source: trunk/src/STLineFinder.cpp @ 2345

Last change on this file since 2345 was 2345, checked in by WataruKawasaki, 13 years ago

New Development: No

JIRA Issue: No

Ready for Test: Yes

Interface Changes: No

What Interface Changed:

Test Programs:

Put in Release Notes: No

Module(s): SD

Description: bugfix.


  • Property svn:eol-style set to native
  • Property svn:keywords set to Author Date Id Revision
File size: 52.6 KB
RevLine 
[297]1//#---------------------------------------------------------------------------
[881]2//# STLineFinder.cc: A class for automated spectral line search
[297]3//#--------------------------------------------------------------------------
4//# Copyright (C) 2004
5//# ATNF
6//#
7//# This program is free software; you can redistribute it and/or modify it
8//# under the terms of the GNU General Public License as published by the Free
9//# Software Foundation; either version 2 of the License, or (at your option)
10//# any later version.
11//#
12//# This program is distributed in the hope that it will be useful, but
13//# WITHOUT ANY WARRANTY; without even the implied warranty of
14//# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General
15//# Public License for more details.
16//#
17//# You should have received a copy of the GNU General Public License along
18//# with this program; if not, write to the Free Software Foundation, Inc.,
19//# 675 Massachusetts Ave, Cambridge, MA 02139, USA.
20//#
21//# Correspondence concerning this software should be addressed as follows:
22//#        Internet email: Malte.Marquarding@csiro.au
23//#        Postal address: Malte Marquarding,
24//#                        Australia Telescope National Facility,
25//#                        P.O. Box 76,
26//#                        Epping, NSW, 2121,
27//#                        AUSTRALIA
28//#
[890]29//# $Id: STLineFinder.cpp 2345 2011-11-08 08:03:54Z WataruKawasaki $
[297]30//#---------------------------------------------------------------------------
31
32
33// ASAP
[894]34#include "STLineFinder.h"
35#include "STFitter.h"
[1642]36#include "IndexedCompare.h"
[297]37
38// STL
[343]39#include <functional>
40#include <algorithm>
[297]41#include <iostream>
[351]42#include <fstream>
[297]43
44using namespace asap;
45using namespace casa;
46using namespace std;
47
[344]48namespace asap {
49
[343]50///////////////////////////////////////////////////////////////////////////////
51//
[881]52// RunningBox -    a running box calculator. This class implements
[1315]53//                 iterations over the specified spectrum and calculates
[351]54//                 running box filter statistics.
[343]55//
56
[351]57class RunningBox {
[331]58   // The input data to work with. Use reference symantics to avoid
[881]59   // an unnecessary copying
[331]60   const casa::Vector<casa::Float>  &spectrum; // a buffer for the spectrum
61   const casa::Vector<casa::Bool>   &mask; // associated mask
62   const std::pair<int,int>         &edge; // start and stop+1 channels
63                                           // to work with
[881]64
[351]65   // statistics for running box filtering
66   casa::Float sumf;       // sum of fluxes
67   casa::Float sumf2;     // sum of squares of fluxes
68   casa::Float sumch;       // sum of channel numbers (for linear fit)
69   casa::Float sumch2;     // sum of squares of channel numbers (for linear fit)
70   casa::Float sumfch;     // sum of flux*(channel number) (for linear fit)
[881]71
[331]72   int box_chan_cntr;     // actual number of channels in the box
73   int max_box_nchan;     // maximum allowed number of channels in the box
74                          // (calculated from boxsize and actual spectrum size)
[351]75   // cache for derivative statistics
76   mutable casa::Bool need2recalculate; // if true, values of the statistics
77                                       // below are invalid
78   mutable casa::Float linmean;  // a value of the linear fit to the
79                                 // points in the running box
80   mutable casa::Float linvariance; // the same for variance
81   int cur_channel;       // the number of the current channel
82   int start_advance;     // number of channel from which the box can
83                          // be moved (the middle of the box, if there is no
[996]84                          // masking)
[351]85public:
86   // set up the object with the references to actual data
87   // as well as the number of channels in the running box
88   RunningBox(const casa::Vector<casa::Float>  &in_spectrum,
89                 const casa::Vector<casa::Bool>   &in_mask,
[996]90                 const std::pair<int,int>         &in_edge,
91                 int in_max_box_nchan) throw(AipsError);
[881]92
[351]93   // access to the statistics
94   const casa::Float& getLinMean() const throw(AipsError);
95   const casa::Float& getLinVariance() const throw(AipsError);
[2163]96   casa::Float aboveMean() const throw(AipsError);
[351]97   int getChannel() const throw();
[881]98
[351]99   // actual number of channels in the box (max_box_nchan, if no channels
100   // are masked)
101   int getNumberOfBoxPoints() const throw();
[297]102
[351]103   // next channel
104   void next() throw(AipsError);
105
106   // checking whether there are still elements
107   casa::Bool haveMore() const throw();
108
109   // go to start
110   void rewind() throw(AipsError);
[881]111
[351]112protected:
[1644]113   // supplementary function to control running mean/median calculations.
114   // It adds a specified channel to the running box and
[351]115   // removes (ch-maxboxnchan+1)'th channel from there
116   // Channels, for which the mask is false or index is beyond the
117   // allowed range, are ignored
118   void advanceRunningBox(int ch) throw(casa::AipsError);
119
120   // calculate derivative statistics. This function is const, because
121   // it updates the cache only
122   void updateDerivativeStatistics() const throw(AipsError);
123};
124
125//
126///////////////////////////////////////////////////////////////////////////////
127
128///////////////////////////////////////////////////////////////////////////////
129//
130// LFAboveThreshold   An algorithm for line detection using running box
131//                    statistics.  Line is detected if it is above the
132//                    specified threshold at the specified number of
133//                    consequtive channels. Prefix LF stands for Line Finder
134//
[352]135class LFAboveThreshold : protected LFLineListOperations {
[331]136   // temporary line edge channels and flag, which is True if the line
137   // was detected in the previous channels.
138   std::pair<int,int> cur_line;
139   casa::Bool is_detected_before;
140   int  min_nchan;                         // A minimum number of consequtive
141                                           // channels, which should satisfy
[996]142                                           // the detection criterion, to be
143                                           // a detection
[881]144   casa::Float threshold;                  // detection threshold - the
[331]145                                           // minimal signal to noise ratio
[351]146   std::list<pair<int,int> > &lines;       // list where detections are saved
147                                           // (pair: start and stop+1 channel)
148   RunningBox *running_box;                // running box filter
[551]149   casa::Vector<Int> signs;                // An array to store the signs of
150                                           // the value - current mean
[996]151                                           // (used to search wings)
[907]152   casa::Int last_sign;                    // a sign (+1, -1 or 0) of the
153                                           // last point of the detected line
154                                           //
[1644]155   bool itsUseMedian;                      // true if median statistics is used
156                                           // to determine the noise level, otherwise
157                                           // it is the mean of the lowest 80% of deviations
158                                           // (default)
159   int itsNoiseSampleSize;                 // sample size used to estimate the noise statistics
160                                           // Negative value means the whole spectrum is used (default)
[331]161public:
[351]162
163   // set up the detection criterion
164   LFAboveThreshold(std::list<pair<int,int> > &in_lines,
165                    int in_min_nchan = 3,
[1644]166                    casa::Float in_threshold = 5,
167                    bool use_median = false,
168                    int noise_sample_size = -1) throw();
[351]169   virtual ~LFAboveThreshold() throw();
[881]170
[331]171   // replace the detection criterion
172   void setCriterion(int in_min_nchan, casa::Float in_threshold) throw();
[297]173
[551]174   // return the array with signs of the value-current mean
175   // An element is +1 if value>mean, -1 if less, 0 if equal.
176   // This array is updated each time the findLines method is called and
177   // is used to search the line wings
178   const casa::Vector<Int>& getSigns() const throw();
179
[331]180   // find spectral lines and add them into list
[344]181   // if statholder is not NULL, the accumulate function of it will be
182   // called for each channel to save statistics
[351]183   //    spectrum, mask and edge - reference to the data
184   //    max_box_nchan  - number of channels in the running box
185   void findLines(const casa::Vector<casa::Float> &spectrum,
[996]186                  const casa::Vector<casa::Bool> &mask,
187                  const std::pair<int,int> &edge,
188                  int max_box_nchan) throw(casa::AipsError);
[351]189
[331]190protected:
[297]191
[331]192   // process a channel: update curline and is_detected before and
193   // add a new line to the list, if necessary using processCurLine()
[351]194   // detect=true indicates that the current channel satisfies the criterion
195   void processChannel(Bool detect, const casa::Vector<casa::Bool> &mask)
196                                                  throw(casa::AipsError);
[297]197
[331]198   // process the interval of channels stored in curline
199   // if it satisfies the criterion, add this interval as a new line
[351]200   void processCurLine(const casa::Vector<casa::Bool> &mask)
201                                                 throw(casa::AipsError);
[924]202
[907]203   // get the sign of runningBox->aboveMean(). The RunningBox pointer
204   // should be defined
205   casa::Int getAboveMeanSign() const throw();
[331]206};
[344]207
208//
209///////////////////////////////////////////////////////////////////////////////
[351]210
[1642]211///////////////////////////////////////////////////////////////////////////////
212//
213// LFNoiseEstimator  a helper class designed to estimate off-line variance
214//                   using statistics depending on the distribution of
215//                   values (e.g. like a median)
216//
217//                   Two statistics are supported: median and an average of
218//                   80% of smallest values.
219//
220
221struct LFNoiseEstimator {
222   // construct an object
223   // size - maximum sample size. After a size number of elements is processed
224   // any new samples would cause the algorithm to drop the oldest samples in the
225   // buffer.
226   explicit LFNoiseEstimator(size_t size);
227
228   // add a new sample
229   // in - the new value
230   void add(float in);
231
232   // median of the distribution
233   float median() const;
234
235   // mean of lowest 80% of the samples
236   float meanLowest80Percent() const;
237
[1644]238   // return true if the buffer is full (i.e. statistics are representative)
239   inline bool filledToCapacity() const { return itsBufferFull;}
240
[1642]241protected:
242   // update cache of sorted indices
243   // (it is assumed that itsSampleNumber points to the newly
244   // replaced element)
245   void updateSortedCache() const;
246
247   // build sorted cache from the scratch
248   void buildSortedCache() const;
249
250   // number of samples accumulated so far
251   // (can be less than the buffer size)
252   size_t numberOfSamples() const;
253
254   // this helper method builds the cache if
255   // necessary using one of the methods
256   void fillCacheIfNecessary() const;
257
258private:
259   // buffer with samples (unsorted)
260   std::vector<float> itsVariances;
261   // current sample number (<=itsVariances.size())
262   size_t itsSampleNumber;
263   // true, if the buffer all values in the sample buffer are used
264   bool itsBufferFull;
265   // cached indices into vector of samples
266   mutable std::vector<size_t> itsSortedIndices;
267   // true if any of the statistics have been obtained at least
268   // once. This flag allows to implement a more efficient way of
269   // calculating statistics, if they are needed at once and not
270   // after each addition of a new element
271   mutable bool itsStatisticsAccessed;
272};
273
274//
275///////////////////////////////////////////////////////////////////////////////
276
277
[331]278} // namespace asap
[297]279
[344]280///////////////////////////////////////////////////////////////////////////////
[343]281//
[1642]282// LFNoiseEstimator  a helper class designed to estimate off-line variance
283//                   using statistics depending on the distribution of
284//                   values (e.g. like a median)
285//
286//                   Two statistics are supported: median and an average of
287//                   80% of smallest values.
288//
289
290// construct an object
291// size - maximum sample size. After a size number of elements is processed
292// any new samples would cause the algorithm to drop the oldest samples in the
293// buffer.
294LFNoiseEstimator::LFNoiseEstimator(size_t size) : itsVariances(size),
295     itsSampleNumber(0), itsBufferFull(false), itsSortedIndices(size),
296     itsStatisticsAccessed(false)
297{
298   AlwaysAssert(size>0,AipsError);
299}
300
301
302// add a new sample
303// in - the new value
304void LFNoiseEstimator::add(float in)
305{
[1670]306   if (isnan(in)) {
307       // normally it shouldn't happen
308       return;
309   }
[1642]310   itsVariances[itsSampleNumber] = in;
311
312   if (itsStatisticsAccessed) {
313       // only do element by element addition if on-the-fly
314       // statistics are needed
315       updateSortedCache();
316   }
317
318   // advance itsSampleNumber now
319   ++itsSampleNumber;
320   if (itsSampleNumber == itsVariances.size()) {
321       itsSampleNumber = 0;
322       itsBufferFull = true;
323   }
324   AlwaysAssert(itsSampleNumber<itsVariances.size(),AipsError);
325}
326
327// number of samples accumulated so far
328// (can be less than the buffer size)
329size_t LFNoiseEstimator::numberOfSamples() const
330{
331  // the number of samples accumulated so far may be less than the
332  // buffer size
333  const size_t nSamples = itsBufferFull ? itsVariances.size(): itsSampleNumber;
[1643]334  AlwaysAssert( (nSamples > 0) && (nSamples <= itsVariances.size()), AipsError);
[1642]335  return nSamples;
336}
337
338// this helper method builds the cache if
339// necessary using one of the methods
340void LFNoiseEstimator::fillCacheIfNecessary() const
341{
342  if (!itsStatisticsAccessed) {
343      if ((itsSampleNumber!=0) || itsBufferFull) {
344          // build the whole cache efficiently
345          buildSortedCache();
346      } else {
347          updateSortedCache();
348      }
349      itsStatisticsAccessed = true;
350  } // otherwise, it is updated in 'add' using on-the-fly method
351}
352
353// median of the distribution
354float LFNoiseEstimator::median() const
355{
356  fillCacheIfNecessary();
357  // the number of samples accumulated so far may be less than the
358  // buffer size
359  const size_t nSamples = numberOfSamples();
360  const size_t medSample = nSamples / 2;
361  AlwaysAssert(medSample < itsSortedIndices.size(), AipsError);
362  return itsVariances[itsSortedIndices[medSample]];
363}
364
365// mean of lowest 80% of the samples
366float LFNoiseEstimator::meanLowest80Percent() const
367{
368  fillCacheIfNecessary();
369  // the number of samples accumulated so far may be less than the
370  // buffer size
371  const size_t nSamples = numberOfSamples();
372  float result = 0;
373  size_t numpt=size_t(0.8*nSamples);
374  if (!numpt) {
375      numpt=nSamples; // no much else left,
376                     // although it is very inaccurate
377  }
378  AlwaysAssert( (numpt > 0) && (numpt<itsSortedIndices.size()), AipsError);
379  for (size_t ch=0; ch<numpt; ++ch) {
380       result += itsVariances[itsSortedIndices[ch]];
381  }
382  result /= float(numpt);
383  return result;
384}
385
386// update cache of sorted indices
387// (it is assumed that itsSampleNumber points to the newly
388// replaced element)
389void LFNoiseEstimator::updateSortedCache() const
390{
391  // the number of samples accumulated so far may be less than the
392  // buffer size
393  const size_t nSamples = numberOfSamples();
394
395  if (itsBufferFull) {
396      // first find the index of the element which is being replaced
397      size_t index = nSamples;
398      for (size_t i=0; i<nSamples; ++i) {
399           AlwaysAssert(i < itsSortedIndices.size(), AipsError);
400           if (itsSortedIndices[i] == itsSampleNumber) {
401               index = i;
402               break;
403           }
404      }
405      AlwaysAssert( index < nSamples, AipsError);
406
407      const vector<size_t>::iterator indStart = itsSortedIndices.begin();
408      // merge this element with preceeding block first
409      if (index != 0) {
410          // merge indices on the basis of variances
411          inplace_merge(indStart,indStart+index,indStart+index+1,
412                        indexedCompare<size_t>(itsVariances.begin()));
413      }
414      // merge with the following block
415      if (index + 1 != nSamples) {
416          // merge indices on the basis of variances
417          inplace_merge(indStart,indStart+index+1,indStart+nSamples,
418                        indexedCompare<size_t>(itsVariances.begin()));
419      }
420  } else {
421      // itsSampleNumber is the index of the new element
422      AlwaysAssert(itsSampleNumber < itsSortedIndices.size(), AipsError);
423      itsSortedIndices[itsSampleNumber] = itsSampleNumber;
424      if (itsSampleNumber >= 1) {
425          // we have to place this new sample in
426          const vector<size_t>::iterator indStart = itsSortedIndices.begin();
427          // merge indices on the basis of variances
428          inplace_merge(indStart,indStart+itsSampleNumber,indStart+itsSampleNumber+1,
429                        indexedCompare<size_t>(itsVariances.begin()));
430      }
431  }
432}
433
434// build sorted cache from the scratch
435void LFNoiseEstimator::buildSortedCache() const
436{
437  // the number of samples accumulated so far may be less than the
438  // buffer size
439  const size_t nSamples = numberOfSamples();
[1643]440  AlwaysAssert(nSamples <= itsSortedIndices.size(), AipsError);
[1642]441  for (size_t i=0; i<nSamples; ++i) {
442       itsSortedIndices[i]=i;
443  }
444
445  // sort indices, but check the array of variances
446  const vector<size_t>::iterator indStart = itsSortedIndices.begin();
447  stable_sort(indStart,indStart+nSamples, indexedCompare<size_t>(itsVariances.begin()));
448}
449
450//
451///////////////////////////////////////////////////////////////////////////////
452
453///////////////////////////////////////////////////////////////////////////////
454//
[881]455// RunningBox -    a running box calculator. This class implements
[351]456//                 interations over the specified spectrum and calculates
457//                 running box filter statistics.
[331]458//
[297]459
[331]460// set up the object with the references to actual data
461// and the number of channels in the running box
[351]462RunningBox::RunningBox(const casa::Vector<casa::Float>  &in_spectrum,
463                       const casa::Vector<casa::Bool>   &in_mask,
[996]464                       const std::pair<int,int>         &in_edge,
465                       int in_max_box_nchan) throw(AipsError) :
[331]466        spectrum(in_spectrum), mask(in_mask), edge(in_edge),
[996]467        max_box_nchan(in_max_box_nchan)
[351]468{
469  rewind();
470}
[331]471
[351]472void RunningBox::rewind() throw(AipsError) {
473  // fill statistics for initial box
474  box_chan_cntr=0; // no channels are currently in the box
475  sumf=0.;           // initialize statistics
476  sumf2=0.;
477  sumch=0.;
478  sumch2=0.;
479  sumfch=0.;
480  int initial_box_ch=edge.first;
481  for (;initial_box_ch<edge.second && box_chan_cntr<max_box_nchan;
482        ++initial_box_ch)
483       advanceRunningBox(initial_box_ch);
[881]484
485  if (initial_box_ch==edge.second)
[351]486      throw AipsError("RunningBox::rewind - too much channels are masked");
487
488  cur_channel=edge.first;
[881]489  start_advance=initial_box_ch-max_box_nchan/2;
[351]490}
491
492// access to the statistics
493const casa::Float& RunningBox::getLinMean() const throw(AipsError)
[331]494{
[351]495  DebugAssert(cur_channel<edge.second, AipsError);
496  if (need2recalculate) updateDerivativeStatistics();
497  return linmean;
[297]498}
499
[351]500const casa::Float& RunningBox::getLinVariance() const throw(AipsError)
501{
502  DebugAssert(cur_channel<edge.second, AipsError);
503  if (need2recalculate) updateDerivativeStatistics();
504  return linvariance;
505}
[331]506
[2163]507casa::Float RunningBox::aboveMean() const throw(AipsError)
[351]508{
509  DebugAssert(cur_channel<edge.second, AipsError);
510  if (need2recalculate) updateDerivativeStatistics();
511  return spectrum[cur_channel]-linmean;
512}
513
514int RunningBox::getChannel() const throw()
515{
516  return cur_channel;
517}
518
519// actual number of channels in the box (max_box_nchan, if no channels
520// are masked)
521int RunningBox::getNumberOfBoxPoints() const throw()
522{
523  return box_chan_cntr;
524}
525
[1644]526// supplementary function to control running mean/median calculations.
527// It adds a specified channel to the running box and
[297]528// removes (ch-max_box_nchan+1)'th channel from there
529// Channels, for which the mask is false or index is beyond the
530// allowed range, are ignored
[351]531void RunningBox::advanceRunningBox(int ch) throw(AipsError)
[297]532{
533  if (ch>=edge.first && ch<edge.second)
534      if (mask[ch]) { // ch is a valid channel
535          ++box_chan_cntr;
[351]536          sumf+=spectrum[ch];
537          sumf2+=square(spectrum[ch]);
[996]538          sumch+=Float(ch);
539          sumch2+=square(Float(ch));
540          sumfch+=spectrum[ch]*Float(ch);
541          need2recalculate=True;
[297]542      }
543  int ch2remove=ch-max_box_nchan;
544  if (ch2remove>=edge.first && ch2remove<edge.second)
545      if (mask[ch2remove]) { // ch2remove is a valid channel
546          --box_chan_cntr;
[351]547          sumf-=spectrum[ch2remove];
[881]548          sumf2-=square(spectrum[ch2remove]);
[996]549          sumch-=Float(ch2remove);
550          sumch2-=square(Float(ch2remove));
551          sumfch-=spectrum[ch2remove]*Float(ch2remove);
552          need2recalculate=True;
[297]553      }
554}
555
[351]556// next channel
557void RunningBox::next() throw(AipsError)
[297]558{
[351]559   AlwaysAssert(cur_channel<edge.second,AipsError);
560   ++cur_channel;
561   if (cur_channel+max_box_nchan/2<edge.second && cur_channel>=start_advance)
562       advanceRunningBox(cur_channel+max_box_nchan/2); // update statistics
[297]563}
564
[351]565// checking whether there are still elements
566casa::Bool RunningBox::haveMore() const throw()
567{
568   return cur_channel<edge.second;
569}
570
571// calculate derivative statistics. This function is const, because
572// it updates the cache only
573void RunningBox::updateDerivativeStatistics() const throw(AipsError)
574{
575  AlwaysAssert(box_chan_cntr, AipsError);
[881]576
[351]577  Float mean=sumf/Float(box_chan_cntr);
578
579  // linear LSF formulae
580  Float meanch=sumch/Float(box_chan_cntr);
581  Float meanch2=sumch2/Float(box_chan_cntr);
582  if (meanch==meanch2 || box_chan_cntr<3) {
583      // vertical line in the spectrum, can't calculate linmean and linvariance
584      linmean=0.;
585      linvariance=0.;
586  } else {
587      Float coeff=(sumfch/Float(box_chan_cntr)-meanch*mean)/
588                (meanch2-square(meanch));
589      linmean=coeff*(Float(cur_channel)-meanch)+mean;
[1670]590      linvariance=sumf2/Float(box_chan_cntr)-square(mean)-
591                    square(coeff)*(meanch2-square(meanch));
592      if (linvariance<0.) {
593          // this shouldn't happen normally, but could be due to round-off error
594          linvariance = 0;
595      } else {
596          linvariance = sqrt(linvariance);
597      }
[351]598  }
599  need2recalculate=False;
600}
601
602
603//
604///////////////////////////////////////////////////////////////////////////////
605
606///////////////////////////////////////////////////////////////////////////////
607//
[1644]608// LFAboveThreshold - a running mean/median algorithm for line detection
[351]609//
610//
611
612
613// set up the detection criterion
614LFAboveThreshold::LFAboveThreshold(std::list<pair<int,int> > &in_lines,
615                                   int in_min_nchan,
[1644]616                                   casa::Float in_threshold,
617                                   bool use_median,
618                                   int noise_sample_size) throw() :
[351]619             min_nchan(in_min_nchan), threshold(in_threshold),
[1644]620             lines(in_lines), running_box(NULL), itsUseMedian(use_median),
621             itsNoiseSampleSize(noise_sample_size) {}
[351]622
623LFAboveThreshold::~LFAboveThreshold() throw()
624{
625  if (running_box!=NULL) delete running_box;
626}
627
628// replace the detection criterion
629void LFAboveThreshold::setCriterion(int in_min_nchan, casa::Float in_threshold)
630                                 throw()
631{
632  min_nchan=in_min_nchan;
633  threshold=in_threshold;
634}
635
[907]636// get the sign of runningBox->aboveMean(). The RunningBox pointer
637// should be defined
638casa::Int LFAboveThreshold::getAboveMeanSign() const throw()
639{
640  const Float buf=running_box->aboveMean();
641  if (buf>0) return 1;
642  if (buf<0) return -1;
643  return 0;
644}
[351]645
[907]646
[297]647// process a channel: update cur_line and is_detected before and
648// add a new line to the list, if necessary
[351]649void LFAboveThreshold::processChannel(Bool detect,
650                 const casa::Vector<casa::Bool> &mask) throw(casa::AipsError)
[297]651{
652  try {
[907]653       if (is_detected_before) {
654          // we have to check that the current detection has the
655          // same sign of running_box->aboveMean
656          // otherwise it could be a spurious detection
657          if (last_sign && last_sign!=getAboveMeanSign())
658              detect=False;
[1315]659       }
660       if (detect) {
661           last_sign=getAboveMeanSign();
662           if (is_detected_before)
663               cur_line.second=running_box->getChannel()+1;
664           else {
665               is_detected_before=True;
666               cur_line.first=running_box->getChannel();
667               cur_line.second=running_box->getChannel()+1;
668           }
669       } else processCurLine(mask);
[297]670  }
671  catch (const AipsError &ae) {
672      throw;
[881]673  }
[297]674  catch (const exception &ex) {
[351]675      throw AipsError(String("LFAboveThreshold::processChannel - STL error: ")+ex.what());
[297]676  }
677}
678
679// process the interval of channels stored in cur_line
680// if it satisfies the criterion, add this interval as a new line
[351]681void LFAboveThreshold::processCurLine(const casa::Vector<casa::Bool> &mask)
[331]682                                   throw(casa::AipsError)
[297]683{
684  try {
[881]685       if (is_detected_before) {
[1315]686           if (cur_line.second-cur_line.first>=min_nchan) {
[996]687               // it was a detection. We need to change the list
688               Bool add_new_line=False;
689               if (lines.size()) {
690                   for (int i=lines.back().second;i<cur_line.first;++i)
691                        if (mask[i]) { // one valid channel in between
692                                //  means that we deal with a separate line
693                            add_new_line=True;
694                            break;
695                        }
696               } else add_new_line=True;
697               if (add_new_line)
698                   lines.push_back(cur_line);
[881]699               else lines.back().second=cur_line.second;
[996]700           }
701           is_detected_before=False;
[881]702       }
[297]703  }
704  catch (const AipsError &ae) {
705      throw;
[881]706  }
[297]707  catch (const exception &ex) {
[351]708      throw AipsError(String("LFAboveThreshold::processCurLine - STL error: ")+ex.what());
[297]709  }
710}
711
[551]712// return the array with signs of the value-current mean
713// An element is +1 if value>mean, -1 if less, 0 if equal.
714// This array is updated each time the findLines method is called and
715// is used to search the line wings
716const casa::Vector<Int>& LFAboveThreshold::getSigns() const throw()
717{
718  return signs;
719}
720
[331]721// find spectral lines and add them into list
[351]722void LFAboveThreshold::findLines(const casa::Vector<casa::Float> &spectrum,
[996]723                              const casa::Vector<casa::Bool> &mask,
724                              const std::pair<int,int> &edge,
725                              int max_box_nchan)
[331]726                        throw(casa::AipsError)
727{
728  const int minboxnchan=4;
[351]729  try {
[331]730
[351]731      if (running_box!=NULL) delete running_box;
732      running_box=new RunningBox(spectrum,mask,edge,max_box_nchan);
[368]733
734      // determine the off-line variance first
735      // an assumption made: lines occupy a small part of the spectrum
[881]736
[1644]737      const size_t noiseSampleSize = itsNoiseSampleSize<0 ? size_t(edge.second-edge.first) :
738                      std::min(size_t(itsNoiseSampleSize), size_t(edge.second-edge.first));
739      DebugAssert(noiseSampleSize,AipsError);
740      const bool globalNoise = (size_t(edge.second - edge.first) == noiseSampleSize);
741      LFNoiseEstimator ne(noiseSampleSize);
[881]742
[1643]743      for (;running_box->haveMore();running_box->next()) {
[1644]744           ne.add(running_box->getLinVariance());
745           if (ne.filledToCapacity()) {
746               break;
747           }
[1643]748      }
[881]749
[1644]750      Float offline_variance = -1; // just a flag that it is unset
751           
752      if (globalNoise) {
753          offline_variance = itsUseMedian ? ne.median() : ne.meanLowest80Percent();
754      }
[881]755
[351]756      // actual search algorithm
757      is_detected_before=False;
[368]758
[551]759      // initiate the signs array
760      signs.resize(spectrum.nelements());
761      signs=Vector<Int>(spectrum.nelements(),0);
762
[369]763      //ofstream os("dbg.dat");
[368]764      for (running_box->rewind();running_box->haveMore();
765                                 running_box->next()) {
[351]766           const int ch=running_box->getChannel();
[1644]767           if (!globalNoise) {
768               // add a next point for a local noise estimate
769               ne.add(running_box->getLinVariance());
770           }   
771           if (running_box->getNumberOfBoxPoints()>=minboxnchan) {
772               if (!globalNoise) {
773                   offline_variance = itsUseMedian ? ne.median() : ne.meanLowest80Percent();
774               }
775               AlwaysAssert(offline_variance>0.,AipsError);
[996]776               processChannel(mask[ch] && (fabs(running_box->aboveMean()) >=
777                  threshold*offline_variance), mask);
[1644]778           } else processCurLine(mask); // just finish what was accumulated before
[907]779
[996]780           signs[ch]=getAboveMeanSign();
[1641]781            //os<<ch<<" "<<spectrum[ch]<<" "<<fabs(running_box->aboveMean())<<" "<<
782            //threshold*offline_variance<<endl;
[351]783      }
[352]784      if (lines.size())
785          searchForWings(lines,signs,mask,edge);
[344]786  }
[351]787  catch (const AipsError &ae) {
788      throw;
[881]789  }
[351]790  catch (const exception &ex) {
791      throw AipsError(String("LFAboveThreshold::findLines - STL error: ")+ex.what());
792  }
[331]793}
794
795//
796///////////////////////////////////////////////////////////////////////////////
797
[343]798///////////////////////////////////////////////////////////////////////////////
799//
[352]800// LFLineListOperations::IntersectsWith  -  An auxiliary object function
801//                to test whether two lines have a non-void intersection
[343]802//
[331]803
[343]804
805// line1 - range of the first line: start channel and stop+1
[352]806LFLineListOperations::IntersectsWith::IntersectsWith(const std::pair<int,int> &in_line1) :
[343]807                          line1(in_line1) {}
808
809
810// return true if line2 intersects with line1 with at least one
811// common channel, and false otherwise
812// line2 - range of the second line: start channel and stop+1
[352]813bool LFLineListOperations::IntersectsWith::operator()(const std::pair<int,int> &line2)
[343]814                          const throw()
815{
816  if (line2.second<line1.first) return false; // line2 is at lower channels
817  if (line2.first>line1.second) return false; // line2 is at upper channels
818  return true; // line2 has an intersection or is adjacent to line1
819}
820
821//
822///////////////////////////////////////////////////////////////////////////////
823
824///////////////////////////////////////////////////////////////////////////////
825//
[352]826// LFLineListOperations::BuildUnion - An auxiliary object function to build a union
[343]827// of several lines to account for a possibility of merging the nearby lines
828//
829
830// set an initial line (can be a first line in the sequence)
[352]831LFLineListOperations::BuildUnion::BuildUnion(const std::pair<int,int> &line1) :
[343]832                             temp_line(line1) {}
833
834// update temp_line with a union of temp_line and new_line
835// provided there is no gap between the lines
[352]836void LFLineListOperations::BuildUnion::operator()(const std::pair<int,int> &new_line)
[343]837                                   throw()
838{
839  if (new_line.first<temp_line.first) temp_line.first=new_line.first;
840  if (new_line.second>temp_line.second) temp_line.second=new_line.second;
841}
842
843// return the result (temp_line)
[352]844const std::pair<int,int>& LFLineListOperations::BuildUnion::result() const throw()
[343]845{
846  return temp_line;
847}
848
849//
850///////////////////////////////////////////////////////////////////////////////
851
852///////////////////////////////////////////////////////////////////////////////
853//
[352]854// LFLineListOperations::LaterThan - An auxiliary object function to test whether a
[343]855// specified line is at lower spectral channels (to preserve the order in
856// the line list)
857//
858
859// setup the line to compare with
[352]860LFLineListOperations::LaterThan::LaterThan(const std::pair<int,int> &in_line1) :
[343]861                         line1(in_line1) {}
862
863// return true if line2 should be placed later than line1
864// in the ordered list (so, it is at greater channel numbers)
[352]865bool LFLineListOperations::LaterThan::operator()(const std::pair<int,int> &line2)
[343]866                          const throw()
867{
868  if (line2.second<line1.first) return false; // line2 is at lower channels
869  if (line2.first>line1.second) return true; // line2 is at upper channels
[881]870
[343]871  // line2 intersects with line1. We should have no such situation in
872  // practice
873  return line2.first>line1.first;
874}
875
876//
877///////////////////////////////////////////////////////////////////////////////
878
879
880///////////////////////////////////////////////////////////////////////////////
881//
[881]882// STLineFinder  -  a class for automated spectral line search
[343]883//
884//
[331]885
[881]886STLineFinder::STLineFinder() throw() : edge(0,0)
[331]887{
[369]888  setOptions();
[331]889}
890
[369]891// set the parameters controlling algorithm
892// in_threshold a single channel threshold default is sqrt(3), which
893//              means together with 3 minimum channels at least 3 sigma
894//              detection criterion
895//              For bad baseline shape, in_threshold may need to be
896//              increased
897// in_min_nchan minimum number of channels above the threshold to report
898//              a detection, default is 3
899// in_avg_limit perform the averaging of no more than in_avg_limit
900//              adjacent channels to search for broad lines
[881]901//              Default is 8, but for a bad baseline shape this
[369]902//              parameter should be decreased (may be even down to a
903//              minimum of 1 to disable this option) to avoid
904//              confusing of baseline undulations with a real line.
[881]905//              Setting a very large value doesn't usually provide
906//              valid detections.
[1644]907// in_box_size  the box size for running mean/median calculation. Default is
[369]908//              1./5. of the whole spectrum size
[1644]909// in_noise_box the box size for off-line noise estimation (if working with
910//              local noise. Negative value means use global noise estimate
911//              Default is -1 (i.e. estimate using the whole spectrum)
912// in_median    true if median statistics is used as opposed to average of
913//              the lowest 80% of deviations (default)
[881]914void STLineFinder::setOptions(const casa::Float &in_threshold,
[369]915                              const casa::Int &in_min_nchan,
[996]916                              const casa::Int &in_avg_limit,
[1644]917                              const casa::Float &in_box_size,
918                              const casa::Float &in_noise_box,
919                              const casa::Bool &in_median) throw()
[369]920{
921  threshold=in_threshold;
922  min_nchan=in_min_nchan;
923  avg_limit=in_avg_limit;
924  box_size=in_box_size;
[1644]925  itsNoiseBox = in_noise_box;
926  itsUseMedian = in_median;
[2012]927
928  useScantable = true;
[369]929}
930
[881]931STLineFinder::~STLineFinder() throw(AipsError) {}
[331]932
[907]933// set scan to work with (in_scan parameter)
934void STLineFinder::setScan(const ScantableWrapper &in_scan) throw(AipsError)
935{
936  scan=in_scan.getCP();
937  AlwaysAssert(!scan.null(),AipsError);
[2012]938}
[924]939
[2012]940// set spectrum data to work with. this is a method to allow linefinder work
941// without setting scantable for the purpose of using linefinder inside some
942// method in scantable class. (Dec 22, 2010 by W.Kawasaki)
943void STLineFinder::setData(const std::vector<float> &in_spectrum)
944{
945  spectrum = Vector<Float>(in_spectrum);
946  useScantable = false;
[907]947}
948
949// search for spectral lines. Number of lines found is returned
950// in_edge and in_mask control channel rejection for a given row
[331]951//   if in_edge has zero length, all channels chosen by mask will be used
952//   if in_edge has one element only, it represents the number of
953//      channels to drop from both sides of the spectrum
954//   in_edge is introduced for convinience, although all functionality
[881]955//   can be achieved using a spectrum mask only
[907]956int STLineFinder::findLines(const std::vector<bool> &in_mask,
[2345]957                            const std::vector<int> &in_edge,
958                            const casa::uInt &whichRow) throw(casa::AipsError)
[331]959{
[2012]960  if (useScantable && scan.null())
[907]961      throw AipsError("STLineFinder::findLines - a scan should be set first,"
962                      " use set_scan");
[924]963
[2012]964  uInt nchan = useScantable ? scan->nchan(scan->getIF(whichRow)) : spectrum.nelements();
[907]965  // set up mask and edge rejection
[924]966  // no mask given...
967  if (in_mask.size() == 0) {
968    mask = Vector<Bool>(nchan,True);
969  } else {
970    // use provided mask
971    mask=Vector<Bool>(in_mask);
972  }
973  if (mask.nelements()!=nchan)
[2012]974      throw AipsError("STLineFinder::findLines - in_scan and in_mask, or in_spectrum "
975                      "and in_mask have different number of spectral channels.");
[1641]976
977  // taking flagged channels into account
[2012]978  if (useScantable) {
979    vector<bool> flaggedChannels = scan->getMask(whichRow);
980    if (flaggedChannels.size()) {
[1641]981      // there is a mask set for this row
982      if (flaggedChannels.size() != mask.nelements()) {
[2012]983          throw AipsError("STLineFinder::findLines - internal inconsistency: number of "
984                          "mask elements do not match the number of channels");
[1641]985      }
986      for (size_t ch = 0; ch<mask.nelements(); ++ch) {
987           mask[ch] &= flaggedChannels[ch];
988      }
[2012]989    }
[1641]990  }
991
[907]992  // number of elements in in_edge
993  if (in_edge.size()>2)
994      throw AipsError("STLineFinder::findLines - the length of the in_edge parameter"
[996]995                      "should not exceed 2");
[907]996      if (!in_edge.size()) {
[881]997           // all spectra, no rejection
[331]998           edge.first=0;
[996]999           edge.second=nchan;
[907]1000      } else {
1001           edge.first=in_edge[0];
[996]1002           if (edge.first<0)
1003               throw AipsError("STLineFinder::findLines - the in_edge parameter has a negative"
1004                                "number of channels to drop");
1005           if (edge.first>=int(nchan))
1006               throw AipsError("STLineFinder::findLines - all channels are rejected by the in_edge parameter");
[907]1007           if (in_edge.size()==2) {
[996]1008               edge.second=in_edge[1];
1009               if (edge.second<0)
1010                   throw AipsError("STLineFinder::findLines - the in_edge parameter has a negative"
1011                                   "number of channels to drop");
[924]1012               edge.second=nchan-edge.second;
[996]1013           } else edge.second=nchan-edge.first;
[369]1014           if (edge.second<0 || (edge.first>=edge.second))
[996]1015               throw AipsError("STLineFinder::findLines - all channels are rejected by the in_edge parameter");
[881]1016       }
[924]1017
[907]1018  //
[924]1019  int max_box_nchan=int(nchan*box_size); // number of channels in running
[331]1020                                                 // box
1021  if (max_box_nchan<2)
[881]1022      throw AipsError("STLineFinder::findLines - box_size is too small");
[331]1023
[1644]1024  // number of elements in the sample for noise estimate
1025  const int noise_box = itsNoiseBox<0 ? -1 : int(nchan * itsNoiseBox);
1026
1027  if ((noise_box!= -1) and (noise_box<2))
1028      throw AipsError("STLineFinder::findLines - noise_box is supposed to be at least 2 elements");
1029
[2012]1030  if (useScantable) {
1031    spectrum.resize();
1032    spectrum = Vector<Float>(scan->getSpectrum(whichRow));
1033  }
[331]1034
1035  lines.resize(0); // search from the scratch
[370]1036  last_row_used=whichRow;
[331]1037  Vector<Bool> temp_mask(mask);
[351]1038
1039  Bool first_pass=True;
[368]1040  Int avg_factor=1; // this number of adjacent channels is averaged together
1041                    // the total number of the channels is not altered
[996]1042                    // instead, min_nchan is also scaled
1043                    // it helps to search for broad lines
[551]1044  Vector<Int> signs; // a buffer for signs of the value - mean quantity
1045                     // see LFAboveThreshold for details
[996]1046                     // We need only signs resulted from last iteration
1047                     // because all previous values may be corrupted by the
1048                     // presence of spectral lines
[344]1049  while (true) {
[351]1050     // a buffer for new lines found at this iteration
[881]1051     std::list<pair<int,int> > new_lines;
[351]1052
1053     try {
[369]1054         // line find algorithm
[1644]1055         LFAboveThreshold lfalg(new_lines,avg_factor*min_nchan, threshold, itsUseMedian,noise_box);
[352]1056         lfalg.findLines(spectrum,temp_mask,edge,max_box_nchan);
[996]1057         signs.resize(lfalg.getSigns().nelements());
1058         signs=lfalg.getSigns();
[368]1059         first_pass=False;
1060         if (!new_lines.size())
[996]1061              throw AipsError("spurious"); // nothing new - use the same
1062                                           // code as for a real exception
[351]1063     }
1064     catch(const AipsError &ae) {
1065         if (first_pass) throw;
[368]1066         // nothing new - proceed to the next step of averaging, if any
[996]1067         // (to search for broad lines)
[1315]1068         if (avg_factor>=avg_limit) break; // averaging up to avg_limit
[996]1069                                          // adjacent channels,
1070                                          // stop after that
1071         avg_factor*=2; // twice as more averaging
1072         subtractBaseline(temp_mask,9);
1073         averageAdjacentChannels(temp_mask,avg_factor);
