1 | //#--------------------------------------------------------------------------- |
---|
2 | //# STLineFinder.cc: A class for automated spectral line search |
---|
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 | //# |
---|
29 | //# $Id: STLineFinder.cpp 2205 2011-07-01 02:27:03Z WataruKawasaki $ |
---|
30 | //#--------------------------------------------------------------------------- |
---|
31 | |
---|
32 | |
---|
33 | // ASAP |
---|
34 | #include "STLineFinder.h" |
---|
35 | #include "STFitter.h" |
---|
36 | #include "IndexedCompare.h" |
---|
37 | |
---|
38 | // STL |
---|
39 | #include <functional> |
---|
40 | #include <algorithm> |
---|
41 | #include <iostream> |
---|
42 | #include <fstream> |
---|
43 | |
---|
44 | using namespace asap; |
---|
45 | using namespace casa; |
---|
46 | using namespace std; |
---|
47 | |
---|
48 | namespace asap { |
---|
49 | |
---|
50 | /////////////////////////////////////////////////////////////////////////////// |
---|
51 | // |
---|
52 | // RunningBox - a running box calculator. This class implements |
---|
53 | // iterations over the specified spectrum and calculates |
---|
54 | // running box filter statistics. |
---|
55 | // |
---|
56 | |
---|
57 | class RunningBox { |
---|
58 | // The input data to work with. Use reference symantics to avoid |
---|
59 | // an unnecessary copying |
---|
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 |
---|
64 | |
---|
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) |
---|
71 | |
---|
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) |
---|
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 |
---|
84 | // masking) |
---|
85 | public: |
---|
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, |
---|
90 | const std::pair<int,int> &in_edge, |
---|
91 | int in_max_box_nchan) throw(AipsError); |
---|
92 | |
---|
93 | // access to the statistics |
---|
94 | const casa::Float& getLinMean() const throw(AipsError); |
---|
95 | const casa::Float& getLinVariance() const throw(AipsError); |
---|
96 | const casa::Float aboveMean() const throw(AipsError); |
---|
97 | int getChannel() const throw(); |
---|
98 | |
---|
99 | // actual number of channels in the box (max_box_nchan, if no channels |
---|
100 | // are masked) |
---|
101 | int getNumberOfBoxPoints() const throw(); |
---|
102 | |
---|
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); |
---|
111 | |
---|
112 | protected: |
---|
113 | // supplementary function to control running mean/median calculations. |
---|
114 | // It adds a specified channel to the running box and |
---|
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 | // |
---|
135 | class LFAboveThreshold : protected LFLineListOperations { |
---|
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 |
---|
142 | // the detection criterion, to be |
---|
143 | // a detection |
---|
144 | casa::Float threshold; // detection threshold - the |
---|
145 | // minimal signal to noise ratio |
---|
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 |
---|
149 | casa::Vector<Int> signs; // An array to store the signs of |
---|
150 | // the value - current mean |
---|
151 | // (used to search wings) |
---|
152 | casa::Int last_sign; // a sign (+1, -1 or 0) of the |
---|
153 | // last point of the detected line |
---|
154 | // |
---|
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) |
---|
161 | public: |
---|
162 | |
---|
163 | // set up the detection criterion |
---|
164 | LFAboveThreshold(std::list<pair<int,int> > &in_lines, |
---|
165 | int in_min_nchan = 3, |
---|
166 | casa::Float in_threshold = 5, |
---|
167 | bool use_median = false, |
---|
168 | int noise_sample_size = -1) throw(); |
---|
169 | virtual ~LFAboveThreshold() throw(); |
---|
170 | |
---|
171 | // replace the detection criterion |
---|
172 | void setCriterion(int in_min_nchan, casa::Float in_threshold) throw(); |
---|
173 | |
---|
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 | |
---|
180 | // find spectral lines and add them into list |
---|
181 | // if statholder is not NULL, the accumulate function of it will be |
---|
182 | // called for each channel to save statistics |
---|
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, |
---|
186 | const casa::Vector<casa::Bool> &mask, |
---|
187 | const std::pair<int,int> &edge, |
---|
188 | int max_box_nchan) throw(casa::AipsError); |
---|
189 | |
---|
190 | protected: |
---|
191 | |
---|
192 | // process a channel: update curline and is_detected before and |
---|
193 | // add a new line to the list, if necessary using processCurLine() |
---|
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); |
---|
197 | |
---|
198 | // process the interval of channels stored in curline |
---|
199 | // if it satisfies the criterion, add this interval as a new line |
---|
200 | void processCurLine(const casa::Vector<casa::Bool> &mask) |
---|
201 | throw(casa::AipsError); |
---|
202 | |
---|
203 | // get the sign of runningBox->aboveMean(). The RunningBox pointer |
---|
204 | // should be defined |
---|
205 | casa::Int getAboveMeanSign() const throw(); |
---|
206 | }; |
---|
207 | |
---|
208 | // |
---|
209 | /////////////////////////////////////////////////////////////////////////////// |
---|
210 | |
---|
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 | |
---|
221 | struct 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 | |
---|
238 | // return true if the buffer is full (i.e. statistics are representative) |
---|
239 | inline bool filledToCapacity() const { return itsBufferFull;} |
---|
240 | |
---|
241 | protected: |
---|
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 | |
---|
258 | private: |
---|
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 | |
---|
278 | } // namespace asap |
---|
279 | |
---|
280 | /////////////////////////////////////////////////////////////////////////////// |
---|
281 | // |
---|
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. |
---|
294 | LFNoiseEstimator::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 |
---|
304 | void LFNoiseEstimator::add(float in) |
---|
305 | { |
---|
306 | if (isnan(in)) { |
---|
307 | // normally it shouldn't happen |
---|
308 | return; |
---|
309 | } |
---|
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) |
---|
329 | size_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; |
---|
334 | AlwaysAssert( (nSamples > 0) && (nSamples <= itsVariances.size()), AipsError); |
---|
335 | return nSamples; |
---|
336 | } |
---|
337 | |
---|
338 | // this helper method builds the cache if |
---|
339 | // necessary using one of the methods |
---|
340 | void 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 |
---|
354 | float 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 |
---|
366 | float 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) |
---|
389 | void 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 |
---|
435 | void 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(); |
---|
440 | AlwaysAssert(nSamples <= itsSortedIndices.size(), AipsError); |
---|
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 | // |
---|
455 | // RunningBox - a running box calculator. This class implements |
---|
456 | // interations over the specified spectrum and calculates |
---|
457 | // running box filter statistics. |
---|
458 | // |
---|
459 | |
---|
460 | // set up the object with the references to actual data |
---|
461 | // and the number of channels in the running box |
---|
462 | RunningBox::RunningBox(const casa::Vector<casa::Float> &in_spectrum, |
---|
463 | const casa::Vector<casa::Bool> &in_mask, |
---|
464 | const std::pair<int,int> &in_edge, |
---|
465 | int in_max_box_nchan) throw(AipsError) : |
---|
466 | spectrum(in_spectrum), mask(in_mask), edge(in_edge), |
---|
467 | max_box_nchan(in_max_box_nchan) |
---|
468 | { |
---|
469 | rewind(); |
---|
470 | } |
---|
471 | |
---|
472 | void 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); |
---|
484 | |
---|
485 | if (initial_box_ch==edge.second) |
---|
486 | throw AipsError("RunningBox::rewind - too much channels are masked"); |
---|
487 | |
---|
488 | cur_channel=edge.first; |
---|
489 | start_advance=initial_box_ch-max_box_nchan/2; |
---|
490 | } |
---|
491 | |
---|
492 | // access to the statistics |
---|
493 | const casa::Float& RunningBox::getLinMean() const throw(AipsError) |
---|
494 | { |
---|
495 | DebugAssert(cur_channel<edge.second, AipsError); |
---|
496 | if (need2recalculate) updateDerivativeStatistics(); |
---|
497 | return linmean; |
---|
498 | } |
---|
499 | |
---|
500 | const casa::Float& RunningBox::getLinVariance() const throw(AipsError) |
---|
501 | { |
---|
502 | DebugAssert(cur_channel<edge.second, AipsError); |
---|
503 | if (need2recalculate) updateDerivativeStatistics(); |
---|
504 | return linvariance; |
---|
505 | } |
---|
506 | |
---|
507 | const casa::Float RunningBox::aboveMean() const throw(AipsError) |
---|
508 | { |
---|
509 | DebugAssert(cur_channel<edge.second, AipsError); |
---|
510 | if (need2recalculate) updateDerivativeStatistics(); |
---|
511 | return spectrum[cur_channel]-linmean; |
---|
512 | } |
---|
513 | |
---|
514 | int 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) |
---|
521 | int RunningBox::getNumberOfBoxPoints() const throw() |
---|
522 | { |
---|
523 | return box_chan_cntr; |
---|
524 | } |
---|
525 | |
---|
526 | // supplementary function to control running mean/median calculations. |
---|
527 | // It adds a specified channel to the running box and |
---|
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 |
---|
531 | void RunningBox::advanceRunningBox(int ch) throw(AipsError) |
---|
532 | { |
---|
533 | if (ch>=edge.first && ch<edge.second) |
---|
534 | if (mask[ch]) { // ch is a valid channel |
---|
535 | ++box_chan_cntr; |
---|
536 | sumf+=spectrum[ch]; |
---|
537 | sumf2+=square(spectrum[ch]); |
---|
538 | sumch+=Float(ch); |
---|
539 | sumch2+=square(Float(ch)); |
---|
540 | sumfch+=spectrum[ch]*Float(ch); |
---|
541 | need2recalculate=True; |
---|
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; |
---|
547 | sumf-=spectrum[ch2remove]; |
---|
548 | sumf2-=square(spectrum[ch2remove]); |
---|
549 | sumch-=Float(ch2remove); |
---|
550 | sumch2-=square(Float(ch2remove)); |
---|
551 | sumfch-=spectrum[ch2remove]*Float(ch2remove); |
---|
552 | need2recalculate=True; |
---|
553 | } |
---|
554 | } |
---|
555 | |
---|
556 | // next channel |
---|
557 | void RunningBox::next() throw(AipsError) |
---|
558 | { |
---|
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 |
---|
563 | } |
---|
564 | |
---|
565 | // checking whether there are still elements |
---|
566 | casa::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 |
---|
573 | void RunningBox::updateDerivativeStatistics() const throw(AipsError) |
---|
574 | { |
---|
575 | AlwaysAssert(box_chan_cntr, AipsError); |
---|
576 | |
---|
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; |
---|
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 | } |
---|
598 | } |
---|
599 | need2recalculate=False; |
---|
600 | } |
---|
601 | |
---|
602 | |
---|
603 | // |
---|
604 | /////////////////////////////////////////////////////////////////////////////// |
---|
605 | |
---|
606 | /////////////////////////////////////////////////////////////////////////////// |
---|
607 | // |
---|
608 | // LFAboveThreshold - a running mean/median algorithm for line detection |
---|
609 | // |
---|
610 | // |
---|
611 | |
---|
612 | |
---|
613 | // set up the detection criterion |
---|
614 | LFAboveThreshold::LFAboveThreshold(std::list<pair<int,int> > &in_lines, |
---|
615 | int in_min_nchan, |
---|
616 | casa::Float in_threshold, |
---|
617 | bool use_median, |
---|
618 | int noise_sample_size) throw() : |
---|
619 | min_nchan(in_min_nchan), threshold(in_threshold), |
---|
620 | lines(in_lines), running_box(NULL), itsUseMedian(use_median), |
---|
621 | itsNoiseSampleSize(noise_sample_size) {} |
---|
622 | |
---|
623 | LFAboveThreshold::~LFAboveThreshold() throw() |
---|
624 | { |
---|
625 | if (running_box!=NULL) delete running_box; |
---|
626 | } |
---|
627 | |
---|
628 | // replace the detection criterion |
---|
629 | void 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 | |
---|
636 | // get the sign of runningBox->aboveMean(). The RunningBox pointer |
---|
637 | // should be defined |
---|
638 | casa::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 | } |
---|
645 | |
---|
646 | |
---|
647 | // process a channel: update cur_line and is_detected before and |
---|
648 | // add a new line to the list, if necessary |
---|
649 | void LFAboveThreshold::processChannel(Bool detect, |
---|
650 | const casa::Vector<casa::Bool> &mask) throw(casa::AipsError) |
---|
651 | { |
---|
652 | try { |
---|
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; |
---|
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); |
---|
670 | } |
---|
671 | catch (const AipsError &ae) { |
---|
672 | throw; |
---|
673 | } |
---|
674 | catch (const exception &ex) { |
---|
675 | throw AipsError(String("LFAboveThreshold::processChannel - STL error: ")+ex.what()); |
---|
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 |
---|
681 | void LFAboveThreshold::processCurLine(const casa::Vector<casa::Bool> &mask) |
---|
682 | throw(casa::AipsError) |
---|
683 | { |
---|
684 | try { |
---|
685 | if (is_detected_before) { |
---|
686 | if (cur_line.second-cur_line.first>=min_nchan) { |
---|
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); |
---|
699 | else lines.back().second=cur_line.second; |
---|
700 | } |
---|
701 | is_detected_before=False; |
---|
702 | } |
---|
703 | } |
---|
704 | catch (const AipsError &ae) { |
---|
705 | throw; |
---|
706 | } |
---|
707 | catch (const exception &ex) { |
---|
708 | throw AipsError(String("LFAboveThreshold::processCurLine - STL error: ")+ex.what()); |
---|
709 | } |
---|
710 | } |
---|
711 | |
---|
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 |
---|
716 | const casa::Vector<Int>& LFAboveThreshold::getSigns() const throw() |
---|
717 | { |
---|
718 | return signs; |
---|
719 | } |
---|
720 | |
---|
721 | // find spectral lines and add them into list |
---|
722 | void LFAboveThreshold::findLines(const casa::Vector<casa::Float> &spectrum, |
---|
723 | const casa::Vector<casa::Bool> &mask, |
---|
724 | const std::pair<int,int> &edge, |
---|
725 | int max_box_nchan) |
---|
726 | throw(casa::AipsError) |
---|
727 | { |
---|
728 | const int minboxnchan=4; |
---|
729 | try { |
---|
730 | |
---|
731 | if (running_box!=NULL) delete running_box; |
---|
732 | running_box=new RunningBox(spectrum,mask,edge,max_box_nchan); |
---|
733 | |
---|
734 | // determine the off-line variance first |
---|
735 | // an assumption made: lines occupy a small part of the spectrum |
---|
736 | |
---|
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); |
---|
742 | |
---|
743 | for (;running_box->haveMore();running_box->next()) { |
---|
744 | ne.add(running_box->getLinVariance()); |
---|
745 | if (ne.filledToCapacity()) { |
---|
746 | break; |
---|
747 | } |
---|
748 | } |
---|
749 | |
---|
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 | } |
---|
755 | |
---|
756 | // actual search algorithm |
---|
757 | is_detected_before=False; |
---|
758 | |
---|
759 | // initiate the signs array |
---|
760 | signs.resize(spectrum.nelements()); |
---|
761 | signs=Vector<Int>(spectrum.nelements(),0); |
---|
762 | |
---|
763 | //ofstream os("dbg.dat"); |
---|
764 | for (running_box->rewind();running_box->haveMore(); |
---|
765 | running_box->next()) { |
---|
766 | const int ch=running_box->getChannel(); |
---|
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); |
---|
776 | processChannel(mask[ch] && (fabs(running_box->aboveMean()) >= |
---|
777 | threshold*offline_variance), mask); |
---|
778 | } else processCurLine(mask); // just finish what was accumulated before |
---|
779 | |
---|
780 | signs[ch]=getAboveMeanSign(); |
---|
781 | //os<<ch<<" "<<spectrum[ch]<<" "<<fabs(running_box->aboveMean())<<" "<< |
---|
782 | //threshold*offline_variance<<endl; |
---|
783 | } |
---|
784 | if (lines.size()) |
---|
785 | searchForWings(lines,signs,mask,edge); |
---|
786 | } |
---|
787 | catch (const AipsError &ae) { |
---|
788 | throw; |
---|
789 | } |
---|
790 | catch (const exception &ex) { |
---|
791 | throw AipsError(String("LFAboveThreshold::findLines - STL error: ")+ex.what()); |
---|
792 | } |
---|
793 | } |
---|
794 | |
---|
795 | // |
---|
796 | /////////////////////////////////////////////////////////////////////////////// |
---|
797 | |
---|
798 | /////////////////////////////////////////////////////////////////////////////// |
---|
799 | // |
---|
800 | // LFLineListOperations::IntersectsWith - An auxiliary object function |
---|
801 | // to test whether two lines have a non-void intersection |
---|
802 | // |
---|
803 | |
---|
804 | |
---|
805 | // line1 - range of the first line: start channel and stop+1 |
---|
806 | LFLineListOperations::IntersectsWith::IntersectsWith(const std::pair<int,int> &in_line1) : |
---|
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 |
---|
813 | bool LFLineListOperations::IntersectsWith::operator()(const std::pair<int,int> &line2) |
---|
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 | // |
---|
826 | // LFLineListOperations::BuildUnion - An auxiliary object function to build a union |
---|
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) |
---|
831 | LFLineListOperations::BuildUnion::BuildUnion(const std::pair<int,int> &line1) : |
---|
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 |
---|
836 | void LFLineListOperations::BuildUnion::operator()(const std::pair<int,int> &new_line) |
---|
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) |
---|
844 | const std::pair<int,int>& LFLineListOperations::BuildUnion::result() const throw() |
---|
845 | { |
---|
846 | return temp_line; |
---|
847 | } |
---|
848 | |
---|
849 | // |
---|
850 | /////////////////////////////////////////////////////////////////////////////// |
---|
851 | |
---|
852 | /////////////////////////////////////////////////////////////////////////////// |
---|
853 | // |
---|
854 | // LFLineListOperations::LaterThan - An auxiliary object function to test whether a |
---|
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 |
---|
860 | LFLineListOperations::LaterThan::LaterThan(const std::pair<int,int> &in_line1) : |
---|
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) |
---|
865 | bool LFLineListOperations::LaterThan::operator()(const std::pair<int,int> &line2) |
---|
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 |
---|
870 | |
---|
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 | // |
---|
882 | // STLineFinder - a class for automated spectral line search |
---|
883 | // |
---|
884 | // |
---|
885 | |
---|
886 | STLineFinder::STLineFinder() throw() : edge(0,0) |
---|
887 | { |
---|
888 | setOptions(); |
---|
889 | } |
---|
890 | |
---|
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 |
---|
901 | // Default is 8, but for a bad baseline shape this |
---|
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. |
---|
905 | // Setting a very large value doesn't usually provide |
---|
906 | // valid detections. |
---|
907 | // in_box_size the box size for running mean/median calculation. Default is |
---|
908 | // 1./5. of the whole spectrum size |
---|
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) |
---|
914 | void STLineFinder::setOptions(const casa::Float &in_threshold, |
---|
915 | const casa::Int &in_min_nchan, |
---|
916 | const casa::Int &in_avg_limit, |
---|
917 | const casa::Float &in_box_size, |
---|
918 | const casa::Float &in_noise_box, |
---|
919 | const casa::Bool &in_median) throw() |
---|
920 | { |
---|
921 | threshold=in_threshold; |
---|
922 | min_nchan=in_min_nchan; |
---|
923 | avg_limit=in_avg_limit; |
---|
924 | box_size=in_box_size; |
---|
925 | itsNoiseBox = in_noise_box; |
---|
926 | itsUseMedian = in_median; |
---|
927 | |
---|
928 | useScantable = true; |
---|
929 | } |
---|
930 | |
---|
931 | STLineFinder::~STLineFinder() throw(AipsError) {} |
---|
932 | |
---|
933 | // set scan to work with (in_scan parameter) |
---|
934 | void STLineFinder::setScan(const ScantableWrapper &in_scan) throw(AipsError) |
---|
935 | { |
---|
936 | scan=in_scan.getCP(); |
---|
937 | AlwaysAssert(!scan.null(),AipsError); |
---|
938 | } |
---|
939 | |
---|
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) |
---|
943 | void STLineFinder::setData(const std::vector<float> &in_spectrum) |
