Conversation
QOL and bugfixes following 2D DIC workshop
Mooseherder Integration
…it doesn't exist.
…clude speckle gen, and added specklegen tests.
…nction, 2.Converted number of periods into a speckle size, add random seed setting in Perlin/fractal noise, 3. Split diagnostics function into just stats and everything chart-related, let the diagnostics function return the plot handle as well. - Created function descriptions in a standard format. - Added an option to create a speckle pattern by first creating a regular circle pattern and then randomly perturbating speckles.
…comments to the examples.
|
@WieraB have we merged in the latest version of dev into this to update everything? Should be my massive refactor from me and some DIC stuff from Joel |
scripts/gengold_specklegen.py
Outdated
| #=============================================================================== | ||
|
|
||
| import pyvale.verif.specklegold as specklegold | ||
| import pyvale.verif.specklegneconst as specklegneconst |
There was a problem hiding this comment.
Might be worth quickly fixing the typo here "specklegenconst" - should be gen I think for consistency with the rest of the package
There was a problem hiding this comment.
I've merged dev into wb-dev now. Also, I amended the typo.
|
The docs don't seem to be adding the specklegen module to the side menu and the examples aren't listed on the page - probably something needs to be added to the rst. |
…specklegen.rst file needed by doxygen.
I think I fixed it now. The changes in |
| sigma = 4.0 | ||
| reduce_overlap = False | ||
| type_gen = "random_disks" | ||
| output_path = "src/pyvale/examples/specklegen/output/ex1a" |
There was a problem hiding this comment.
As default we should probably be using pathlib and Path objects to manage all directories. If you look at some of the sensor sim examples we create a standard directory called 'pyvale-output' in the current working directory:
output_path = Path.cwd() / "pyvale-output"
if not output_path.is_dir():
output_path.mkdir(parents=True, exist_ok=True)
|
|
||
| print('Start') | ||
|
|
||
| assert theme in ['black_on_white', 'white_on_black'], "Theme should be either 'black_on_white' or 'white_on_black'." |
There was a problem hiding this comment.
If you want to restrict the options a user can select I would probably use and enum:
import enum
class Theme(enum.Enum):
BLACK_ON_WHITE = enum.auto()
| subfolder = f"/{type_gen}_{speckle_size}_{screen_size_width}_{screen_size_height}_{bit_depth}_{theme}_{seed}_{sigma}_{reduce_overlap}" | ||
| print(subfolder) | ||
| save_path = output_path + subfolder | ||
| if not os.path.exists(save_path): |
There was a problem hiding this comment.
See comment above about pathlib
| # We now generate the speckle pattern using the specified parameters. | ||
| # The background and foreground colours are set based on the chosen theme and bit depth. | ||
|
|
||
| speckle_area = np.pi * (speckle_size / 2) ** 2 |
There was a problem hiding this comment.
These 3 calculations should probably be automated for the user and hidden within the generate_speckles function - the user should only need to specify the black-white ratio and all these calculations are done for them in the function
|
|
||
| print("") | ||
| print('Starting speckle pattern diagnostics...') | ||
| results = specklegen.speckle_pattern_statistics(image, dynamic_range) |
There was a problem hiding this comment.
Probably easier for the user to just specify the number of bits and the dynamic range is calculated in the function.
| plt.rc('text', usetex=True) | ||
| fontsize1= 17 | ||
|
|
||
| fig, axes = plt.subplots(1, 1, figsize=(7.0, 5.0)) |
There was a problem hiding this comment.
In the sensor sim module (probably moving to the python common module soon) we have a standard dataclass for formatting plots: https://github.com/Computer-Aided-Validation-Laboratory/pyvale/blob/main/src/pyvale/sensorsim/visualopts.py.
Have a look at some of the plotting functions for the sensor sim module and see if you can make it consistent across modules.
| print("") | ||
| print('Starting speckle pattern diagnostics...') | ||
| results = specklegen.speckle_pattern_statistics(image, dynamic_range) | ||
| plots = specklegen.speckle_pattern_plots(image, dynamic_range, save_path) |
There was a problem hiding this comment.
Can we also include the images in the docs
| Optimal parameters for horisontal Gaussian fit. | ||
| popt_V : array | ||
| Optimal parameters for vertical Gaussian fit. | ||
| h_profile : np.ndarray |
There was a problem hiding this comment.
In the docstring I will generally try and make sure I list what the dimensions of the array mean - here is it just: shape=(num_px_y,num_px_x)?
| sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), '..'))) | ||
| import pyvale.specklegen as specklegen | ||
|
|
||
| width_range = np.arange(10, 1000, 250) |
There was a problem hiding this comment.
by convention constants outside of functions should be in CAPS. That being said there is an easier way to send a consistent set of variables to all functions using @pytest.fixture. See examples here:https://github.com/Computer-Aided-Validation-Laboratory/pyvale/blob/main/tests/mooseherder/mooseherd_test.py
| if dist_sq <= radius_sq: | ||
| assert img[y, x] == fg_colour | ||
|
|
||
| @pytest.mark.parametrize("total_speckles", total_speckles_range) |
There was a problem hiding this comment.
Generally I would only use parametrize if I was wanting to test multiple combinations of variables. If you are only testing one variable combination like you have here then you want pytest.fixture. For an example of when to use parameterize see:
@pytest.mark.parametrize(
('n_para','expected'),
(
(0, 1),
(-1,1),
(2.5,2),
(os.cpu_count()+1,os.cpu_count()) # type: ignore
)
)
def test_set_num_para_sims(n_para: int, expected: int, herd: MooseHerd):
herd.set_num_para_sims(n_para)
assert herd._n_para_sims == expected
3 participants
Speckle generation module consisting of the following: