the new k_n_star
00196         float bottom_log = 1 - std::pow (epsilon_n_star, static_cast<float>(m));
00197         if (bottom_log == 0)
00198           k_n_star = 1;
00199         else if (bottom_log == 1)
00200           k_n_star = T_N;
00201         else
00202           k_n_star = static_cast<int> (ceil (log (0.05) / log (bottom_log)));
00203         // It seems weird to have very few iterations, so do have a few (totally empirical)
00204         k_n_star = (std::max)(k_n_star, 2 * m);
00205       }
00206     }
00207 
00208     ++iterations_;
00209     if (debug_verbosity_level > 1)
00210       PCL_DEBUG ("[pcl::ProgressiveSampleConsensus::computeModel] Trial %d out of %d: %d inliers (best is: %d so far).\n", iterations_, k_n_star, I_N, I_N_best);
00211     if (iterations_ > max_iterations_)
00212     {
00213       if (debug_verbosity_level > 0)
00214         PCL_DEBUG ("[pcl::ProgressiveSampleConsensus::computeModel] RANSAC reached the maximum number of trials.\n");
00215       break;
00216     }
00217   }
00218 
00219   if (debug_verbosity_level > 0)
00220     PCL_DEBUG ("[pcl::ProgressiveSampleConsensus::computeModel] Model: %zu size, %d inliers.\n", model_.size (), I_N_best);
00221 
00222   if (model_.empty ())
00223   {
00224     inliers_.clear ();
00225     return (false);
00226   }
00227 
00228   // Get the set of inliers that correspond to the best model found so far
00229   //sac_model_->selectWithinDistance (model_coefficients_, threshold_, inliers_);
00230   return (true);
00231 }
00232 
00233 #define PCL_INSTANTIATE_ProgressiveSampleConsensus(T) template class PCL_EXPORTS pcl::ProgressiveSampleConsensus<T>;
00234 
00235 #endif    // PCL_SAMPLE_CONSENSUS_IMPL_PROSAC_H_


pcl
Author(s): Open Perception
autogenerated on Mon Oct 6 2014 03:17:28