---|
944 | { |
---|
945 | spectrum = Vector<Float>(in_spectrum); |
---|
946 | useScantable = false; |
---|
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 |
---|
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 |
---|
955 | // can be achieved using a spectrum mask only |
---|
956 | int STLineFinder::findLines(const std::vector<bool> &in_mask, |
---|
957 | const std::vector<int> &in_edge, |
---|
958 | const casa::uInt &whichRow) throw(casa::AipsError) |
---|
959 | { |
---|
960 | if (useScantable && scan.null()) |
---|
961 | throw AipsError("STLineFinder::findLines - a scan should be set first," |
---|
962 | " use set_scan"); |
---|
963 | |
---|
964 | uInt nchan = useScantable ? scan->nchan(scan->getIF(whichRow)) : spectrum.nelements(); |
---|
965 | // set up mask and edge rejection |
---|
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) |
---|
974 | throw AipsError("STLineFinder::findLines - in_scan and in_mask, or in_spectrum " |
---|
975 | "and in_mask have different number of spectral channels."); |
---|
976 | |
---|
977 | // taking flagged channels into account |
---|
978 | if (useScantable) { |
---|
979 | vector<bool> flaggedChannels = scan->getMask(whichRow); |
---|
980 | if (flaggedChannels.size()) { |
---|
981 | // there is a mask set for this row |
---|
982 | if (flaggedChannels.size() != mask.nelements()) { |
---|
983 | throw AipsError("STLineFinder::findLines - internal inconsistency: number of " |
---|
984 | "mask elements do not match the number of channels"); |
---|
985 | } |
---|
986 | for (size_t ch = 0; ch<mask.nelements(); ++ch) { |
---|
987 | mask[ch] &= flaggedChannels[ch]; |
---|
988 | } |
---|
989 | } |
---|
990 | } |
---|
991 | |
---|
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" |
---|
995 | "should not exceed 2"); |
---|
996 | if (!in_edge.size()) { |
---|
997 | // all spectra, no rejection |
---|
998 | edge.first=0; |
---|
999 | edge.second=nchan; |
---|
1000 | } else { |
---|
1001 | edge.first=in_edge[0]; |
---|
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"); |
---|
1007 | if (in_edge.size()==2) { |
---|
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"); |
---|
1012 | edge.second=nchan-edge.second; |
---|
1013 | } else edge.second=nchan-edge.first; |
---|
1014 | if (edge.second<0 || (edge.first>=edge.second)) |
---|
1015 | throw AipsError("STLineFinder::findLines - all channels are rejected by the in_edge parameter"); |
---|
1016 | } |
---|
1017 | |
---|
1018 | // |
---|
1019 | int max_box_nchan=int(nchan*box_size); // number of channels in running |
---|
1020 | // box |
---|
1021 | if (max_box_nchan<2) |
---|
1022 | throw AipsError("STLineFinder::findLines - box_size is too small"); |
---|
1023 | |
---|
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 | |
---|
1030 | if (useScantable) { |
---|
1031 | spectrum.resize(); |
---|
1032 | spectrum = Vector<Float>(scan->getSpectrum(whichRow)); |
---|
1033 | } |
---|
1034 | |
---|
1035 | lines.resize(0); // search from the scratch |
---|
1036 | last_row_used=whichRow; |
---|
1037 | Vector<Bool> temp_mask(mask); |
---|
1038 | |
---|
1039 | Bool first_pass=True; |
---|
1040 | Int avg_factor=1; // this number of adjacent channels is averaged together |
---|
1041 | // the total number of the channels is not altered |
---|
1042 | // instead, min_nchan is also scaled |
---|
1043 | // it helps to search for broad lines |
---|
1044 | Vector<Int> signs; // a buffer for signs of the value - mean quantity |
---|
1045 | // see LFAboveThreshold for details |
---|
1046 | // We need only signs resulted from last iteration |
---|
1047 | // because all previous values may be corrupted by the |
---|
1048 | // presence of spectral lines |
---|
1049 | while (true) { |
---|
1050 | // a buffer for new lines found at this iteration |
---|
1051 | std::list<pair<int,int> > new_lines; |
---|
1052 | |
---|
1053 | try { |
---|
1054 | // line find algorithm |
---|
1055 | LFAboveThreshold lfalg(new_lines,avg_factor*min_nchan, threshold, itsUseMedian,noise_box); |
---|
1056 | lfalg.findLines(spectrum,temp_mask,edge,max_box_nchan); |
