##// END OF EJS Templates
schain
RSS
Fix typo in removeInterference
Juan C. Espinoza -
r1295:200e540eee3f
parent child
Show More
Side by Side Unified
Context file:
r1295:200e540eee3f Loading diff...:
Show file before | Show file after | Hide comments
@@ -1,882 +1,882
1 1 import time
2 2 import itertools
3 3
4 4 import numpy
5 5
6 6 from schainpy.model.proc.jroproc_base import ProcessingUnit, MPDecorator, Operation
7 7 from schainpy.model.data.jrodata import Spectra
8 8 from schainpy.model.data.jrodata import hildebrand_sekhon
9 9 from schainpy.utils import log
10 10
11 11
12 12 class SpectraProc(ProcessingUnit):
13 13
14 14
15 15 def __init__(self):
16 16
17 17 ProcessingUnit.__init__(self)
18 18
19 19 self.buffer = None
20 20 self.firstdatatime = None
21 21 self.profIndex = 0
22 22 self.dataOut = Spectra()
23 23 self.id_min = None
24 24 self.id_max = None
25 25 self.setupReq = False #Agregar a todas las unidades de proc
26 26
27 27 def __updateSpecFromVoltage(self):
28 28
29 29 self.dataOut.timeZone = self.dataIn.timeZone
30 30 self.dataOut.dstFlag = self.dataIn.dstFlag
31 31 self.dataOut.errorCount = self.dataIn.errorCount
32 32 self.dataOut.useLocalTime = self.dataIn.useLocalTime
33 33 try:
34 34 self.dataOut.processingHeaderObj = self.dataIn.processingHeaderObj.copy()
35 35 except:
36 36 pass
37 37 self.dataOut.radarControllerHeaderObj = self.dataIn.radarControllerHeaderObj.copy()
38 38 self.dataOut.systemHeaderObj = self.dataIn.systemHeaderObj.copy()
39 39 self.dataOut.channelList = self.dataIn.channelList
40 40 self.dataOut.heightList = self.dataIn.heightList
41 41 self.dataOut.dtype = numpy.dtype([('real', '<f4'), ('imag', '<f4')])
42 42
43 43 self.dataOut.nBaud = self.dataIn.nBaud
44 44 self.dataOut.nCode = self.dataIn.nCode
45 45 self.dataOut.code = self.dataIn.code
46 46 self.dataOut.nProfiles = self.dataOut.nFFTPoints
47 47
48 48 self.dataOut.flagDiscontinuousBlock = self.dataIn.flagDiscontinuousBlock
49 49 self.dataOut.utctime = self.firstdatatime
50 50 # asumo q la data esta decodificada
51 51 self.dataOut.flagDecodeData = self.dataIn.flagDecodeData
52 52 # asumo q la data esta sin flip
53 53 self.dataOut.flagDeflipData = self.dataIn.flagDeflipData
54 54 self.dataOut.flagShiftFFT = False
55 55
56 56 self.dataOut.nCohInt = self.dataIn.nCohInt
57 57 self.dataOut.nIncohInt = 1
58 58
59 59 self.dataOut.windowOfFilter = self.dataIn.windowOfFilter
60 60
61 61 self.dataOut.frequency = self.dataIn.frequency
62 62 self.dataOut.realtime = self.dataIn.realtime
63 63
64 64 self.dataOut.azimuth = self.dataIn.azimuth
65 65 self.dataOut.zenith = self.dataIn.zenith
66 66
67 67 self.dataOut.beam.codeList = self.dataIn.beam.codeList
68 68 self.dataOut.beam.azimuthList = self.dataIn.beam.azimuthList
69 69 self.dataOut.beam.zenithList = self.dataIn.beam.zenithList
70 70
71 71 def __getFft(self):
72 72 """
73 73 Convierte valores de Voltaje a Spectra
74 74
75 75 Affected:
76 76 self.dataOut.data_spc
77 77 self.dataOut.data_cspc
78 78 self.dataOut.data_dc
79 79 self.dataOut.heightList
80 80 self.profIndex
81 81 self.buffer
82 82 self.dataOut.flagNoData
83 83 """
84 84 fft_volt = numpy.fft.fft(
85 85 self.buffer, n=self.dataOut.nFFTPoints, axis=1)
86 86 fft_volt = fft_volt.astype(numpy.dtype('complex'))
87 87 dc = fft_volt[:, 0, :]
88 88
89 89 # calculo de self-spectra
90 90 fft_volt = numpy.fft.fftshift(fft_volt, axes=(1,))
91 91 spc = fft_volt * numpy.conjugate(fft_volt)
92 92 spc = spc.real
93 93
94 94 blocksize = 0
95 95 blocksize += dc.size
96 96 blocksize += spc.size
97 97
98 98 cspc = None
99 99 pairIndex = 0
100 100 if self.dataOut.pairsList != None:
101 101 # calculo de cross-spectra
102 102 cspc = numpy.zeros(
103 103 (self.dataOut.nPairs, self.dataOut.nFFTPoints, self.dataOut.nHeights), dtype='complex')
104 104 for pair in self.dataOut.pairsList:
105 105 if pair[0] not in self.dataOut.channelList:
106 106 raise ValueError("Error getting CrossSpectra: pair 0 of %s is not in channelList = %s" % (
107 107 str(pair), str(self.dataOut.channelList)))
108 108 if pair[1] not in self.dataOut.channelList:
109 109 raise ValueError("Error getting CrossSpectra: pair 1 of %s is not in channelList = %s" % (
110 110 str(pair), str(self.dataOut.channelList)))
111 111
112 112 cspc[pairIndex, :, :] = fft_volt[pair[0], :, :] * \
113 113 numpy.conjugate(fft_volt[pair[1], :, :])
114 114 pairIndex += 1
115 115 blocksize += cspc.size
116 116
117 117 self.dataOut.data_spc = spc
118 118 self.dataOut.data_cspc = cspc
119 119 self.dataOut.data_dc = dc
120 120 self.dataOut.blockSize = blocksize
121 121 self.dataOut.flagShiftFFT = False
122 122
123 123 def run(self, nProfiles=None, nFFTPoints=None, pairsList=[], ippFactor=None, shift_fft=False):
124 124
125 125 if self.dataIn.type == "Spectra":
126 126 self.dataOut.copy(self.dataIn)
127 127 if shift_fft:
128 128 #desplaza a la derecha en el eje 2 determinadas posiciones
129 129 shift = int(self.dataOut.nFFTPoints/2)
130 130 self.dataOut.data_spc = numpy.roll(self.dataOut.data_spc, shift , axis=1)
131 131
132 132 if self.dataOut.data_cspc is not None:
133 133 #desplaza a la derecha en el eje 2 determinadas posiciones
134 134 self.dataOut.data_cspc = numpy.roll(self.dataOut.data_cspc, shift, axis=1)
135 135
136 136 return True
137 137
138 138 if self.dataIn.type == "Voltage":
139 139
140 140 self.dataOut.flagNoData = True
141 141
142 142 if nFFTPoints == None:
143 143 raise ValueError("This SpectraProc.run() need nFFTPoints input variable")
144 144
145 145 if nProfiles == None:
146 146 nProfiles = nFFTPoints
147 147
148 148 if ippFactor == None:
149 149 ippFactor = 1
150 150
151 151 self.dataOut.ippFactor = ippFactor
152 152
153 153 self.dataOut.nFFTPoints = nFFTPoints
154 154 self.dataOut.pairsList = pairsList
155 155
156 156 if self.buffer is None:
157 157 self.buffer = numpy.zeros((self.dataIn.nChannels,
158 158 nProfiles,
159 159 self.dataIn.nHeights),
160 160 dtype='complex')
161 161
162 162 if self.dataIn.flagDataAsBlock:
163 163 nVoltProfiles = self.dataIn.data.shape[1]
164 164
165 165 if nVoltProfiles == nProfiles:
166 166 self.buffer = self.dataIn.data.copy()
167 167 self.profIndex = nVoltProfiles
168 168
169 169 elif nVoltProfiles < nProfiles:
170 170
171 171 if self.profIndex == 0:
172 172 self.id_min = 0
173 173 self.id_max = nVoltProfiles
174 174
175 175 self.buffer[:, self.id_min:self.id_max,
176 176 :] = self.dataIn.data
177 177 self.profIndex += nVoltProfiles
178 178 self.id_min += nVoltProfiles
179 179 self.id_max += nVoltProfiles
180 180 else:
181 181 raise ValueError("The type object %s has %d profiles, it should just has %d profiles" % (
182 182 self.dataIn.type, self.dataIn.data.shape[1], nProfiles))
183 183 self.dataOut.flagNoData = True
184 184 return 0
185 185 else:
186 186 self.buffer[:, self.profIndex, :] = self.dataIn.data.copy()
187 187 self.profIndex += 1
188 188
189 189 if self.firstdatatime == None:
190 190 self.firstdatatime = self.dataIn.utctime
191 191
192 192 if self.profIndex == nProfiles:
193 193 self.__updateSpecFromVoltage()
194 194 self.__getFft()
195 195
196 196 self.dataOut.flagNoData = False
197 197 self.firstdatatime = None
198 198 self.profIndex = 0
199 199
200 200 return True
201 201
202 202 raise ValueError("The type of input object '%s' is not valid" % (
203 203 self.dataIn.type))
204 204
205 205 def __selectPairs(self, pairsList):
206 206
207 207 if not pairsList:
208 208 return
209 209
210 210 pairs = []
211 211 pairsIndex = []
212 212
213 213 for pair in pairsList:
214 214 if pair[0] not in self.dataOut.channelList or pair[1] not in self.dataOut.channelList:
215 215 continue
216 216 pairs.append(pair)
217 217 pairsIndex.append(pairs.index(pair))
218 218
219 219 self.dataOut.data_cspc = self.dataOut.data_cspc[pairsIndex]
220 220 self.dataOut.pairsList = pairs
221 221
222 222 return
223 223
224 224 def selectFFTs(self, minFFT, maxFFT ):
225 225 """
226 226 Selecciona un bloque de datos en base a un grupo de valores de puntos FFTs segun el rango
227 227 minFFT<= FFT <= maxFFT
228 228 """
229 229
230 230 if (minFFT > maxFFT):
231 231 raise ValueError("Error selecting heights: Height range (%d,%d) is not valid" % (minFFT, maxFFT))
232 232
233 233 if (minFFT < self.dataOut.getFreqRange()[0]):
234 234 minFFT = self.dataOut.getFreqRange()[0]
235 235
236 236 if (maxFFT > self.dataOut.getFreqRange()[-1]):
237 237 maxFFT = self.dataOut.getFreqRange()[-1]
238 238
239 239 minIndex = 0
240 240 maxIndex = 0
241 241 FFTs = self.dataOut.getFreqRange()
242 242
243 243 inda = numpy.where(FFTs >= minFFT)
244 244 indb = numpy.where(FFTs <= maxFFT)
245 245
246 246 try:
247 247 minIndex = inda[0][0]
248 248 except:
249 249 minIndex = 0
250 250
251 251 try:
252 252 maxIndex = indb[0][-1]
253 253 except:
254 254 maxIndex = len(FFTs)
255 255
256 256 self.selectFFTsByIndex(minIndex, maxIndex)
257 257
258 258 return 1
259 259
260 260 def getBeaconSignal(self, tauindex=0, channelindex=0, hei_ref=None):
261 261 newheis = numpy.where(
262 262 self.dataOut.heightList > self.dataOut.radarControllerHeaderObj.Taus[tauindex])
263 263
264 264 if hei_ref != None:
265 265 newheis = numpy.where(self.dataOut.heightList > hei_ref)
266 266
267 267 minIndex = min(newheis[0])
268 268 maxIndex = max(newheis[0])
269 269 data_spc = self.dataOut.data_spc[:, :, minIndex:maxIndex + 1]
270 270 heightList = self.dataOut.heightList[minIndex:maxIndex + 1]
271 271
272 272 # determina indices
273 273 nheis = int(self.dataOut.radarControllerHeaderObj.txB /
274 274 (self.dataOut.heightList[1] - self.dataOut.heightList[0]))
275 275 avg_dB = 10 * \
276 276 numpy.log10(numpy.sum(data_spc[channelindex, :, :], axis=0))
277 277 beacon_dB = numpy.sort(avg_dB)[-nheis:]
278 278 beacon_heiIndexList = []
279 279 for val in avg_dB.tolist():
280 280 if val >= beacon_dB[0]:
281 281 beacon_heiIndexList.append(avg_dB.tolist().index(val))
282 282
283 283 #data_spc = data_spc[:,:,beacon_heiIndexList]
284 284 data_cspc = None
285 285 if self.dataOut.data_cspc is not None:
286 286 data_cspc = self.dataOut.data_cspc[:, :, minIndex:maxIndex + 1]
287 287 #data_cspc = data_cspc[:,:,beacon_heiIndexList]
288 288
289 289 data_dc = None
290 290 if self.dataOut.data_dc is not None:
291 291 data_dc = self.dataOut.data_dc[:, minIndex:maxIndex + 1]
292 292 #data_dc = data_dc[:,beacon_heiIndexList]
293 293
294 294 self.dataOut.data_spc = data_spc
295 295 self.dataOut.data_cspc = data_cspc
296 296 self.dataOut.data_dc = data_dc
297 297 self.dataOut.heightList = heightList
298 298 self.dataOut.beacon_heiIndexList = beacon_heiIndexList
299 299
300 300 return 1
301 301
302 302 def selectFFTsByIndex(self, minIndex, maxIndex):
303 303 """
304 304
305 305 """
306 306
307 307 if (minIndex < 0) or (minIndex > maxIndex):
308 308 raise ValueError("Error selecting heights: Index range (%d,%d) is not valid" % (minIndex, maxIndex))
309 309
310 310 if (maxIndex >= self.dataOut.nProfiles):
311 311 maxIndex = self.dataOut.nProfiles-1
312 312
313 313 #Spectra
314 314 data_spc = self.dataOut.data_spc[:,minIndex:maxIndex+1,:]
315 315
316 316 data_cspc = None
317 317 if self.dataOut.data_cspc is not None:
318 318 data_cspc = self.dataOut.data_cspc[:,minIndex:maxIndex+1,:]
319 319
320 320 data_dc = None
321 321 if self.dataOut.data_dc is not None:
322 322 data_dc = self.dataOut.data_dc[minIndex:maxIndex+1,:]
323 323
324 324 self.dataOut.data_spc = data_spc
325 325 self.dataOut.data_cspc = data_cspc
326 326 self.dataOut.data_dc = data_dc
327 327
328 328 self.dataOut.ippSeconds = self.dataOut.ippSeconds*(self.dataOut.nFFTPoints / numpy.shape(data_cspc)[1])
329 329 self.dataOut.nFFTPoints = numpy.shape(data_cspc)[1]
330 330 self.dataOut.profilesPerBlock = numpy.shape(data_cspc)[1]
331 331
332 332 return 1
333 333
334 334 def getNoise(self, minHei=None, maxHei=None, minVel=None, maxVel=None):
335 335 # validacion de rango
336 336 if minHei == None:
337 337 minHei = self.dataOut.heightList[0]
338 338
339 339 if maxHei == None:
340 340 maxHei = self.dataOut.heightList[-1]
341 341
342 342 if (minHei < self.dataOut.heightList[0]) or (minHei > maxHei):
343 343 print('minHei: %.2f is out of the heights range' % (minHei))
344 344 print('minHei is setting to %.2f' % (self.dataOut.heightList[0]))
345 345 minHei = self.dataOut.heightList[0]
346 346
347 347 if (maxHei > self.dataOut.heightList[-1]) or (maxHei < minHei):
348 348 print('maxHei: %.2f is out of the heights range' % (maxHei))
349 349 print('maxHei is setting to %.2f' % (self.dataOut.heightList[-1]))
350 350 maxHei = self.dataOut.heightList[-1]
351 351
352 352 # validacion de velocidades
353 353 velrange = self.dataOut.getVelRange(1)
354 354
355 355 if minVel == None:
356 356 minVel = velrange[0]
357 357
358 358 if maxVel == None:
359 359 maxVel = velrange[-1]
360 360
361 361 if (minVel < velrange[0]) or (minVel > maxVel):
362 362 print('minVel: %.2f is out of the velocity range' % (minVel))
363 363 print('minVel is setting to %.2f' % (velrange[0]))
364 364 minVel = velrange[0]
365 365
366 366 if (maxVel > velrange[-1]) or (maxVel < minVel):
367 367 print('maxVel: %.2f is out of the velocity range' % (maxVel))
368 368 print('maxVel is setting to %.2f' % (velrange[-1]))
369 369 maxVel = velrange[-1]
370 370
371 371 # seleccion de indices para rango
372 372 minIndex = 0
373 373 maxIndex = 0
374 374 heights = self.dataOut.heightList
375 375
376 376 inda = numpy.where(heights >= minHei)
377 377 indb = numpy.where(heights <= maxHei)
378 378
379 379 try:
380 380 minIndex = inda[0][0]
381 381 except:
382 382 minIndex = 0
383 383
384 384 try:
385 385 maxIndex = indb[0][-1]
386 386 except:
387 387 maxIndex = len(heights)
388 388
389 389 if (minIndex < 0) or (minIndex > maxIndex):
390 390 raise ValueError("some value in (%d,%d) is not valid" % (
391 391 minIndex, maxIndex))
392 392
393 393 if (maxIndex >= self.dataOut.nHeights):
394 394 maxIndex = self.dataOut.nHeights - 1
395 395
396 396 # seleccion de indices para velocidades
397 397 indminvel = numpy.where(velrange >= minVel)
398 398 indmaxvel = numpy.where(velrange <= maxVel)
399 399 try:
400 400 minIndexVel = indminvel[0][0]
401 401 except:
402 402 minIndexVel = 0
403 403
404 404 try:
405 405 maxIndexVel = indmaxvel[0][-1]
406 406 except:
407 407 maxIndexVel = len(velrange)
408 408
409 409 # seleccion del espectro
410 410 data_spc = self.dataOut.data_spc[:,
411 411 minIndexVel:maxIndexVel + 1, minIndex:maxIndex + 1]
412 412 # estimacion de ruido
413 413 noise = numpy.zeros(self.dataOut.nChannels)
414 414
415 415 for channel in range(self.dataOut.nChannels):
416 416 daux = data_spc[channel, :, :]
417 417 sortdata = numpy.sort(daux, axis=None)
418 418 noise[channel] = hildebrand_sekhon(sortdata, self.dataOut.nIncohInt)
419 419
420 420 self.dataOut.noise_estimation = noise.copy()
421 421
422 422 return 1
423 423
424 424 class removeDC(Operation):
425 425
426 426 def run(self, dataOut, mode=2):
427 427 self.dataOut = dataOut
428 428 jspectra = self.dataOut.data_spc
429 429 jcspectra = self.dataOut.data_cspc
430 430
431 431 num_chan = jspectra.shape[0]
432 432 num_hei = jspectra.shape[2]
433 433
434 434 if jcspectra is not None:
435 435 jcspectraExist = True
436 436 num_pairs = jcspectra.shape[0]
437 437 else:
438 438 jcspectraExist = False
439 439
440 440 freq_dc = int(jspectra.shape[1] / 2)
441 441 ind_vel = numpy.array([-2, -1, 1, 2]) + freq_dc
442 442 ind_vel = ind_vel.astype(int)
443 443
444 444 if ind_vel[0] < 0:
445 445 ind_vel[list(range(0, 1))] = ind_vel[list(range(0, 1))] + self.num_prof
446 446
447 447 if mode == 1:
448 448 jspectra[:, freq_dc, :] = (
449 449 jspectra[:, ind_vel[1], :] + jspectra[:, ind_vel[2], :]) / 2 # CORRECCION
450 450
451 451 if jcspectraExist:
452 452 jcspectra[:, freq_dc, :] = (
453 453 jcspectra[:, ind_vel[1], :] + jcspectra[:, ind_vel[2], :]) / 2
454 454
455 455 if mode == 2:
456 456
457 457 vel = numpy.array([-2, -1, 1, 2])
458 458 xx = numpy.zeros([4, 4])
459 459
460 460 for fil in range(4):
461 461 xx[fil, :] = vel[fil]**numpy.asarray(list(range(4)))
462 462
463 463 xx_inv = numpy.linalg.inv(xx)
464 464 xx_aux = xx_inv[0, :]
465 465
466 466 for ich in range(num_chan):
467 467 yy = jspectra[ich, ind_vel, :]
468 468 jspectra[ich, freq_dc, :] = numpy.dot(xx_aux, yy)
469 469
470 470 junkid = jspectra[ich, freq_dc, :] <= 0
471 471 cjunkid = sum(junkid)
472 472
473 473 if cjunkid.any():
474 474 jspectra[ich, freq_dc, junkid.nonzero()] = (
475 475 jspectra[ich, ind_vel[1], junkid] + jspectra[ich, ind_vel[2], junkid]) / 2
476 476
477 477 if jcspectraExist:
478 478 for ip in range(num_pairs):
479 479 yy = jcspectra[ip, ind_vel, :]
480 480 jcspectra[ip, freq_dc, :] = numpy.dot(xx_aux, yy)
481 481
482 482 self.dataOut.data_spc = jspectra
483 483 self.dataOut.data_cspc = jcspectra
484 484
485 485 return self.dataOut
486 486
487 487 class removeInterference(Operation):
488 488
489 489 def removeInterference2(self):
490 490
491 491 cspc = self.dataOut.data_cspc
492 492 spc = self.dataOut.data_spc
493 493 Heights = numpy.arange(cspc.shape[2])
494 494 realCspc = numpy.abs(cspc)
495 495
496 496 for i in range(cspc.shape[0]):
497 497 LinePower= numpy.sum(realCspc[i], axis=0)
498 498 Threshold = numpy.amax(LinePower)-numpy.sort(LinePower)[len(Heights)-int(len(Heights)*0.1)]
499 499 SelectedHeights = Heights[ numpy.where( LinePower < Threshold ) ]
500 500 InterferenceSum = numpy.sum( realCspc[i,:,SelectedHeights], axis=0 )
501 501 InterferenceThresholdMin = numpy.sort(InterferenceSum)[int(len(InterferenceSum)*0.98)]
502 502 InterferenceThresholdMax = numpy.sort(InterferenceSum)[int(len(InterferenceSum)*0.99)]
503 503
504 504
505 505 InterferenceRange = numpy.where( ([InterferenceSum > InterferenceThresholdMin]))# , InterferenceSum < InterferenceThresholdMax]) )
506 506 #InterferenceRange = numpy.where( ([InterferenceRange < InterferenceThresholdMax]))
507 507 if len(InterferenceRange)<int(cspc.shape[1]*0.3):
508 508 cspc[i,InterferenceRange,:] = numpy.NaN
509 509
510 510 self.dataOut.data_cspc = cspc
511 511
512 def removeInterference(self):
512 def removeInterference(self, interf = 2, hei_interf = None, nhei_interf = None, offhei_interf = None):
513 513
514 514 jspectra = self.dataOut.data_spc
515 515 jcspectra = self.dataOut.data_cspc
516 516 jnoise = self.dataOut.getNoise()
517 517 num_incoh = self.dataOut.nIncohInt
518 518
519 519 num_channel = jspectra.shape[0]
520 520 num_prof = jspectra.shape[1]
521 521 num_hei = jspectra.shape[2]
522 522
523 523 # hei_interf
524 524 if hei_interf is None:
525 525 count_hei = int(num_hei / 2)
526 526 hei_interf = numpy.asmatrix(list(range(count_hei))) + num_hei - count_hei
527 527 hei_interf = numpy.asarray(hei_interf)[0]
528 528 # nhei_interf
