ndsets = np.int(sys.argv[2])
nth = np.int(sys.argv[3])
start = np.int(sys.argv[4])
name = sys.argv[5]
binning = 0
data = np.zeros([nsets*ndsets*nth, (2048-512)//pow(2, binning),
(2448-400)//pow(2, binning)], dtype='float32')
theta = np.zeros(nsets*ndsets*nth, dtype='float32')
strs = ['098','099','100']
for j in range(nsets):
name0 = name[:-3]+strs[j]#name[:-2]+str(np.int(name[-2:])+j)
print(name0)
for k in range(ndsets):
print(j,k)
idstart = j*ndsets*nth+k*nth
data[idstart:idstart+nth] = np.load(name0+'ti_bin'+str(binning)+str(k)+'.npy')[:,256:-256,200:-200].astype('float32')
theta[idstart:idstart+nth] = np.load(name0+'_theta'+str(k)+'.npy').astype('float32')
data[np.isnan(data)] = 0
data = np.ascontiguousarray(data[start::2])
theta = np.ascontiguousarray(theta[start::2])
ngpus = 4
pnz = 8
nitercg = 32
center = 1256-200
res = tomoalign.cg(data, theta, pnz, center, ngpus, nitercg, padding=True)
name+='/'+str(len(theta))
dxchange.write_tiff_stack(res['u'], name+'/results_cg'+str(start)+'/u/r', overwrite=True)
2448//pow(2, binning)], dtype='float32')
theta = np.zeros(ndsets*nth, dtype='float32')
for k in range(ndsets):
data[k*nth:(k+1)*nth] = np.load(fname+'_bin' +
str(binning)+str(k)+'.npy').astype('float32')
theta[k*nth:(k+1)*nth] = np.load(fname+'_theta' +
str(k)+'.npy').astype('float32')
#data = data[:,512-8:512+8]
data[np.isnan(data)] = 0
data -= np.mean(data)
ngpus = 4
pnz = 4
ptheta = 10
niteradmm = [96, 48, 24] # number of iterations in the ADMM scheme
# starting window size in optical flow estimation
startwin = [256, 128, 64]
# step for decreasing the window size in optical flow estimtion
stepwin = [2, 2, 2]
center = (1202)/pow(2, binning)
fname += '/revision_cg_nop'+'_'+str(binning)
data = np.ascontiguousarray(data.astype('float32'))
theta = np.ascontiguousarray(theta.astype('float32'))
#dxchange.write_tiff_stack(
#data, fname+'/data/d', overwrite=True)
pprot = 100
res = tomoalign.cg(data, theta, pprot, pnz, center, ngpus, 64)
dxchange.write_tiff_stack(res['u'], fname+'/results/u/r', overwrite=True)
ndsets = np.int(sys.argv[1])
nth = np.int(sys.argv[2])
fname = sys.argv[3]
binning = 1
data = np.zeros(
[ndsets * nth, 2048 // pow(2, binning), 2448 // pow(2, binning)],
dtype='float32')
theta = np.zeros(ndsets * nth, dtype='float32')
for k in range(ndsets):
data[k * nth:(k + 1) *
nth] = np.load(fname + '_bin' + str(binning) + str(k) +
'.npy').astype('float32')
theta[k * nth:(k + 1) * nth] = np.load(fname + '_theta' + str(k) +
'.npy').astype('float32')
data[np.isnan(data)] = 0
# data -= np.mean(data)
ngpus = 4
nitercg = 32
pnz = 8
center = centers[fname] / pow(2, binning)
data = np.ascontiguousarray(data)
theta = np.ascontiguousarray(theta)
res = tomoalign.cg(data, theta, pnz, center, ngpus, nitercg)
dxchange.write_tiff_stack(res['u'],
fname + '/results_cg/u/r',
overwrite=True)
binning = 1
data = np.zeros([ndsets * nth, 320, 544], dtype='float32')
theta = np.zeros(ndsets * nth, dtype='float32')
for k in range(ndsets):
data[k * nth:(k + 1) *
nth] = np.load(fname + '_bin' + str(binning) + str(k) +
'.npy').astype('float32')
theta[k * nth:(k + 1) * nth] = np.load(fname + '_theta' + str(k) +
'.npy').astype('float32')
data[np.isnan(data)] = 0
#data-=np.mean(data)
ngpus = 4
pprot = 190
nitercg = 64
pnz = 80
center = 272
res = tomoalign.cg(data,
theta,
pprot,
pnz,
center,
ngpus,
nitercg,
padding=False)
dxchange.write_tiff_stack(res['u'],
fname + '/results_cg/u/r',
overwrite=True)