Effects.py 26.9 KB
Newer Older
Fang Yuedong's avatar
Fang Yuedong committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
import galsim
from matplotlib.pyplot import flag
import numpy as np
from numpy.core.fromnumeric import mean, size
from numpy.random import Generator, PCG64
import math
from numba import jit


def AddOverscan(GSImage, overscan=1000, gain=1, widthl=27, widthr=27, widtht=8, widthb=8, read_noise=5):
    """
    Add overscan/gain; gain=e-/ADU
    widthl: left pre-scan width
    widthr: right pre-scan width
    widtht: top over-scan width (the top of nd-array with small row-index)
    widthb: bottom over-scan width (the bottom of nd-array with large row-index)
    """
    imgshape = GSImage.array.shape
    newimg = galsim.Image(imgshape[1]+widthl+widthr, imgshape[0]+widtht+widthb, init_value=0)
    rng = galsim.UniformDeviate()
    NoiseOS = galsim.GaussianNoise(rng, sigma=read_noise)
    newimg.addNoise(NoiseOS)
    newimg = (newimg+overscan)/gain
    newimg.array[widtht:(widtht+imgshape[0]),widthl:(widthl+imgshape[1])] = GSImage.array
    newimg.wcs = GSImage.wcs
    # if GSImage.wcs is not None:
    #     newwcs = GSImage.wcs.withOrigin(galsim.PositionD(widthl,widtht))
    #     newimg.wcs = newwcs
    # else:
    #     pass
    return newimg


def DefectivePixels(GSImage, IfHotPix=True, IfDeadPix=True, fraction=1E-4, seed=20210304, biaslevel=500):
    # Also called bad pixels, including hot pixels and dead pixels
    # Hot Pixel > 20e-/s
    # Dead Pixel < 70%*Mean
    rgf = Generator(PCG64(int(seed*1.1)))
    if IfHotPix==True and IfDeadPix==True:
        HotFraction = rgf.random()             # fraction in total bad pixels
    elif IfHotPix==False and IfDeadPix==False:
        return GSImage
    elif IfHotPix==True:
        HotFraction = 1
    else:
        HotFraction = 0

    NPix = GSImage.array.size
    NPixBad = int(NPix*fraction)
    NPixHot = int(NPix*fraction*HotFraction)
    NPixDead = NPixBad-NPixHot
    
    NPix_y,NPix_x = GSImage.array.shape
    mean = np.mean(GSImage.array)
    rgp = Generator(PCG64(int(seed)))
    yxposfrac = rgp.random((NPixBad,2))
    YPositHot = np.array(NPix_y*yxposfrac[0:NPixHot,0]).astype(np.int32)
    XPositHot = np.array(NPix_x*yxposfrac[0:NPixHot,1]).astype(np.int32)
    YPositDead = np.array(NPix_y*yxposfrac[NPixHot:,0]).astype(np.int32)
    XPositDead = np.array(NPix_x*yxposfrac[NPixHot:,1]).astype(np.int32)

    rgh = Generator(PCG64(int(seed*1.2)))
    rgd = Generator(PCG64(int(seed*1.3)))
    if IfHotPix==True:
        GSImage.array[YPositHot,XPositHot] += rgh.gamma(2,25*150,size=NPixHot)
    if IfDeadPix==True:
        GSImage.array[YPositDead,XPositDead] = rgd.random(NPixDead)*(mean-biaslevel)*0.7+biaslevel+rgp.standard_normal()*5
    return GSImage


def BadColumns(GSImage, seed=20210309, chipid=1):
    # Set bad column values
    ysize,xsize = GSImage.array.shape
    subarr = GSImage.array[int(ysize*0.1):int(ysize*0.12), int(xsize*0.1):int(xsize*0.12)]
    meanimg = np.median(subarr)
    stdimg = np.std(subarr)
    seed += chipid
    rgn = Generator(PCG64(int(seed)))
    rgcollen = Generator(PCG64(int(seed*1.1)))
    rgxpos = Generator(PCG64(int(seed*1.2)))
    rgdn = Generator(PCG64(int(seed*1.3)))

    nbadsecA,nbadsecD = rgn.integers(low=1, high=5, size=2) 
    collen = rgcollen.integers(low=int(ysize*0.1), high=int(ysize*0.7), size=(nbadsecA+nbadsecD)) 
    xposit = rgxpos.integers(low=int(xsize*0.05), high=int(xsize*0.95), size=(nbadsecA+nbadsecD))
    signs = 2*rgdn.integers(0,2,size=(nbadsecA+nbadsecD))-1
    dn = rgdn.integers(low=meanimg*0.4, high=meanimg*0.8, size=(nbadsecA+nbadsecD))*signs
    for badcoli in range(nbadsecA):
        GSImage.array[(ysize-collen[badcoli]):ysize,xposit[badcoli]:(xposit[badcoli]+1)] += np.random.random((collen[badcoli],1))*stdimg*3+dn[badcoli]
    for badcoli in range(nbadsecD):
        GSImage.array[0:collen[badcoli+nbadsecA],xposit[badcoli+nbadsecA]:(xposit[badcoli+nbadsecA]+1)] += np.random.random((collen[badcoli+nbadsecA],1))*stdimg*3+dn[badcoli+nbadsecA]
    return GSImage


def AddBiasNonUniform16(GSImage, bias_level = 500, nsecy = 2, nsecx=8, seed=202102):
    # Generate Bias and its non-uniformity, and add the 16 bias values to the GS-Image
    rg = Generator(PCG64(int(seed)))
    Random16 = (rg.random(nsecy*nsecx)-0.5)*20
    if int(bias_level)==0:
        BiasLevel = np.zeros((nsecy,nsecx))
    elif bias_level>0:
        BiasLevel = Random16.reshape((nsecy,nsecx)) + bias_level
    arrshape = GSImage.array.shape
    secsize_x = int(arrshape[1]/nsecx)
    secsize_y = int(arrshape[0]/nsecy)
    for rowi in range(nsecy):
        for coli in range(nsecx):
            GSImage.array[rowi*secsize_y:(rowi+1)*secsize_y,coli*secsize_x:(coli+1)*secsize_x] += BiasLevel[rowi,coli]
    return GSImage


def MakeBiasNcomb(npix_x, npix_y, bias_level=500, ncombine=1, read_noise=5, gain=1, seed=202102):
    # Start with 0 value bias GS-Image
    ncombine=int(ncombine)
    BiasSngImg0 = galsim.Image(npix_x, npix_y, init_value=0)
    BiasSngImg = AddBiasNonUniform16(BiasSngImg0, 
                bias_level=bias_level, 
                nsecy = 2, nsecx=8, 
                seed=int(seed))
    BiasCombImg = BiasSngImg*ncombine
    rng = galsim.UniformDeviate()
    NoiseBias = galsim.GaussianNoise(rng=rng, sigma=read_noise*ncombine**0.5)
    BiasCombImg.addNoise(NoiseBias)
    if ncombine == 1:
        BiasTag = 'Single'
        pass
    elif ncombine >1:
        BiasCombImg /= ncombine
        BiasTag = 'Combine'
    # BiasCombImg.replaceNegative(replace_value=0)
    # BiasCombImg.quantize()
    return BiasCombImg, BiasTag


def ApplyGainNonUniform16(GSImage, gain=1, nsecy = 2, nsecx=8, seed=202102):
    # Generate Gain non-uniformity, and multipy the different factors (mean~1 with sigma~1%) to the GS-Image
    rg = Generator(PCG64(int(seed)))
    Random16 = (rg.random(nsecy*nsecx)-0.5)*0.04+1   # sigma~1%
    Gain16 = Random16.reshape((nsecy,nsecx))/gain
    print("Gain of 16 channels: ",Gain16)
    arrshape = GSImage.array.shape
    secsize_x = int(arrshape[1]/nsecx)
    secsize_y = int(arrshape[0]/nsecy)
    for rowi in range(nsecy):
        for coli in range(nsecx):
            GSImage.array[rowi*secsize_y:(rowi+1)*secsize_y,coli*secsize_x:(coli+1)*secsize_x] *= Gain16[rowi,coli]
    return GSImage