529 529 if (nhei_interf == None):
530 530 nhei_interf = 5
531 531 if (nhei_interf < 1):
532 532 nhei_interf = 1
533 533 if (nhei_interf > count_hei):
534 534 nhei_interf = count_hei
535 535 if (offhei_interf == None):
536 536 offhei_interf = 0
537 537
538 538 ind_hei = list(range(num_hei))
539 539 # mask_prof = numpy.asarray(range(num_prof - 2)) + 1
540 540 # mask_prof[range(num_prof/2 - 1,len(mask_prof))] += 1
541 541 mask_prof = numpy.asarray(list(range(num_prof)))
542 542 num_mask_prof = mask_prof.size
543 543 comp_mask_prof = [0, num_prof / 2]
544 544
545 545 # noise_exist: Determina si la variable jnoise ha sido definida y contiene la informacion del ruido de cada canal
546 546 if (jnoise.size < num_channel or numpy.isnan(jnoise).any()):
547 547 jnoise = numpy.nan
548 548 noise_exist = jnoise[0] < numpy.Inf
549 549
550 550 # Subrutina de Remocion de la Interferencia
551 551 for ich in range(num_channel):
552 552 # Se ordena los espectros segun su potencia (menor a mayor)
553 553 power = jspectra[ich, mask_prof, :]
554 554 power = power[:, hei_interf]
555 555 power = power.sum(axis=0)
556 556 psort = power.ravel().argsort()
557 557
558 558 # Se estima la interferencia promedio en los Espectros de Potencia empleando
559 559 junkspc_interf = jspectra[ich, :, hei_interf[psort[list(range(
560 560 offhei_interf, nhei_interf + offhei_interf))]]]
561 561
562 562 if noise_exist:
563 563 # tmp_noise = jnoise[ich] / num_prof
564 564 tmp_noise = jnoise[ich]
565 565 junkspc_interf = junkspc_interf - tmp_noise
566 566 #junkspc_interf[:,comp_mask_prof] = 0
567 567
568 568 jspc_interf = junkspc_interf.sum(axis=0) / nhei_interf
569 569 jspc_interf = jspc_interf.transpose()
570 570 # Calculando el espectro de interferencia promedio
571 571 noiseid = numpy.where(
572 572 jspc_interf <= tmp_noise / numpy.sqrt(num_incoh))
573 573 noiseid = noiseid[0]
574 574 cnoiseid = noiseid.size
575 575 interfid = numpy.where(
576 576 jspc_interf > tmp_noise / numpy.sqrt(num_incoh))
577 577 interfid = interfid[0]
578 578 cinterfid = interfid.size
579 579
580 580 if (cnoiseid > 0):
581 581 jspc_interf[noiseid] = 0
582 582
583 583 # Expandiendo los perfiles a limpiar
584 584 if (cinterfid > 0):
585 585 new_interfid = (
586 586 numpy.r_[interfid - 1, interfid, interfid + 1] + num_prof) % num_prof
587 587 new_interfid = numpy.asarray(new_interfid)
588 588 new_interfid = {x for x in new_interfid}
589 589 new_interfid = numpy.array(list(new_interfid))
590 590 new_cinterfid = new_interfid.size
591 591 else:
592 592 new_cinterfid = 0
593 593
594 594 for ip in range(new_cinterfid):
595 595 ind = junkspc_interf[:, new_interfid[ip]].ravel().argsort()
596 596 jspc_interf[new_interfid[ip]
597 597 ] = junkspc_interf[ind[nhei_interf // 2], new_interfid[ip]]
598 598
599 599 jspectra[ich, :, ind_hei] = jspectra[ich, :,
600 600 ind_hei] - jspc_interf # Corregir indices
601 601
602 602 # Removiendo la interferencia del punto de mayor interferencia
603 603 ListAux = jspc_interf[mask_prof].tolist()
604 604 maxid = ListAux.index(max(ListAux))
605 605
606 606 if cinterfid > 0:
607 607 for ip in range(cinterfid * (interf == 2) - 1):
608 608 ind = (jspectra[ich, interfid[ip], :] < tmp_noise *
609 609 (1 + 1 / numpy.sqrt(num_incoh))).nonzero()
610 610 cind = len(ind)
611 611
612 612 if (cind > 0):
613 613 jspectra[ich, interfid[ip], ind] = tmp_noise * \
614 614 (1 + (numpy.random.uniform(cind) - 0.5) /
615 615 numpy.sqrt(num_incoh))
616 616
617 617 ind = numpy.array([-2, -1, 1, 2])
618 618 xx = numpy.zeros([4, 4])
619 619
620 620 for id1 in range(4):
621 621 xx[:, id1] = ind[id1]**numpy.asarray(list(range(4)))
622 622
623 623 xx_inv = numpy.linalg.inv(xx)
624 624 xx = xx_inv[:, 0]
625 625 ind = (ind + maxid + num_mask_prof) % num_mask_prof