1074         continue;
[1315]1075     }
[368]1076     keepStrongestOnly(temp_mask,new_lines,max_box_nchan);
[343]1077     // update the list (lines) merging intervals, if necessary
[344]1078     addNewSearchResult(new_lines,lines);
1079     // get a new mask
[881]1080     temp_mask=getMask();
[343]1081  }
[881]1082
[551]1083  // an additional search for wings because in the presence of very strong
1084  // lines temporary mean used at each iteration will be higher than
1085  // the true mean
[881]1086
[551]1087  if (lines.size())
1088      LFLineListOperations::searchForWings(lines,signs,mask,edge);
[881]1089
[331]1090  return int(lines.size());
1091}
1092
[369]1093// auxiliary function to fit and subtract a polynomial from the current
[890]1094// spectrum. It uses the Fitter class. This action is required before
[369]1095// reducing the spectral resolution if the baseline shape is bad
[881]1096void STLineFinder::subtractBaseline(const casa::Vector<casa::Bool> &temp_mask,
[369]1097                      const casa::Int &order) throw(casa::AipsError)
1098{
1099  AlwaysAssert(spectrum.nelements(),AipsError);
1100  // use the fact that temp_mask excludes channels rejected at the edge
[890]1101  Fitter sdf;
[369]1102  std::vector<float> absc(spectrum.nelements());
[996]1103  for (unsigned int i=0;i<absc.size();++i)
[369]1104       absc[i]=float(i)/float(spectrum.nelements());
1105  std::vector<float> spec;
1106  spectrum.tovector(spec);
1107  std::vector<bool> std_mask;
1108  temp_mask.tovector(std_mask);
1109  sdf.setData(absc,spec,std_mask);
1110  sdf.setExpression("poly",order);
[2196]1111  if (!sdf.lfit()) return; // fit failed, use old spectrum
[881]1112  spectrum=casa::Vector<casa::Float>(sdf.getResidual());
[369]1113}
1114
[368]1115// auxiliary function to average adjacent channels and update the mask
1116// if at least one channel involved in summation is masked, all
1117// output channels will be masked. This function works with the
1118// spectrum and edge fields of this class, but updates the mask
1119// array specified, rather than the field of this class
1120// boxsize - a number of adjacent channels to average
[881]1121void STLineFinder::averageAdjacentChannels(casa::Vector<casa::Bool> &mask2update,
[368]1122                                   const casa::Int &boxsize)
1123                            throw(casa::AipsError)
1124{
1125  DebugAssert(mask2update.nelements()==spectrum.nelements(), AipsError);
1126  DebugAssert(boxsize!=0,AipsError);
[881]1127
[368]1128  for (int n=edge.first;n<edge.second;n+=boxsize) {
1129       DebugAssert(n<spectrum.nelements(),AipsError);
1130       int nboxch=0; // number of channels currently in the box
1131       Float mean=0; // buffer for mean calculations
1132       for (int k=n;k<n+boxsize && k<edge.second;++k)
1133            if (mask2update[k]) {  // k is a valid channel
[996]1134                mean+=spectrum[k];
1135                ++nboxch;
[881]1136            }
[368]1137       if (nboxch<boxsize) // mask these channels
1138           for (int k=n;k<n+boxsize && k<edge.second;++k)
[996]1139                mask2update[k]=False;
[368]1140       else {
1141          mean/=Float(boxsize);
[996]1142           for (int k=n;k<n+boxsize && k<edge.second;++k)
1143                spectrum[k]=mean;
[368]1144       }
1145  }
1146}
[331]1147
[368]1148
[297]1149// get the mask to mask out all lines that have been found (default)
1150// if invert=true, only channels belong to lines will be unmasked
1151// Note: all channels originally masked by the input mask (in_mask
1152//       in setScan) or dropped out by the edge parameter (in_edge
1153//       in setScan) are still excluded regardless on the invert option
[881]1154std::vector<bool> STLineFinder::getMask(bool invert)
[297]1155                                        const throw(casa::AipsError)
1156{
1157  try {
[2012]1158    if (useScantable) {
1159      if (scan.null())
1160        throw AipsError("STLineFinder::getMask - a scan should be set first,"
1161                        " use set_scan followed by find_lines");
1162      DebugAssert(mask.nelements()==scan->getChannels(last_row_used), AipsError);
1163    }
1164    /*
1165    if (!lines.size())
1166       throw AipsError("STLineFinder::getMask - one have to search for "
[996]1167                           "lines first, use find_lines");
[2012]1168    */
1169    std::vector<bool> res_mask(mask.nelements());
1170    // iterator through lines
1171    std::list<std::pair<int,int> >::const_iterator cli=lines.begin();
1172    for (int ch=0;ch<int(res_mask.size());++ch) {
1173      if (ch<edge.first || ch>=edge.second) res_mask[ch]=false;
1174      else if (!mask[ch]) res_mask[ch]=false;
1175      else {
1176        res_mask[ch]=!invert; // no line by default
1177        if (cli!=lines.end())
1178          if (ch>=cli->first && ch<cli->second)
1179            res_mask[ch]=invert; // this is a line
1180      }
1181      if (cli!=lines.end())
1182        if (ch>=cli->second)
1183          ++cli; // next line in the list
1184    }
1185    return res_mask;
[297]1186  }
1187  catch (const AipsError &ae) {
[2012]1188    throw;
[881]1189  }
[297]1190  catch (const exception &ex) {
[2012]1191    throw AipsError(String("STLineFinder::getMask - STL error: ")+ex.what());
[297]1192  }
1193}
1194
[370]1195// get range for all lines found. The same units as used in the scan
1196// will be returned (e.g. velocity instead of channels).
[881]1197std::vector<double> STLineFinder::getLineRanges()
[297]1198                             const throw(casa::AipsError)
1199{
[2012]1200  std::vector<double> vel;
1201  if (useScantable) {
1202    // convert to required abscissa units
1203    vel = scan->getAbcissa(last_row_used);
1204  } else {
[2081]1205    for (uInt i = 0; i < spectrum.nelements(); ++i)
[2012]1206      vel.push_back((double)i);
1207  }
[370]1208  std::vector<int> ranges=getLineRangesInChannels();
1209  std::vector<double> res(ranges.size());
1210
1211  std::vector<int>::const_iterator cri=ranges.begin();
1212  std::vector<double>::iterator outi=res.begin();
1213  for (;cri!=ranges.end() && outi!=res.end();++cri,++outi)
1214       if (uInt(*cri)>=vel.size())
[881]1215          throw AipsError("STLineFinder::getLineRanges - getAbcissa provided less channels than reqired");
[370]1216       else *outi=vel[*cri];
1217  return res;
1218}
1219
1220// The same as getLineRanges, but channels are always used to specify
1221// the range
[881]1222std::vector<int> STLineFinder::getLineRangesInChannels()
[370]1223                                   const throw(casa::AipsError)
1224{
[297]1225  try {
[2012]1226    if (useScantable) {
1227      if (scan.null())
1228        throw AipsError("STLineFinder::getLineRangesInChannels - a scan should be set first,"
1229                        " use set_scan followed by find_lines");
1230      DebugAssert(mask.nelements()==scan->getChannels(last_row_used), AipsError);
1231    }
[881]1232
[2012]1233    if (!lines.size())
1234      throw AipsError("STLineFinder::getLineRangesInChannels - one have to search for "
1235                      "lines first, use find_lines");
[881]1236
[2012]1237    std::vector<int> res(2*lines.size());
1238    // iterator through lines & result
1239    std::list<std::pair<int,int> >::const_iterator cli = lines.begin();
1240    std::vector<int>::iterator ri = res.begin();
1241    for (; cli != lines.end() && ri != res.end(); ++cli,++ri) {
1242      *ri = cli->first;
1243      if (++ri != res.end())
1244        *ri = cli->second - 1;
1245    }
1246    return res;
1247  } catch (const AipsError &ae) {
1248    throw;
1249  } catch (const exception &ex) {
1250    throw AipsError(String("STLineFinder::getLineRanges - STL error: ") + ex.what());
[297]1251  }
1252}
[331]1253
[370]1254
1255
[368]1256// an auxiliary function to remove all lines from the list, except the
1257// strongest one (by absolute value). If the lines removed are real,
[881]1258// they will be find again at the next iteration. This approach
1259// increases the number of iterations required, but is able to remove
[1315]1260// spurious detections likely to occur near strong lines.