def MakeFlatSmooth(GSBounds, seed):
    rg = Generator(PCG64(int(seed)))
    r1,r2,r3,r4 = rg.random(4)
    a1 = -0.5 + 0.2*r1
    a2 = -0.5 + 0.2*r2
    a3 = r3+5
    a4 = r4+5
    xmin,xmax,ymin,ymax = GSBounds.getXMin(), GSBounds.getXMax(), GSBounds.getYMin(), GSBounds.getYMax()
    Flty, Fltx = np.mgrid[ymin:(ymax+1), xmin:(xmax+1)]
    rg = Generator(PCG64(int(seed)))
    p1,p2,bg=rg.poisson(1000, 3)
    Fltz = 1e-6*(a1 * (Fltx-p1) ** 2 + a2 * (Flty-p2) ** 2 - a3*Fltx - a4*Flty) + bg*20
    FlatImg = galsim.ImageF(Fltz)
    return FlatImg


def MakeFlatNcomb(flat_single_image, ncombine=1, read_noise=5, gain=1, overscan=500, biaslevel=500, seed_bias=20210311):
    ncombine=int(ncombine)
    FlatCombImg = flat_single_image*ncombine
    rng = galsim.UniformDeviate()
    NoiseFlatPoi = galsim.PoissonNoise(rng=rng, sky_level=0)
    FlatCombImg.addNoise(NoiseFlatPoi)
    NoiseFlatReadN = galsim.GaussianNoise(rng=rng, sigma=read_noise*ncombine**0.5)
    FlatCombImg.addNoise(NoiseFlatReadN)
    # NoiseFlat = galsim.CCDNoise(rng, gain=gain, read_noise=read_noise*ncombine**0.5, sky_level=0)
    for i in range(ncombine):
        FlatCombImg = AddBiasNonUniform16(
            FlatCombImg, 
            bias_level=biaslevel, 
            nsecy=2, nsecx=8, 
            seed=seed_bias)
    if ncombine == 1:
        FlatTag = 'Single'
        pass
    elif ncombine >1:
        FlatCombImg /= ncombine
        FlatTag = 'Combine'
    # FlatCombImg.replaceNegative(replace_value=0)
    # FlatCombImg.quantize()
    return FlatCombImg, FlatTag


def MakeDarkNcomb(npix_x, npix_y, overscan=500, bias_level=500, seed_bias=202102, darkpsec=0.02, exptime=150, ncombine=10, read_noise=5, gain=1):
    ncombine=int(ncombine)
    darkpix = darkpsec*exptime
    DarkSngImg = galsim.Image(npix_x, npix_y, init_value=darkpix)
    rng = galsim.UniformDeviate()
    NoiseDarkPoi = galsim.PoissonNoise(rng=rng, sky_level=0)
    NoiseReadN = galsim.GaussianNoise(rng=rng, sigma=read_noise*ncombine**0.5)
    DarkCombImg = DarkSngImg*ncombine
    DarkCombImg.addNoise(NoiseDarkPoi)
    DarkCombImg.addNoise(NoiseReadN)
    for i in range(ncombine):
        DarkCombImg = AddBiasNonUniform16(
            DarkCombImg, 
            bias_level=bias_level, 
            nsecy = 2, nsecx=8, 
            seed=int(seed_bias))
    if ncombine == 1:
        DarkTag = 'Single'
        pass
    elif ncombine >1:
        DarkCombImg /= ncombine
        DarkTag = 'Combine'
    # DarkCombImg.replaceNegative(replace_value=0)
    # DarkCombImg.quantize()
    return DarkCombImg, DarkTag


def PRNU_Img(xsize, ysize, sigma=0.01, seed=202101):
    rg = Generator(PCG64(int(seed)))
    prnuarr = rg.normal(1, sigma, (ysize,xsize))
    prnuimg = galsim.ImageF(prnuarr)
    return prnuimg


def NonLinearity(GSImage, beta1=1E-7, beta2=1E-10):
    NonLinear_f = lambda x, beta_1, beta_2: x - beta_1*x*x + beta_2*x*x*x
    GSImage.applyNonlinearity(NonLinear_f, beta1, beta2)
    return GSImage


def chargeflow(ndarr, fullwell=10E4):
    size_y,size_x = ndarr.shape
    satpos_y = np.where(ndarr>=fullwell)[0]
    satpos_x = np.where(ndarr>=fullwell)[1]
    Nsatpix = len(satpos_y)

    if Nsatpix==0:
        # make no change for the image array
        return ndarr

    else:    
        for i in range(Nsatpix):
            ClumpFullwell=True
            satcharge = ndarr[satpos_y[i],satpos_x[i]]-fullwell
            ndarr[satpos_y[i],satpos_x[i]] = fullwell
            
            satpos_yi0 = satpos_y[i]

            # Define the x,y=0,0 element of image array is the lower-left corner. So y decreases being downward.

            # print('Charge Clump moves down')
            chargedn = ((np.random.random()-0.5)*0.05+0.5)*satcharge
            chargeup = satcharge - chargedn
            fwi = 1
            aa = np.log(chargedn/fullwell)**3*0.9   # blooming length begin to has e- less than fullwell
            if aa < 0.05:
                ClumpFullwell=False
            # Test
            try:
                while ClumpFullwell==True:
                    if satpos_y[i]<=0:
                        break
                    if ndarr[satpos_y[i]-1,satpos_x[i]]<fullwell:
                        ndarr[satpos_y[i]-1,satpos_x[i]] = ndarr[satpos_y[i]-1,satpos_x[i]] + chargedn
                        if ndarr[satpos_y[i]-1,satpos_x[i]]>=fullwell:
                            fx = 0.5*(math.exp(math.log(fwi)**3/aa)+np.exp(-1*math.log(fwi)**3/aa))
                            if fx>5:
                                fx=5
                            faa= 0.5*(math.exp(aa/aa)+np.exp(-1*aa/aa))
                            rand_frac = 1-0.1*(fx-1)/(faa-1)

                            chargedn = ndarr[satpos_y[i]-1,satpos_x[i]] - fullwell
                            ndarr[satpos_y[i]-1,satpos_x[i]] = fullwell*rand_frac
                            satpos_y[i] = satpos_y[i]-1
                            if satpos_y[i]<0:
                                ClumpFullwell=False
                                break
                        else:
                            ClumpFullwell=False
                        fwi += 1
                    else:
                        satpos_y[i] = satpos_y[i]-1
                        if satpos_y[i]<0:
                            ClumpFullwell=False
                            break
                  
                
                # print('Charge Clump moves up')
                ClumpFullwell=True
                satpos_y[i] = satpos_yi0
                fwi = 1
                aa = np.log(chargeup/fullwell)**3*0.9   # blooming length at which it begins to have e- less than fullwell
                if aa < 0.05:
                    ClumpFullwell=False
                while ClumpFullwell==True:
                    if satpos_y[i]>=size_y-1:
                        break
                    if ndarr[satpos_y[i]+1,satpos_x[i]]<fullwell:
                        ndarr[satpos_y[i]+1,satpos_x[i]] = ndarr[satpos_y[i]+1,satpos_x[i]] + chargeup
                        if ndarr[satpos_y[i]+1,satpos_x[i]]>=fullwell:
                            fx = 0.5*(math.exp(math.log(fwi)**3/aa)+np.exp(-1*math.log(fwi)**3/aa))
                            if fx>5:
                                fx=5
                            faa= 0.5*(math.exp(aa/aa)+np.exp(-1*aa/aa))
                            rand_frac = 1-0.1*(fx-1)/(faa-1)
                            chargeup = ndarr[satpos_y[i]+1,satpos_x[i]] - fullwell
                            ndarr[satpos_y[i]+1,satpos_x[i]] = fullwell*rand_frac
                            satpos_y[i] = satpos_y[i]+1
                            if satpos_y[i]>=size_y-1:
                                ClumpFullwell=False
                                break
                        else:
                            ClumpFullwell=False
                        fwi += 1
                    else:
                        satpos_y[i] = satpos_y[i]+1
                        if satpos_y[i]>=size_y-1:
                            ClumpFullwell=False
                            break
            except Exception as e:
                print(e)
                print(fwi, aa)
                pass
          
    return ndarr


def SaturBloom(GSImage, nsect_x=1, nsect_y=1, fullwell=10e4):
    """
    To simulate digital detector's saturation and blooming effect. The blooming is along the read-out direction, perpendicular to the charge transfer direction. Charge clumpy overflows the pixel well will flow to two oposite directions with nearly same charges.
    Work together with chargeflow() function.
    Parameters:
      GSImage: a GalSim Image of a chip;
      nsect_x: number of sections divided along x direction;
      nsect_y: number of sections divided along y direction.
    Return: a saturated image array with the same size of input.
    """
    imgarr = GSImage.array
    size_y, size_x = imgarr.shape
    subsize_y = int(size_y/nsect_y)
    subsize_x = int(size_x/nsect_x)

    for i in range(nsect_y):
        for j in range(nsect_x):
            subimg = imgarr[subsize_y*i:subsize_y*(i+1), subsize_x*j:subsize_x*(j+1)]

            subimg = chargeflow(subimg, fullwell=fullwell)

            imgarr[subsize_y*i:subsize_y*(i+1), subsize_x*j:subsize_x*(j+1)] = subimg

    return GSImage


def readout16(GSImage, rowi=0, coli=0, overscan_value=0):
    # readout image as 16 outputs of sub-images plus prescan & overscan.
    # assuming image width and height sizes are both even.
    # assuming image has 2 columns and 8 rows of output channels. 
    # 00  01
    # 10  11
    # 20  21
    # ...
    # return: GS Image Object
    npix_y,npix_x = GSImage.array.shape
    subheight = int(8+npix_y/2+8)
    subwidth = int(16+npix_x/8+27)
    OutputSubimg = galsim.ImageUS(subwidth, subheight, init_value=overscan_value)
    if rowi<4 and coli==0:
        subbounds = galsim.BoundsI(1, int(npix_x/2),  int(npix_y/8*rowi+1), int(npix_y/8*(rowi+1)))
        subbounds = subbounds.shift(galsim.PositionI(GSImage.bounds.getXMin()-1, GSImage.bounds.getYMin()-1))
        subimg = GSImage[subbounds]
        OutputSubimg.array[27:int(npix_y/8)+27,8:int(npix_x/2)+8] = subimg.array
    elif rowi<4 and coli==1:
        subbounds = galsim.BoundsI(npix_x/2+1, npix_x,  npix_y/8*rowi+1, npix_y/8*(rowi+1))
        subbounds = subbounds.shift(galsim.PositionI(GSImage.bounds.getXMin()-1, GSImage.bounds.getYMin()-1))
        subimg = GSImage[subbounds]
        OutputSubimg.array[27:int(npix_y/8)+27,8:int(npix_x/2)+8] = subimg.array
    elif rowi>=4 and rowi<8 and coli==0:
        subbounds = galsim.BoundsI(1, npix_x/2,  npix_y/8*rowi+1, npix_y/8*(rowi+1))
        subbounds = subbounds.shift(galsim.PositionI(GSImage.bounds.getXMin()-1, GSImage.bounds.getYMin()-1))
        subimg = GSImage[subbounds]
        OutputSubimg.array[16:int(npix_y/8)+16,8:int(npix_x/2)+8] = subimg.array
    elif rowi>=4 and rowi<8 and coli==1:
        subbounds = galsim.BoundsI(npix_x/2+1, npix_x,  npix_y/8*rowi+1, npix_y/8*(rowi+1))
        subbounds = subbounds.shift(galsim.PositionI(GSImage.bounds.getXMin()-1, GSImage.bounds.getYMin()-1))
        subimg = GSImage[subbounds]
        OutputSubimg.array[16 :int(npix_y/8)+16,8:int(npix_x/2)+8] = subimg.array
    else:
        print("\n\033[31mError: "+"Wrong rowi or coli assignment. Permitted: 0<=rowi<=7, 0<=coli<=1."+"\033[0m\n")
        return OutputSubimg
    return OutputSubimg


def CTE_Effect(GSImage, threshold=27, direction='column'):
    # Devide the image into 4 sections and apply CTE effect with different trail directions.
    # GSImage: a GalSim Image object.
    size_y,size_x = GSImage.array.shape
    size_sect_y = int(size_y/2)
    size_sect_x = int(size_x/2)
    imgarr = GSImage.array
    if direction == 'column':
        imgarr[0:size_sect_y,:]      = CTEModelColRow(imgarr[0:size_sect_y,:],      trail_direction='down', direction='column', threshold=threshold)
        imgarr[size_sect_y:size_y,:] = CTEModelColRow(imgarr[size_sect_y:size_y,:], trail_direction='up',   direction='column', threshold=threshold)
    elif direction == 'row':
        imgarr[:,0:size_sect_x]      = CTEModelColRow(imgarr[:,0:size_sect_x],      trail_direction='right', direction='row', threshold=threshold)
        imgarr[:,size_sect_x:size_x] = CTEModelColRow(imgarr[:,size_sect_x:size_x], trail_direction='left',  direction='row', threshold=threshold)
    return GSImage