626 626 yy = jspectra[ich, mask_prof[ind], :]
627 627 jspectra[ich, mask_prof[maxid], :] = numpy.dot(
628 628 yy.transpose(), xx)
629 629
630 630 indAux = (jspectra[ich, :, :] < tmp_noise *
631 631 (1 - 1 / numpy.sqrt(num_incoh))).nonzero()
632 632 jspectra[ich, indAux[0], indAux[1]] = tmp_noise * \
633 633 (1 - 1 / numpy.sqrt(num_incoh))
634 634
635 635 # Remocion de Interferencia en el Cross Spectra
636 636 if jcspectra is None:
637 637 return jspectra, jcspectra
638 638 num_pairs = int(jcspectra.size / (num_prof * num_hei))
639 639 jcspectra = jcspectra.reshape(num_pairs, num_prof, num_hei)
640 640
641 641 for ip in range(num_pairs):
642 642
643 643 #-------------------------------------------
644 644
645 645 cspower = numpy.abs(jcspectra[ip, mask_prof, :])
646 646 cspower = cspower[:, hei_interf]
647 647 cspower = cspower.sum(axis=0)
648 648
649 649 cspsort = cspower.ravel().argsort()
650 650 junkcspc_interf = jcspectra[ip, :, hei_interf[cspsort[list(range(
651 651 offhei_interf, nhei_interf + offhei_interf))]]]
652 652 junkcspc_interf = junkcspc_interf.transpose()
653 653 jcspc_interf = junkcspc_interf.sum(axis=1) / nhei_interf
654 654
655 655 ind = numpy.abs(jcspc_interf[mask_prof]).ravel().argsort()
656 656
657 657 median_real = int(numpy.median(numpy.real(
658 658 junkcspc_interf[mask_prof[ind[list(range(3 * num_prof // 4))]], :])))
659 659 median_imag = int(numpy.median(numpy.imag(
660 660 junkcspc_interf[mask_prof[ind[list(range(3 * num_prof // 4))]], :])))
661 661 comp_mask_prof = [int(e) for e in comp_mask_prof]
662 662 junkcspc_interf[comp_mask_prof, :] = numpy.complex(
663 663 median_real, median_imag)
664 664
665 665 for iprof in range(num_prof):
666 666 ind = numpy.abs(junkcspc_interf[iprof, :]).ravel().argsort()
667 667 jcspc_interf[iprof] = junkcspc_interf[iprof, ind[nhei_interf // 2]]
668 668
669 669 # Removiendo la Interferencia
670 670 jcspectra[ip, :, ind_hei] = jcspectra[ip,
671 671 :, ind_hei] - jcspc_interf
672 672
673 673 ListAux = numpy.abs(jcspc_interf[mask_prof]).tolist()
674 674 maxid = ListAux.index(max(ListAux))
675 675
676 676 ind = numpy.array([-2, -1, 1, 2])
677 677 xx = numpy.zeros([4, 4])
678 678
679 679 for id1 in range(4):
680 680 xx[:, id1] = ind[id1]**numpy.asarray(list(range(4)))
681 681
682 682 xx_inv = numpy.linalg.inv(xx)
683 683 xx = xx_inv[:, 0]
684 684
685 685 ind = (ind + maxid + num_mask_prof) % num_mask_prof
686 686 yy = jcspectra[ip, mask_prof[ind], :]
687 687 jcspectra[ip, mask_prof[maxid], :] = numpy.dot(yy.transpose(), xx)
688 688
689 689 # Guardar Resultados
690 690 self.dataOut.data_spc = jspectra
691 691 self.dataOut.data_cspc = jcspectra
692 692
693 693 return 1
694 694
695 695 def run(self, dataOut, interf = 2,hei_interf = None, nhei_interf = None, offhei_interf = None, mode=1):
696 696
697 697 self.dataOut = dataOut
698 698
699 699 if mode == 1:
700 700 self.removeInterference(interf = 2,hei_interf = None, nhei_interf = None, offhei_interf = None)
701 701 elif mode == 2:
702 702 self.removeInterference2()
703 703
704 704 return self.dataOut
705 705
706 706
707 707 class IncohInt(Operation):
708 708
709 709 __profIndex = 0
710 710 __withOverapping = False
711 711
712 712 __byTime = False
713 713 __initime = None
714 714 __lastdatatime = None
715 715 __integrationtime = None
716 716
717 717 __buffer_spc = None
718 718 __buffer_cspc = None
719 719 __buffer_dc = None
720 720
721 721 __dataReady = False
722 722
723 723 __timeInterval = None
724 724
725 725 n = None
726 726
727 727 def __init__(self):
728 728
729 729 Operation.__init__(self)
730 730
731 731 def setup(self, n=None, timeInterval=None, overlapping=False):
732 732 """
733 733 Set the parameters of the integration class.
