pixel_offset_stddev=0.05)
# Make a noise injector producing 2% gaussian additive noise
noise_injector = vlab.noise_injector("gaussian", sigma=.02)
# Make an synthetic image generation pipeline
image_generator = vlab.SyntheticImageGenerator(speckle_image=speckle_image,
image_deformer=image_deformer,
downsampler=downsampler,
noise_injector=noise_injector,
n=n_imgs)
# Put it into an image stack
image_stack = dic.ImageStack(image_generator)
# Now, make a mesh. Make sure to use enough elements
mesher = dic.Mesher(deg_n=3, deg_e=3, type="spline")
#mesh = mesher.mesh(image_stack) # Use this if you want to mesh with a GUI
mesh = mesher.mesh(image_stack,
Xc1=50,
Xc2=450,
Yc1=50,
Yc2=450,
n_ely=8,
n_elx=8,
GUI=False)
# Prepare the analysis input and initiate the analysis
input = dic.DICInput(mesh, image_stack)
input.tol = 1e-6
dic_job = dic.DICAnalysis(input)
logging.basicConfig(format='%(name)s:%(levelname)s:%(message)s', level=logging.INFO) # Path to folder containing images #path = r'./example_data/' # Use this formatting on Linux and Mac OS path = r'F:\GraduateProject\Python\PCA\speckle\\' # path = r'c:\path\to\example_data\\' # Use this formatting on Windows # Generate image instance containing all images found in the folder # images = dic.IO.image_stack_from_folder(path, file_type='.tif') images = dic.IO.image_stack_from_folder(path, file_type='.tif') images.set_filter(dic.filtering.lowpass_gaussian, sigma=1.) # Generate mesh # 这里Xc1，Xc2，Yc1,Yc2是ROI的大小，需要根据输入图象的大小修改 mesher = dic.Mesher(deg_e=3, deg_n=3) # mesh = mesher.mesh(images,Xc1=200,Xc2=1050,Yc1=200,Yc2=650,n_ely=8,n_elx=8, GUI=True) mesh = mesher.mesh(images, Xc1=20, Xc2=105, Yc1=20, Yc2=65, n_ely=8, n_elx=8, GUI=True) # Instantiate settings object and set some settings manually settings = dic.DICInput(mesh, images) # 图像数量 settings.max_nr_im = 2 settings.ref_update = [15] settings.maxit = 20 # If you want to access the residual fields after the analysis, this should be set to True settings.store_internals = False # Instantiate job object job = dic.DICAnalysis(settings)
def run_DIC(folder_path, n_images, filter, filter_sigma, maxim, maxit, padding,
poly_order, converge):
# Set the amount of info printed to terminal during analysis
logging.basicConfig(format='%(name)s:%(levelname)s:%(message)s',
level=logging.INFO)
# Path to folder containing images
path = folder_path
# Generate image instance containing all images found in the folder
images = dic.IO.image_stack_from_folder(path, file_type='.png')
images.use_every_n_image(int(n_images))
if filter == "Lowpass Gaussian":
images.set_filter(dic.filtering.lowpass_gaussian,
sigma=int(filter_sigma))
elif filter == "Highpass Gaussian":
images.set_filter(dic.filtering.highpass_gaussian,
sigma=int(filter_sigma))
elif filter == "Homomorphic Median":
images.set_filter(dic.filtering.homomorphic_median,
sigma=int(filter_sigma))
else:
pass
# Generate mesh
mesher = dic.Mesher(deg_e=3, deg_n=3, type="q4") #type="b_splines")
mesh = mesher.mesh(images,
Xc1=316,
Xc2=523,
Yc1=209,
Yc2=1055,
n_ely=18,
n_elx=5,
GUI=True)
# Instantiate settings object and set some settings manually (first video)
settings = dic.DICInput(mesh, images)
if bool(maxim):
settings.max_nr_im = int(maxim)
settings.ref_update = [15]
settings.maxit = int(maxit)
settings.tol = 1.e-6
settings.interpolation_order = int(poly_order)
# If you want to access the residual fields after the analysis, this should be set to True
settings.store_internals = True
# This setting defines the behaviour when convergence is not obtained
if converge == "Ignore":
settings.noconvergence = "ignore"
elif converge == "Update":
settings.noconvergence = "update"
elif converge == "Break":
settings.noconvergence = "break"
else:
pass
# Instantiate job object
job = dic.DICAnalysis(settings)
# Running DIC analysis
dic_results = job.run()
return images, dic_results
import logging
# Set the amount of info printed to terminal during analysis
logging.basicConfig(format='%(name)s:%(levelname)s:%(message)s',
level=logging.INFO)
# Path to folder containing images
path = r'./example_data/' # Use this formatting on Linux and Mac OS
#path = r'c:\path\to\example_data\\' # Use this formatting on Windows
# Generate image instance containing all images found in the folder
images = dic.IO.image_stack_from_folder(path, file_type='.tif')
#images.set_filter(dic.filtering.lowpass_gaussian, sigma=1.)
# Generate mesh
mesher = dic.Mesher(deg_e=3, deg_n=3, type="q4")
# If you want to see use a GUI, set GUI=True below
mesh = mesher.mesh(images,
Xc1=316,
Xc2=523,
Yc1=209,
Yc2=1055,
n_ely=36,
n_elx=9,
GUI=False)
# If you want to visualize nodes label
mesh.visualize_node()
# Instantiate settings object and set some settings manually
# Instantiate an image deformer
image_deformer = vlab.image_deformer.imageDeformer_rigid_body(amp=(shift_amp, 0.))
# Shift the images. Note that the first one is not shifted
shifted_images = image_deformer(img, n_imgs)
# We now add noise to the images
shifted_noisy_images = [noise_injector(image) for image in shifted_images]
# And put them into a stack which is the formatting needed for the DIC analysis
image_stack = dic.image_stack_from_list(shifted_noisy_images)
# Generate mesh
# We here use Q4 elements
mesher = dic.Mesher(type="q4")
# If you want to use a GUI, set GUI=True
mesh = mesher.mesh(image_stack, Xc1=20, Xc2=480, Yc1=20, Yc2=480, n_ely=num_elms, n_elx=num_elms, GUI=False)
# Instantiate settings object and set some settings manually
settings = dic.DICInput(mesh, image_stack)
# We allow the solver to use as many iterations as it needs
settings.maxit = 100
# This tolerance is the larges increment in nodal position [pixels]
settings.tol = 1e-6
# We use fourth order splines for grey-scale interpolation, minimizing interpolation bias
settings.interpolation_order = 4
# Instantiate job object
job = dic.DICAnalysis(settings)
# Running DIC analysis