---|
1057 | signs.resize(lfalg.getSigns().nelements()); |
---|
1058 | signs=lfalg.getSigns(); |
---|
1059 | first_pass=False; |
---|
1060 | if (!new_lines.size()) |
---|
1061 | throw AipsError("spurious"); // nothing new - use the same |
---|
1062 | // code as for a real exception |
---|
1063 | } |
---|
1064 | catch(const AipsError &ae) { |
---|
1065 | if (first_pass) throw; |
---|
1066 | // nothing new - proceed to the next step of averaging, if any |
---|
1067 | // (to search for broad lines) |
---|
1068 | if (avg_factor>=avg_limit) break; // averaging up to avg_limit |
---|
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; |
---|
1075 | } |
---|
1076 | keepStrongestOnly(temp_mask,new_lines,max_box_nchan); |
---|
1077 | // update the list (lines) merging intervals, if necessary |
---|
1078 | addNewSearchResult(new_lines,lines); |
---|
1079 | // get a new mask |
---|
1080 | temp_mask=getMask(); |
---|
1081 | } |
---|
1082 | |
---|
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 |
---|
1086 | |
---|
1087 | if (lines.size()) |
---|
1088 | LFLineListOperations::searchForWings(lines,signs,mask,edge); |
---|
1089 | |
---|
1090 | return int(lines.size()); |
---|
1091 | } |
---|
1092 | |
---|
1093 | // auxiliary function to fit and subtract a polynomial from the current |
---|
1094 | // spectrum. It uses the Fitter class. This action is required before |
---|
1095 | // reducing the spectral resolution if the baseline shape is bad |
---|
1096 | void STLineFinder::subtractBaseline(const casa::Vector<casa::Bool> &temp_mask, |
---|
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 |
---|
1101 | Fitter sdf; |
---|
1102 | std::vector<float> absc(spectrum.nelements()); |
---|
1103 | for (unsigned int i=0;i<absc.size();++i) |
---|
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); |
---|
1111 | if (!sdf.lfit()) return; // fit failed, use old spectrum |
---|
1112 | spectrum=casa::Vector<casa::Float>(sdf.getResidual()); |
---|
1113 | } |
---|
1114 | |
---|
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 |
---|
1121 | void STLineFinder::averageAdjacentChannels(casa::Vector<casa::Bool> &mask2update, |
---|
1122 | const casa::Int &boxsize) |
---|
1123 | throw(casa::AipsError) |
---|
1124 | { |
---|
1125 | DebugAssert(mask2update.nelements()==spectrum.nelements(), AipsError); |
---|
1126 | DebugAssert(boxsize!=0,AipsError); |
---|
1127 | |
---|
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 |
---|
1134 | mean+=spectrum[k]; |
---|
1135 | ++nboxch; |
---|
1136 | } |
---|
1137 | if (nboxch<boxsize) // mask these channels |
---|
1138 | for (int k=n;k<n+boxsize && k<edge.second;++k) |
---|
1139 | mask2update[k]=False; |
---|
1140 | else { |
---|
1141 | mean/=Float(boxsize); |
---|
1142 | for (int k=n;k<n+boxsize && k<edge.second;++k) |
---|
1143 | spectrum[k]=mean; |
---|
1144 | } |
---|
1145 | } |
---|
1146 | } |
---|
1147 | |
---|
1148 | |
---|
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 |
---|
1154 | std::vector<bool> STLineFinder::getMask(bool invert) |
---|
1155 | const throw(casa::AipsError) |
---|
1156 | { |
---|
1157 | try { |
---|
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 " |
---|
1167 | "lines first, use find_lines"); |
---|
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; |
---|
1186 | } |
---|
1187 | catch (const AipsError &ae) { |
---|
1188 | throw; |
---|
1189 | } |
---|
1190 | catch (const exception &ex) { |
---|
1191 | throw AipsError(String("STLineFinder::getMask - STL error: ")+ex.what()); |
---|
1192 | } |
---|
1193 | } |
---|
1194 | |
---|
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). |
---|
1197 | std::vector<double> STLineFinder::getLineRanges() |
---|
1198 | const throw(casa::AipsError) |
---|
1199 | { |
---|
1200 | std::vector<double> vel; |
---|
1201 | if (useScantable) { |
---|
1202 | // convert to required abscissa units |
---|
1203 | vel = scan->getAbcissa(last_row_used); |
---|
1204 | } else { |
---|
1205 | for (uInt i = 0; i < spectrum.nelements(); ++i) |
---|
1206 | vel.push_back((double)i); |
---|
1207 | } |
---|