734 734
735 735 Inputs:
736 736
737 737 n : Number of coherent integrations
738 738 timeInterval : Time of integration. If the parameter "n" is selected this one does not work
739 739 overlapping :
740 740
741 741 """
742 742
743 743 self.__initime = None
744 744 self.__lastdatatime = 0
745 745
746 746 self.__buffer_spc = 0
747 747 self.__buffer_cspc = 0
748 748 self.__buffer_dc = 0
749 749
750 750 self.__profIndex = 0
751 751 self.__dataReady = False
752 752 self.__byTime = False
753 753
754 754 if n is None and timeInterval is None:
755 755 raise ValueError("n or timeInterval should be specified ...")
756 756
757 757 if n is not None:
758 758 self.n = int(n)
759 759 else:
760 760
761 761 self.__integrationtime = int(timeInterval)
762 762 self.n = None
763 763 self.__byTime = True
764 764
765 765 def putData(self, data_spc, data_cspc, data_dc):
766 766 """
767 767 Add a profile to the __buffer_spc and increase in one the __profileIndex
768 768
769 769 """
770 770
771 771 self.__buffer_spc += data_spc
772 772
773 773 if data_cspc is None:
774 774 self.__buffer_cspc = None
775 775 else:
776 776 self.__buffer_cspc += data_cspc
777 777
778 778 if data_dc is None:
779 779 self.__buffer_dc = None
780 780 else:
781 781 self.__buffer_dc += data_dc
782 782
783 783 self.__profIndex += 1
784 784
785 785 return
786 786
787 787 def pushData(self):
788 788 """
789 789 Return the sum of the last profiles and the profiles used in the sum.
790 790
791 791 Affected:
792 792
793 793 self.__profileIndex
794 794
795 795 """
796 796
797 797 data_spc = self.__buffer_spc
798 798 data_cspc = self.__buffer_cspc
799 799 data_dc = self.__buffer_dc
800 800 n = self.__profIndex
801 801
802 802 self.__buffer_spc = 0
803 803 self.__buffer_cspc = 0
804 804 self.__buffer_dc = 0
805 805 self.__profIndex = 0
806 806
807 807 return data_spc, data_cspc, data_dc, n
808 808
809 809 def byProfiles(self, *args):
810 810
811 811 self.__dataReady = False
812 812 avgdata_spc = None
813 813 avgdata_cspc = None
814 814 avgdata_dc = None
815 815
816 816 self.putData(*args)
817 817
818 818 if self.__profIndex == self.n:
819 819
820 820 avgdata_spc, avgdata_cspc, avgdata_dc, n = self.pushData()
821 821 self.n = n
822 822 self.__dataReady = True
823 823
824 824 return avgdata_spc, avgdata_cspc, avgdata_dc
825 825
826 826 def byTime(self, datatime, *args):
827 827
828 828 self.__dataReady = False
829 829 avgdata_spc = None
830 830 avgdata_cspc = None
831 831 avgdata_dc = None
832 832
833 833 self.putData(*args)
834 834
835 835 if (datatime - self.__initime) >= self.__integrationtime:
836 836 avgdata_spc, avgdata_cspc, avgdata_dc, n = self.pushData()
837 837 self.n = n
838 838 self.__dataReady = True
839 839
840 840 return avgdata_spc, avgdata_cspc, avgdata_dc
841 841
842 842 def integrate(self, datatime, *args):
843 843
844 844 if self.__profIndex == 0:
845 845 self.__initime = datatime
846 846
847 847 if self.__byTime:
848 848 avgdata_spc, avgdata_cspc, avgdata_dc = self.byTime(
849 849 datatime, *args)
850 850 else:
851 851 avgdata_spc, avgdata_cspc, avgdata_dc = self.byProfiles(*args)
852 852
853 853 if not self.__dataReady:
854 854 return None, None, None, None
855 855
856 856 return self.__initime, avgdata_spc, avgdata_cspc, avgdata_dc
857 857
858 858 def run(self, dataOut, n=None, timeInterval=None, overlapping=False):
859 859 if n == 1:
860 860 return dataOut
861 861
862 862 dataOut.flagNoData = True
863 863
864 864 if not self.isConfig:
865 865 self.setup(n, timeInterval, overlapping)
866 866 self.isConfig = True
867 867
868 868 avgdatatime, avgdata_spc, avgdata_cspc, avgdata_dc = self.integrate(dataOut.utctime,
869 869 dataOut.data_spc,
870 870 dataOut.data_cspc,
871 871 dataOut.data_dc)
872 872
873 873 if self.__dataReady:
874 874
875 875 dataOut.data_spc = avgdata_spc
876 876 dataOut.data_cspc = avgdata_cspc
877 877 dataOut.data_dc = avgdata_dc
878 878 dataOut.nIncohInt *= self.n
879 879 dataOut.utctime = avgdatatime
880 880 dataOut.flagNoData = False
881 881
882 882 return dataOut No newline at end of file
General Comments 0
You need to be logged in to leave comments. Login now