[368]1261// Later a better criterion may be implemented, e.g.
1262// taking into consideration the brightness of different lines. Now
[881]1263// use the simplest solution
[368]1264// temp_mask - mask to work with (may be different from original mask as
1265// the lines previously found may be masked)
1266// lines2update - a list of lines to work with
1267//                 nothing will be done if it is empty
1268// max_box_nchan - channels in the running box for baseline filtering
[881]1269void STLineFinder::keepStrongestOnly(const casa::Vector<casa::Bool> &temp_mask,
[996]1270                  std::list<std::pair<int, int> > &lines2update,
1271                  int max_box_nchan)
[368]1272                                   throw (casa::AipsError)
1273{
1274  try {
1275      if (!lines2update.size()) return; // ignore an empty list
1276
1277      // current line
1278      std::list<std::pair<int,int> >::iterator li=lines2update.begin();
1279      // strongest line
1280      std::list<std::pair<int,int> >::iterator strongli=lines2update.begin();
1281      // the flux (absolute value) of the strongest line
1282      Float peak_flux=-1; // negative value - a flag showing uninitialized
1283                          // value
1284      // the algorithm below relies on the list being ordered
1285      Float tmp_flux=-1; // a temporary peak
1286      for (RunningBox running_box(spectrum,temp_mask,edge,max_box_nchan);
1287           running_box.haveMore(); running_box.next()) {
1288
1289           if (li==lines2update.end()) break; // no more lines
[996]1290           const int ch=running_box.getChannel();
1291           if (ch>=li->first && ch<li->second)
1292               if (temp_mask[ch] && tmp_flux<fabs(running_box.aboveMean()))
1293                   tmp_flux=fabs(running_box.aboveMean());
1294           if (ch==li->second-1) {
1295               if (peak_flux<tmp_flux) { // if peak_flux=-1, this condition
1296                   peak_flux=tmp_flux;   // will be satisfied
1297                   strongli=li;
1298               }
1299               ++li;
1300               tmp_flux=-1;
1301           }
[881]1302      }
[368]1303      std::list<std::pair<int,int> > res;
1304      res.splice(res.end(),lines2update,strongli);
1305      lines2update.clear();
1306      lines2update.splice(lines2update.end(),res);
1307  }
1308  catch (const AipsError &ae) {
1309      throw;
[881]1310  }
[368]1311  catch (const exception &ex) {
[881]1312      throw AipsError(String("STLineFinder::keepStrongestOnly - STL error: ")+ex.what());
[368]1313  }
1314
1315}
1316
[352]1317//
1318///////////////////////////////////////////////////////////////////////////////
1319
1320
1321///////////////////////////////////////////////////////////////////////////////
1322//
1323// LFLineListOperations - a class incapsulating  operations with line lists
1324//                        The LF prefix stands for Line Finder
1325//
1326
[331]1327// concatenate two lists preserving the order. If two lines appear to
1328// be adjacent, they are joined into the new one
[352]1329void LFLineListOperations::addNewSearchResult(const std::list<pair<int, int> > &newlines,
[881]1330                         std::list<std::pair<int, int> > &lines_list)
[331]1331                        throw(AipsError)
1332{
1333  try {
1334      for (std::list<pair<int,int> >::const_iterator cli=newlines.begin();
1335           cli!=newlines.end();++cli) {
[881]1336
[996]1337           // the first item, which has a non-void intersection or touches
1338           // the new line
1339           std::list<pair<int,int> >::iterator pos_beg=find_if(lines_list.begin(),
1340                          lines_list.end(), IntersectsWith(*cli));
1341           // the last such item
1342           std::list<pair<int,int> >::iterator pos_end=find_if(pos_beg,
1343                          lines_list.end(), not1(IntersectsWith(*cli)));
[343]1344
1345           // extract all lines which intersect or touch a new one into
[996]1346           // a temporary buffer. This may invalidate the iterators
1347           // line_buffer may be empty, if no lines intersects with a new
1348           // one.
1349           std::list<pair<int,int> > lines_buffer;
1350           lines_buffer.splice(lines_buffer.end(),lines_list, pos_beg, pos_end);
[343]1351
[996]1352           // build a union of all intersecting lines
1353           pair<int,int> union_line=for_each(lines_buffer.begin(),
1354                   lines_buffer.end(),BuildUnion(*cli)).result();
[881]1355
[996]1356           // search for a right place for the new line (union_line) and add
1357           std::list<pair<int,int> >::iterator pos2insert=find_if(lines_list.begin(),
1358                          lines_list.end(), LaterThan(union_line));
1359           lines_list.insert(pos2insert,union_line);
[331]1360      }
1361  }
1362  catch (const AipsError &ae) {
1363      throw;
[881]1364  }
[331]1365  catch (const exception &ex) {
[352]1366      throw AipsError(String("LFLineListOperations::addNewSearchResult - STL error: ")+ex.what());
[331]1367  }
1368}
[344]1369
1370// extend all line ranges to the point where a value stored in the
1371// specified vector changes (e.g. value-mean change its sign)
1372// This operation is necessary to include line wings, which are below
1373// the detection threshold. If lines becomes adjacent, they are
1374// merged together. Any masked channel stops the extension
[352]1375void LFLineListOperations::searchForWings(std::list<std::pair<int, int> > &newlines,
1376           const casa::Vector<casa::Int> &signs,
[996]1377           const casa::Vector<casa::Bool> &mask,
1378           const std::pair<int,int> &edge) throw(casa::AipsError)
[344]1379{
1380  try {
1381      for (std::list<pair<int,int> >::iterator li=newlines.begin();
1382           li!=newlines.end();++li) {
[996]1383           // update the left hand side
1384           for (int n=li->first-1;n>=edge.first;--n) {
1385                if (!mask[n]) break;
1386                if (signs[n]==signs[li->first] && signs[li->first])
1387                    li->first=n;
1388                else break;
1389           }
1390           // update the right hand side
1391           for (int n=li->second;n<edge.second;++n) {
1392                if (!mask[n]) break;
1393                if (signs[n]==signs[li->second-1] && signs[li->second-1])
1394                    li->second=n;
1395                else break;
1396           }
[344]1397      }
1398      // need to search for possible mergers.
1399      std::list<std::pair<int, int> >  result_buffer;
1400      addNewSearchResult(newlines,result_buffer);
1401      newlines.clear();
1402      newlines.splice(newlines.end(),result_buffer);
1403  }
1404  catch (const AipsError &ae) {
1405      throw;
[881]1406  }
[344]1407  catch (const exception &ex) {
[352]1408      throw AipsError(String("LFLineListOperations::extendLines - STL error: ")+ex.what());
[344]1409  }
1410}
[352]1411
1412//
1413///////////////////////////////////////////////////////////////////////////////
Note: See TracBrowser for help on using the repository browser.