FastLineRecognition Class Reference

Class performing optimized hough transform to recognize lines. More...

#include <FastLineRecognition.h>

Classes
struct	Cluster
	cluster of intersection points More...
struct	GeomData
struct	Point

Public Member Functions
	FastLineRecognition (double cw_a=0., double cw_b=0.)
void	getPatterns (const edm::DetSetVector< TotemRPRecHit > &input, double _z0, double threshold, edm::DetSet< TotemRPUVPattern > &patterns)
void	resetGeometry (const CTPPSGeometry *_g)
	~FastLineRecognition ()

Private Member Functions
GeomData	getGeomData (unsigned int id)
	expects raw detector id More...
bool	getOneLine (const std::vector< Point > &points, double threshold, Cluster &result)

Private Attributes
double	chw_a
	cluster half widths in a and b More...
double	chw_b
const CTPPSGeometry *	geometry
	pointer to the geometry More...
std::map< unsigned int, GeomData >	geometryMap
	map: raw detector id --> GeomData More...
double	threshold
	weight threshold for accepting pattern candidates (clusters) More...
double	z0
	"typical" z More...

Static Private Attributes
static const double	sigma0 = 66E-3 / sqrt(12.)
	the uncertainty of 1-hit cluster, in mm More...

Detailed Description

Class performing optimized hough transform to recognize lines.

Definition at line 23 of file FastLineRecognition.h.

Constructor & Destructor Documentation

FastLineRecognition()

FastLineRecognition::FastLineRecognition	(	double	cw_a = 0.,
		double	cw_b = 0.
	)

Definition at line 57 of file FastLineRecognition.cc.

    : chw_a(cw_a / 2.), chw_b(cw_b / 2.), geometry(nullptr) {}

~FastLineRecognition()

FastLineRecognition::~FastLineRecognition

(

)

Definition at line 62 of file FastLineRecognition.cc.

Member Function Documentation

getGeomData()

FastLineRecognition::GeomData FastLineRecognition::getGeomData ( unsigned int  id )

private

expects raw detector id

Definition at line 66 of file FastLineRecognition.cc.

                                                                            {
  // result already buffered?
  map<unsigned int, GeomData>::iterator it = geometryMap.find(id);
  if (it != geometryMap.end())
    return it->second;
 
  // calculate it
  const auto &d = geometry->localToGlobalDirection(id, CTPPSGeometry::Vector(0., 1., 0.));
  DetGeomDesc::Translation c = geometry->sensor(TotemRPDetId(id))->translation();
  GeomData gd;
  gd.z = c.z();
  gd.s = d.x() * c.x() + d.y() * c.y();
 
  geometryMap[id] = gd;
 
  return gd;

References c, ztail::d, geometryMap, triggerObjects_cff::id, FastLineRecognition::GeomData::s, and FastLineRecognition::GeomData::z.

Referenced by getPatterns().

getOneLine()

bool FastLineRecognition::getOneLine ( const std::vector< Point > &  points, double  threshold, FastLineRecognition::Cluster &  result  )

private

gets the most significant pattern in the (remaining) points returns true when a pattern was found

Definition at line 173 of file FastLineRecognition.cc.

                                                                         {
#if CTPPS_DEBUG > 0
  printf("\tFastLineRecognition::getOneLine\n");
#endif
 
  if (points.size() < 2)
    return false;
 
  vector<Cluster> clusters;
 
  // go through all the combinations of measured points
  for (vector<Point>::const_iterator it1 = points.begin(); it1 != points.end(); ++it1) {
    if (!it1->usable)
      continue;
 
    for (vector<Point>::const_iterator it2 = it1; it2 != points.end(); ++it2) {
      if (!it2->usable)
        continue;
 
      const double &z1 = it1->z;
      const double &z2 = it2->z;
 
      if (z1 == z2)
        continue;
 
      const double &p1 = it1->h;
      const double &p2 = it2->h;
 
      const double &w1 = it1->w;
      const double &w2 = it2->w;
 
      // calculate intersection
      double a = (p2 - p1) / (z2 - z1);
      double b = p1 - z1 * a;
      double w = w1 + w2;
 