@jit()
def CTEModelColRow(img, trail_direction = 'up', direction='column', threshold=27):

    #total trail flux vs (pixel flux)^1/2 is approximately linear
    #total trail flux = trail_a * (pixel flux)^1/2 + trail_b
    #trail pixel flux = pow(0.5,x)/0.5, normalize to 1
    trail_a =  5.651803799619966
    trail_b = -2.671933068990729

    sh1 = img.shape[0]
    sh2 = img.shape[1]
    n_img = img*0
    idx = np.where(img<threshold)
    if len(idx[0]) == 0:
        pass
    elif len(idx[0])>0:
        n_img[idx] = img[idx]

    yidx,xidx = np.where(img>=threshold)
    if len(yidx) == 0:
        pass
    elif len(yidx)>0:
        # print(index)
        for i, j in zip(yidx,xidx):
            f = img[i,j]
            trail_f = (np.sqrt(f)*trail_a + trail_b)*0.5
            # trail_f=5E-5*f**1.5
            xy_num = 10
            all_trail = np.zeros(xy_num)
            
            xy_upstr = np.arange(1,xy_num,1)

            # all_trail_pix = np.sum(pow(0.5,xy_upstr)/0.5)
            all_trail_pix = 0
            for m in xy_upstr:
                a1=12.97059491  
                b1=0.54286652
                c1=0.69093105
                a2=2.77298856
                b2=0.11231055
                c2=-0.01038675
                # t_pow = 0
                am=1
                bm=1
                t_pow = am*np.exp(-bm*m)
                # if m < 5:
                #     t_pow = a1*np.exp(-b1*m)+c1
                # else:
                #     t_pow = a2*np.exp(-b2*m)+c2
                if t_pow <0:
                    t_pow = 0

                all_trail_pix += t_pow
                all_trail[m] = t_pow


            trail_pix_eff = trail_f/all_trail_pix

            all_trail = trail_pix_eff*all_trail

            all_trail[0] = f - trail_f
            
            for m in np.arange(0,xy_num,1):
                if direction == 'column':
                    if trail_direction == 'down':
                        y_pos = i + m
                    elif trail_direction == 'up':
                        y_pos = i - m
                    if y_pos < 0 or y_pos >=sh1:
                        break
                    n_img[y_pos,j] = n_img[y_pos,j] + all_trail[m]
                elif direction == 'row':
                    if trail_direction == 'left':
                        x_pos = j - m
                    elif trail_direction == 'right':
                        x_pos = j + m
                    if x_pos < 0 or x_pos >=sh2:
                        break
                    n_img[i,x_pos] = n_img[i,x_pos] + all_trail[m]

    return n_img



#---------- For Cosmic-Ray Simulation ------------
#---------- Zhang Xin ----------------------------

def getYValue(collection, x):
    index = 0;
    if (collection.shape[1] == 2):
        while(x>collection[index,0] and index < collection.shape[0]):
            index= index + 1;
        if (index == collection.shape[0] or index == 0):
            return 0;

        deltX = collection[index, 0] - collection[index-1, 0];
        deltY = collection[index, 1] - collection[index-1, 1];

        if deltX == 0:
            return (collection[index, 1] + collection[index-1, 1])/2.0
        else:
            a = deltY/deltX;
            return a * (x - collection[index-1,0]) + collection[index-1, 1];
    return 0;


def selectCosmicRayCollection(attachedSizes, xLen, yLen,cr_pixelRatio,CR_max_size):

    normalRay = 0.90
    nnormalRay = 1-normalRay
    max_nrayLen = 100
    pixelNum = int(xLen *  yLen * cr_pixelRatio * normalRay );
    pixelNum_n = int(xLen *  yLen * cr_pixelRatio * nnormalRay )
    CRPixelNum = 0;
    
    maxValue = max(attachedSizes[:,1])
    maxValue += 0.1;

    cr_event_num = 0;
    CRs = np.zeros(pixelNum);
    while (CRPixelNum < pixelNum):
        x = CR_max_size * np.random.random();
        y = maxValue * np.random.random();
        if (y <= getYValue(attachedSizes, x)):
            CRs[cr_event_num] = np.ceil(x);
            cr_event_num = cr_event_num + 1;
            CRPixelNum = CRPixelNum + round(x);

    while (CRPixelNum < pixelNum + pixelNum_n):
        nx = np.random.random()*(max_nrayLen-CR_max_size)+CR_max_size
        CRs[cr_event_num] = np.ceil(nx);
        cr_event_num = cr_event_num + 1;
        CRPixelNum = CRPixelNum + np.ceil(nx);

    return   CRs[0:cr_event_num];


def defineEnergyForCR(cr_event_size):
    import random
    sigma = 0.6 / 2.355;
    mean = 3.3;

    energys = np.zeros(cr_event_size);
    for i in np.arange(cr_event_size):
        energy_index = random.normalvariate(mean,sigma);
        energys[i] = pow(10, energy_index);

    return energys;

def convCR(CRmap=None, addPSF=None, sp_n = 4):
    sh = CRmap.shape

    # sp_n = 4
    subCRmap = np.zeros(np.array(sh)*sp_n)
    pix_v0 = 1/(sp_n*sp_n)
    for i in np.arange(sh[0]):
        i_st = sp_n*i
        for j in np.arange(sh[1]):
            if CRmap[i,j] ==0:
                continue
            j_st = sp_n*j
            pix_v1 = CRmap[i,j]*pix_v0
            for m in np.arange(sp_n):
                for n in np.arange(sp_n):
                    subCRmap[i_st+m, j_st + n] = pix_v1

    m_size = addPSF.shape[0]

    subCRmap_n = np.zeros(np.array(subCRmap.shape) + m_size -1)

    for i in np.arange(subCRmap.shape[0]):
        for j in np.arange(subCRmap.shape[1]):
            if subCRmap[i,j]>0:
                convPix = addPSF*subCRmap[i,j]
                subCRmap_n[i:i+m_size,j:j+m_size]+=convPix

    CRmap_n = np.zeros((np.array(subCRmap_n.shape)/sp_n).astype(np.int32))
    sh_n = CRmap_n.shape

    for i in np.arange(sh_n[0]):
        i_st = sp_n*i
        for j in np.arange(sh_n[1]):
            p_v = 0
            j_st=sp_n*j
            for m in np.arange(sp_n):
                for n in np.arange(sp_n):
                    p_v += subCRmap_n[i_st+m, j_st + n]

            CRmap_n[i,j] = p_v

    return CRmap_n


def produceCR_Map(xLen, yLen, exTime, cr_pixelRatio, gain, attachedSizes, seed=20210317):
    # Return: an 2-D numpy array
    # attachedSizes = np.loadtxt('./wfc-cr-attachpixel.dat');
    np.random.seed(seed)
    CR_max_size = 20.0;
    cr_size = selectCosmicRayCollection(attachedSizes, xLen, yLen, cr_pixelRatio, CR_max_size);

    cr_event_size = cr_size.shape[0];
    cr_energys = defineEnergyForCR(cr_event_size);

    CRmap = np.zeros([yLen, xLen]);

    ## produce conv kernel
    from astropy.modeling.models import Gaussian2D
    o_size = 4
    sp_n = 8

    m_size = o_size*sp_n+1
    m_cen = o_size*sp_n/2
    sigma_psf = 0.2*sp_n
    addPSF_ = Gaussian2D(1, m_cen, m_cen, sigma_psf, sigma_psf)
    yp, xp = np.mgrid[0:m_size, 0:m_size]
    addPSF = addPSF_(xp, yp)
    convKernel = addPSF/addPSF.sum()

