#%% compute subregions where to apply more efficiently facrtorization algorithms
fovs, mcoef, distanceMatrix=m.partition_FOV_KMeans(tradeoff_weight=.7,fx=.25,fy=.25,n_clusters=4,max_iter=500);
plt.imshow(fovs)
#%% create a denoised version of the movie, nice to visualize
if True:
m2=m.copy()
m2.IPCA_denoise(components = 100, batch = 1000)
m2.playMovie(frate=.05,magnification=4,gain=10.0)
#%%
#%% compute spatial components via NMF
initTime=time.time()
space_spcomps,time_comps=m.NonnegativeMatrixFactorization(n_components=20,beta=1,tol=5e-7);
print 'elapsed time:' + str(time.time()-initTime)
matrixMontage(np.asarray(space_spcomps),cmap=plt.cm.gray) # visualize components
#%% compute spatial components via ICA PCA
initTime=time.time()
spcomps=m.IPCA_stICA(components=10,mu=.5);
print 'elapsed time:' + str(time.time()-initTime)
matrixMontage(spcomps,cmap=plt.cm.gray) # visualize components
#%% extract ROIs from spatial components
#_masks,masks_grouped=m.extractROIsFromPCAICA(spcomps, numSTD=6, gaussiansigmax=2 , gaussiansigmay=2)
_masks,_=m.extractROIsFromPCAICA(spcomps, numSTD=10.0, gaussiansigmax=1 , gaussiansigmay=1)
matrixMontage(np.asarray(_masks),cmap=plt.cm.gray)
#%% extract single ROIs from each mask
