[PCL]模型拟合方法——随机采样一致性

SACSegmentation封装了多种Ransac方法，包括：

RandomSampleConsensus，

LeastMedianSquares，

MEstimatorSampleConsensus

ProgressiveSampleConsensus，

RandomizedRandomSampleConsensus，

RandomizedMEstimatorSampleConsensus，

MaximumLikelihoodSampleConsensus

1.PCL所谓的平行线判断，是已知一个法向量，判断面与之平行。

2.PCL直线Ransac拟合，为啥只需要设置一个距离阈值？因为默认值迭代50次

 

template <typename PointT> void
pcl::SACSegmentation<PointT>::initSAC (const int method_type)
{
  if (sac_)
    sac_.reset ();
  // Build the sample consensus method
  switch (method_type)
  {
    case SAC_RANSAC:
    default:
    {
      PCL_DEBUG ("[pcl::%s::initSAC] Using a method of type: SAC_RANSAC with a model threshold of %f
", getClassName ().c_str (), threshold_);
      sac_.reset (new RandomSampleConsensus<PointT> (model_, threshold_));
      break;
    }
    case SAC_LMEDS:
    {
      PCL_DEBUG ("[pcl::%s::initSAC] Using a method of type: SAC_LMEDS with a model threshold of %f
", getClassName ().c_str (), threshold_);
      sac_.reset (new LeastMedianSquares<PointT> (model_, threshold_));
      break;
    }
    case SAC_MSAC:
    {
      PCL_DEBUG ("[pcl::%s::initSAC] Using a method of type: SAC_MSAC with a model threshold of %f
", getClassName ().c_str (), threshold_);
      sac_.reset (new MEstimatorSampleConsensus<PointT> (model_, threshold_));
      break;
    }
    case SAC_RRANSAC:
    {
      PCL_DEBUG ("[pcl::%s::initSAC] Using a method of type: SAC_RRANSAC with a model threshold of %f
", getClassName ().c_str (), threshold_);
      sac_.reset (new RandomizedRandomSampleConsensus<PointT> (model_, threshold_));
      break;
    }
    case SAC_RMSAC:
    {
      PCL_DEBUG ("[pcl::%s::initSAC] Using a method of type: SAC_RMSAC with a model threshold of %f
", getClassName ().c_str (), threshold_);
      sac_.reset (new RandomizedMEstimatorSampleConsensus<PointT> (model_, threshold_));
      break;
    }
    case SAC_MLESAC:
    {
      PCL_DEBUG ("[pcl::%s::initSAC] Using a method of type: SAC_MLESAC with a model threshold of %f
", getClassName ().c_str (), threshold_);
      sac_.reset (new MaximumLikelihoodSampleConsensus<PointT> (model_, threshold_));
      break;
    }
    case SAC_PROSAC:
    {
      PCL_DEBUG ("[pcl::%s::initSAC] Using a method of type: SAC_PROSAC with a model threshold of %f
", getClassName ().c_str (), threshold_);
      sac_.reset (new ProgressiveSampleConsensus<PointT> (model_, threshold_));
      break;
    }
  }
  // Set the Sample Consensus parameters if they are given/changed
  if (sac_->getProbability () != probability_)
  {
    PCL_DEBUG ("[pcl::%s::initSAC] Setting the desired probability to %f
", getClassName ().c_str (), probability_);
    sac_->setProbability (probability_);
  }
  if (max_iterations_ != -1 && sac_->getMaxIterations () != max_iterations_)
  {
    PCL_DEBUG ("[pcl::%s::initSAC] Setting the maximum number of iterations to %d
", getClassName ().c_str (), max_iterations_);
    sac_->setMaxIterations (max_iterations_);
  }
  if (samples_radius_ > 0.)
  {
    PCL_DEBUG ("[pcl::%s::initSAC] Setting the maximum sample radius to %f
", getClassName ().c_str (), samples_radius_);
    // Set maximum distance for radius search during random sampling
    model_->setSamplesMaxDist (samples_radius_, samples_radius_search_);
  }
}