    #################################


    for i in np.arange(cr_event_size):
        xPos = round((xLen - 1)* np.random.random());
        yPos = round((yLen - 1)* np.random.random());
        cr_lens = int(cr_size[i]);
        if cr_lens ==0:
            continue;
        pix_energy = cr_energys[i]/gain/cr_lens;
        pos_angle = 1/2*math.pi*np.random.random();

        crMatrix = np.zeros([cr_lens+1, cr_lens + 1])

        for j in np.arange(cr_lens):
            x_n = int(np.cos(pos_angle)*j - np.sin(pos_angle)*0);
            if x_n < 0:
                x_n = x_n + cr_lens+1
            y_n = int(np.sin(pos_angle)*j + np.cos(pos_angle)*0);
            if x_n<0 or x_n >cr_lens or y_n < 0 or y_n > cr_lens:
                continue;
            crMatrix[y_n,x_n] = pix_energy;

        crMatrix_n = convCR(crMatrix, convKernel, sp_n)
        # crMatrix_n = crMatrix

        xpix = np.arange(crMatrix_n.shape[0]) + int(xPos)
        ypix = np.arange(crMatrix_n.shape[1]) + int(yPos)

        sh = CRmap.shape
        okx = (xpix >= 0) & (xpix < sh[1])
        oky = (ypix >= 0) & (ypix < sh[0])

        sly = slice(ypix[oky].min(), ypix[oky].max()+1)
        slx = slice(xpix[okx].min(), xpix[okx].max()+1)
        CRmap[sly, slx] += crMatrix_n[oky,:][:,okx]


    return CRmap.astype(np.int32);


def ShutterEffectArr(GSImage, t_shutter=1.3, dist_bearing=735, dt=1E-3):
    # Generate Shutter-Effect normalized image
    # t_shutter: time of shutter movement
    # dist_bearing: distance between two bearings of shutter leaves
    # dt: delta_t of sampling

    from scipy import interpolate

    SampleNumb = int(t_shutter/dt+1)
    DistHalf = dist_bearing/2

    t = np.arange(SampleNumb)*dt
    a_arr = 5.84*np.sin(2*math.pi/t_shutter*t)
    v = np.zeros(SampleNumb)
    theta = np.zeros(SampleNumb)
    x = np.arange(SampleNumb)/(SampleNumb-1)*dist_bearing
    s = np.zeros(SampleNumb)
    s1 = np.zeros(SampleNumb)
    s2 = np.zeros(SampleNumb)
    brt  = np.zeros(SampleNumb)
    idx = np.arange(SampleNumb)
    sidx = np.zeros(SampleNumb)
    s1idx = np.zeros(SampleNumb)
    s2idx = np.zeros(SampleNumb)

    v[0] = 0
    theta[0] = 0

    for i in range(SampleNumb-1):
        v[i+1] = v[i]+a_arr[i]*dt
        theta[i+1] = theta[i]+v[i]*dt
        s1[i] = DistHalf*np.cos(theta[i])
        s2[i] = dist_bearing-DistHalf*np.cos(theta[i])
        s1idx[i] = int(s1[i]/dist_bearing*(SampleNumb))
        s2idx[i] = int(s2[i]/dist_bearing*(SampleNumb))
        brt[(idx>s1idx[i]) & (idx<s2idx[i])] += dt

    brt = brt*2+(150-t_shutter*2)

    x = (x-dist_bearing/2)*100

    intp = interpolate.splrep(x, brt, s=0)

    xmin = GSImage.bounds.getXMin()
    xmax = GSImage.bounds.getXMax()
    ymin = GSImage.bounds.getYMin()
    ymax = GSImage.bounds.getYMax()
    if xmin<np.min(x) or xmax>np.max(x):
        raise LookupError("Out of focal-plane bounds in X-direction.")
    if ymin<-25331 or ymax>25331:
        raise LookupError("Out of focal-plane bounds in Y-direction.")
    sizex = xmax-xmin+1
    sizey = ymax-ymin+1
    xnewgrid = np.mgrid[xmin:(xmin+sizex)]
    expeffect = interpolate.splev(xnewgrid, intp, der=0)
    expeffect /= np.max(expeffect)
    exparrnormal = np.tile(expeffect, (sizey,1))
    # GSImage *= exparrnormal

    return exparrnormal