#if CTPPS_DEBUG > 0
      printf("\t\t\tz: 1=%+5.1f, 2=%+5.1f | U/V: 1=%+6.3f, 2=%+6.3f | a=%+6.3f rad, b=%+6.3f mm, w=%.1f\n",
             z1,
             z2,
             p1,
             p2,
             a,
             b,
             w);
#endif
 
      // add it to the appropriate cluster
      bool newCluster = true;
      for (unsigned int k = 0; k < clusters.size(); k++) {
        Cluster &c = clusters[k];
        if (c.S1 < 1. || c.Sw <= 0.)
          continue;
 
#if CTPPS_DEBUG > 0
        if (k < 10)
          printf("\t\t\t\ttest cluster %u at a=%+6.3f, b=%+6.3f : %+6.3f, %+6.3f : %i, %i\n",
                 k,
                 c.Saw / c.Sw,
                 c.Sbw / c.Sw,
                 chw_a,
                 chw_b,
                 (std::abs(a - c.Saw / c.Sw) < chw_a),
                 (std::abs(b - c.Sbw / c.Sw) < chw_b));
#endif
 
        if ((std::abs(a - c.Saw / c.Sw) < chw_a) && (std::abs(b - c.Sbw / c.Sw) < chw_b)) {
          newCluster = false;
          clusters[k].add(&(*it1), &(*it2), a, b, w);
#if CTPPS_DEBUG > 0
          printf("\t\t\t\t--> cluster %u\n", k);
#endif
          break;
        }
      }
 
      // make new cluster
      if (newCluster) {
#if CTPPS_DEBUG > 0
        printf("\t\t\t\t--> new cluster %lu\n", clusters.size());
#endif
        clusters.push_back(Cluster());
        clusters.back().add(&(*it1), &(*it2), a, b, w);
      }
    }
  }
 
#if CTPPS_DEBUG > 0
  printf("\t\tclusters: %lu\n", clusters.size());
#endif
 
  // find the cluster with highest weight
  unsigned int mk = 0;
  double mw = -1.;
  for (unsigned int k = 0; k < clusters.size(); k++) {
    double w = 0;
    for (vector<const Point *>::iterator it = clusters[k].contents.begin(); it != clusters[k].contents.end(); ++it)
      w += (*it)->w;
    clusters[k].weight = w;
 
    if (w > mw) {
      mw = w;
      mk = k;
    }
  }
 
#if CTPPS_DEBUG > 0
  printf("\t\tmw = %.1f, mk = %u\n", mw, mk);
#endif
 
  // rerturn result
  if (mw >= threshold) {
    result = clusters[mk];
 
    return true;
  } else
    return false;

References a, funct::abs(), b, c, chw_a, chw_b, bsc_activity_cfg::clusters, relmon_rootfiles_spy::contents, if(), dqmdumpme::k, p1, p2, HLT_FULL_cff::points, mps_fire::result, threshold, w, w2, and testProducerWithPsetDescEmpty_cfi::z2.

Referenced by getPatterns().

getPatterns()

void FastLineRecognition::getPatterns	(	const edm::DetSetVector< TotemRPRecHit > &	input,
		double	_z0,
		double	threshold,
		edm::DetSet< TotemRPUVPattern > &	patterns
	)

Definition at line 86 of file FastLineRecognition.cc.

                                                                          {
  // build collection of points in the global coordinate system
  std::vector<Point> points;
  for (auto &ds : input) {
    unsigned int detId = ds.detId();
 
    for (auto &h : ds) {
      const TotemRPRecHit *hit = &h;
      const GeomData &gd = getGeomData(detId);
 
      double p = hit->position() + gd.s;
      double z = gd.z - z0;
      double w = sigma0 / hit->sigma();
 
      points.push_back(Point(detId, hit, p, z, w));
    }
  }
 
#if CTPPS_DEBUG > 0
  printf(">> FastLineRecognition::getPatterns(z0 = %E)\n", z0);
  printf(">>>>>>>>>>>>>>>>>>>\n");
#endif
 
  // reset output
  patterns.clear();
 
  Cluster c;
  while (getOneLine(points, threshold, c)) {
    // convert cluster to pattern and save it
    TotemRPUVPattern pattern;
    pattern.setA(c.Saw / c.Sw);
    pattern.setB(c.Sbw / c.Sw);
    pattern.setW(c.weight);
 
#if CTPPS_DEBUG > 0
    printf("\tpoints of the selected cluster: %lu\n", c.contents.size());
#endif
 
    for (auto &pit : c.contents) {
#if CTPPS_DEBUG > 0
      printf("\t\t%.1f\n", pit->z);
#endif
      pattern.addHit(pit->detId, *(pit->hit));
    }
 
    patterns.push_back(pattern);
 
#if CTPPS_DEBUG > 0
    unsigned int u_points_b = 0;
    for (vector<Point>::iterator dit = points.begin(); dit != points.end(); ++dit)
      if (dit->usable)
        u_points_b++;
    printf("\tusable points before: %u\n", u_points_b);
#endif
 
    // remove points belonging to the recognized line
    for (vector<const Point *>::iterator hit = c.contents.begin(); hit != c.contents.end(); ++hit) {
      for (vector<Point>::iterator dit = points.begin(); dit != points.end(); ++dit) {
        //printf("\t\t1: %.2f, %p vs. 2: %.2f, %p\n", (*hit)->z, (*hit)->hit, dit->z, dit->hit);
        if ((*hit)->hit == dit->hit) {
          dit->usable = false;
          //points.erase(dit);
          break;
        }
      }
    }
 
#if CTPPS_DEBUG > 0
    unsigned int u_points_a = 0;
    for (vector<Point>::iterator dit = points.begin(); dit != points.end(); ++dit)
      if (dit->usable)
        u_points_a++;
    printf("\tusable points after: %u\n", u_points_a);
#endif
  }
 
#if CTPPS_DEBUG > 0
  printf("patterns at end: %lu\n", patterns.size());
  printf("<<<<<<<<<<<<<<<<<<<\n");
#endif

References c, edm::DetSet< T >::clear(), getGeomData(), getOneLine(), h, input, AlCaHLTBitMon_ParallelJobs::p, topSingleLeptonDQM_PU_cfi::pattern, HLT_FULL_cff::points, edm::DetSet< T >::push_back(), FastLineRecognition::GeomData::s, sigma0, edm::DetSet< T >::size(), threshold, w, z, FastLineRecognition::GeomData::z, and z0.

Referenced by TotemRPUVPatternFinder::recognizeAndSelect().

resetGeometry()

void FastLineRecognition::resetGeometry ( const CTPPSGeometry *  _g )

inline

Definition at line 34 of file FastLineRecognition.h.

Referenced by TotemRPUVPatternFinder::produce().

Member Data Documentation

chw_a

double FastLineRecognition::chw_a

private

cluster half widths in a and b

Definition at line 52 of file FastLineRecognition.h.

Referenced by getOneLine().

chw_b

double FastLineRecognition::chw_b

private

Definition at line 52 of file FastLineRecognition.h.

Referenced by getOneLine().

geometry

const CTPPSGeometry* FastLineRecognition::geometry

private

pointer to the geometry

Definition at line 58 of file FastLineRecognition.h.

geometryMap

std::map<unsigned int, GeomData> FastLineRecognition::geometryMap

private

map: raw detector id --> GeomData

Definition at line 66 of file FastLineRecognition.h.

Referenced by getGeomData().

sigma0

const double FastLineRecognition::sigma0 = 66E-3 / sqrt(12.)

staticprivate

the uncertainty of 1-hit cluster, in mm

Definition at line 46 of file FastLineRecognition.h.

Referenced by getPatterns().

threshold

double FastLineRecognition::threshold

private

weight threshold for accepting pattern candidates (clusters)

Definition at line 55 of file FastLineRecognition.h.

Referenced by utils.StatisticalTest::get_status(), getOneLine(), and getPatterns().

z0

double FastLineRecognition::z0

private

"typical" z

Definition at line 49 of file FastLineRecognition.h.

Referenced by getPatterns().