def getSVDdata(ops):
mov, max_proj = sparsedetect.get_mov(ops)
ops['max_proj'] = max_proj
nbins, Lyc, Lxc = np.shape(mov)
sig = ops['diameter']/10. # PICK UP
for j in range(nbins):
mov[j,:,:] = gaussian_filter(mov[j,:,:], sig)
# compute noise variance across frames
sdmov = sparsedetect.get_sdmov(mov, ops)
mov /= sdmov
mov = np.reshape(mov, (-1,Lyc*Lxc))
if 1:
# compute covariance of binned frames
cov = mov @ mov.transpose() / mov.shape[1]
cov = cov.astype('float32')
nsvd_for_roi = min(ops['nbinned'], int(cov.shape[0]/2))
u, s, v = np.linalg.svd(cov)
u = u[:, :nsvd_for_roi]
U = u.transpose() @ mov
else:
U = mov
u = []
U = np.reshape(U, (-1,Lyc,Lxc))
U = np.transpose(U, (1, 2, 0)).copy()
return ops, U, sdmov, u
def getSVDproj(ops, u):
mov, _ = sparsedetect.get_mov(ops)
nbins, Lyc, Lxc = np.shape(mov)
if ('smooth_masks' in ops) and ops['smooth_masks']:
sig = np.maximum([.5, .5], ops['diameter']/20.)
for j in range(nbins):
mov[j,:,:] = ndimage.gaussian_filter(mov[j,:,:], sig)
if 1:
#sdmov = np.ones((Lyc*Lxc,), 'float32')
sdmov = sparsedetect.get_sdmov(mov, ops)
mov/=sdmov
mov = np.reshape(mov, (-1,Lyc*Lxc))
U = u.transpose() @ mov
U = U.transpose().copy().reshape((Lyc,Lxc,-1))
else:
U = np.transpose(mov, (1, 2, 0)).copy()
return U, sdmov