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()) |
---|
1215 | throw AipsError("STLineFinder::getLineRanges - getAbcissa provided less channels than reqired"); |
---|
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 |
---|
1222 | std::vector<int> STLineFinder::getLineRangesInChannels() |
---|
1223 | const throw(casa::AipsError) |
---|
1224 | { |
---|
1225 | try { |
---|
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 | } |
---|
1232 | |
---|
1233 | if (!lines.size()) |
---|
1234 | throw AipsError("STLineFinder::getLineRangesInChannels - one have to search for " |
---|
1235 | "lines first, use find_lines"); |
---|
1236 | |
---|
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()); |
---|
1251 | } |
---|
1252 | } |
---|
1253 | |
---|
1254 | |
---|
1255 | |
---|
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, |
---|
1258 | // they will be find again at the next iteration. This approach |
---|
1259 | // increases the number of iterations required, but is able to remove |
---|
1260 | // spurious detections likely to occur near strong lines. |
---|
1261 | // Later a better criterion may be implemented, e.g. |
---|
1262 | // taking into consideration the brightness of different lines. Now |
---|
1263 | // use the simplest solution |
---|
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 |
---|
1269 | void STLineFinder::keepStrongestOnly(const casa::Vector<casa::Bool> &temp_mask, |
---|
1270 | std::list<std::pair<int, int> > &lines2update, |
---|
1271 | int max_box_nchan) |
---|
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 |
---|
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 | } |
---|
1302 | } |
---|
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; |
---|
1310 | } |
---|
1311 | catch (const exception &ex) { |
---|
1312 | throw AipsError(String("STLineFinder::keepStrongestOnly - STL error: ")+ex.what()); |
---|
1313 | } |
---|
1314 | |
---|
1315 | } |
---|
1316 | |
---|
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 | |
---|
1327 | // concatenate two lists preserving the order. If two lines appear to |
---|
1328 | // be adjacent, they are joined into the new one |
---|
1329 | void LFLineListOperations::addNewSearchResult(const std::list<pair<int, int> > &newlines, |
---|
1330 | std::list<std::pair<int, int> > &lines_list) |
---|
1331 | throw(AipsError) |
---|
1332 | { |
---|
1333 | try { |
---|
1334 | for (std::list<pair<int,int> >::const_iterator cli=newlines.begin(); |
---|
1335 | cli!=newlines.end();++cli) { |
---|
1336 | |
---|
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))); |
---|
1344 | |
---|
1345 | // extract all lines which intersect or touch a new one into |
---|
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); |
---|
1351 | |
---|
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(); |
---|
1355 | |
---|
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); |
---|
1360 | } |
---|
1361 | } |
---|
1362 | catch (const AipsError &ae) { |
---|
1363 | throw; |
---|
1364 | } |
---|
1365 | catch (const exception &ex) { |
---|
1366 | throw AipsError(String("LFLineListOperations::addNewSearchResult - STL error: ")+ex.what()); |
---|
1367 | } |
---|
1368 | } |
---|
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 |
---|
1375 | void LFLineListOperations::searchForWings(std::list<std::pair<int, int> > &newlines, |
---|
1376 | const casa::Vector<casa::Int> &signs, |
---|
1377 | const casa::Vector<casa::Bool> &mask, |
---|
1378 | const std::pair<int,int> &edge) throw(casa::AipsError) |
---|
1379 | { |
---|
1380 | try { |
---|
1381 | for (std::list<pair<int,int> >::iterator li=newlines.begin(); |
---|
1382 | li!=newlines.end();++li) { |
---|
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 | } |
---|
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; |
---|
1406 | } |
---|
1407 | catch (const exception &ex) { |
---|
1408 | throw AipsError(String("LFLineListOperations::extendLines - STL error: ")+ex.what()); |
---|
1409 | } |
---|
1410 | } |
---|
1411 | |
---|
1412 | // |
---|
1413 | /////////////////////////////////////////////////////////////////////////////// |
---|