PySilCam

This is documentation, auto-generated by Sphinx, for the PySilCam python interface and analysis for the SINTEF SilCam.

You can find out more about the system and higher-level descriptions on the wiki, here.

The code repository can be found here.

Top-Level functions and entry points

pysilcam.__main__.silcam()

Main entry point function to acquire/process images from the SilCam.

Use this function in command line arguments according to the below documentation.

Usage:

silcam acquire <configfile> <datapath> silcam process <configfile> <datapath> [–nbimages=<number of images>] [–nomultiproc] [–appendstats] silcam realtime <configfile> <datapath> [–discwrite] [–nomultiproc] [–appendstats] [–discread] silcam -h | –help silcam –version

Parameters

  • Acquire images (acquire) –

  • Process images (process) –

  • Acquire images from the camera and process them in real time (realtime) –

Options:
--nbimages=<number of images>

Number of images to process.

--discwrite

Write images to disc.

--nomultiproc

Deactivate multiprocessing.

--appendstats

Appends data to output STATS.h5 file. If not specified, the STATS.h5 file will be overwritten!

-h –help Show this screen. –version Show version. –discread emergency disc read version of realtime analysis, to be run seperately but at

the same time as silcam acquire

pysilcam.silcreport.silcreport()

Generate a report figure for a processed dataset from the SilCam.

You can access this function from the command line using the below documentation.

Usage:
silcam-report <configfile> <statsfile> [–type=<particle_type>]

[–dpi=<dpi>] [–monitor] [–filter-stats=<y1,x1,y2,x2>]

Parameters

  • configfile – The config filename associated with the data

  • statsfile – The -STATS.h5 filename associated with the data

Options:
--type=<particle_type>

The particle type to summarise. Can be: ‘all’, ‘oil’, or ‘gas’

--dpi=<dpi>

DPI resolution of figure (default is 600)

--monitor

Enables continuous monitoring (requires display)

–filter-stats=<y1,x1,y2,x2> Filters stats file to only include particle

from the specified coordinate rectangle. (y1,x1) are the lower-left corner, and (y2,x2) are the upper-right corner of the rectangle.

-h –help Show this screen.

Main processing functions

Processing and analysis is performed using the below functions.

pysilcam.__main__.processImage(nnmodel, class_labels, image, settings, logger, gui)

Proceses an image

Parameters

  • nnmodel (tensorflow model object) – loaded using sccl.load_model()

  • class_labels (str) – loaded using sccl.load_model()

  • image (tuple) – tuple contianing (i, timestamp, imc) where i is an int referring to the image number timestamp is the image timestamp obtained from passing the filename imc is the background-corrected image obtained using the backgrounder generator

  • settings (PySilcamSettings) – Settings read from a .ini file

  • logger (logger object) – logger object created using configure_logger()

  • gui=None (Class object) – Queue used to pass information between process thread and GUI initialised in ProcThread within guicals.py

Returns

stats dataframe containing particle statistics

Return type

stats_all (DataFrame)

pysilcam.process.statextract(imc, settings, timestamp, nnmodel, class_labels)

extracts statistics of particles in imc (raw corrected image)

Parameters

  • imc – background-corrected image

  • timestamp – timestamp of image collection

  • settings – PySilCam settings

  • nnmodel – loaded tensorflow model from silcam_classify

  • class_labels – lables of particle classes in tensorflow model

Returns

(list of particle statistics for every particle, according to Partstats class) imbw : segmented image saturation : percentage saturation of image

Return type

stats

pysilcam.process.image2blackwhite_fast(imc, greythresh)

converts corrected image (imc) to a binary image using greythresh as the threshold value (fixed scaling of greythresh is done inside)

Parameters

  • imc – background-corrected image

  • greythresh – threshold multiplier (greythresh is multiplied by 50th percentile of the image histogram)

Returns

segmented image (binary image)

Return type

imbw

pysilcam.process.image2blackwhite_accurate(imc, greythresh)

converts corrected image (imc) to a binary image using greythresh as the threshold value (some auto-scaling of greythresh is done inside)

Parameters

  • imc – background-corrected image

  • greythresh – threshold multiplier (greythresh is multiplied by 50th percentile of the image histogram)

Returns

segmented image (binary image)

Return type

imbw

pysilcam.process.clean_bw(imbw, min_area)

cleans up particles which are too small and particles touching the border

Parameters

  • imbw – segmented image

  • min_area – minimum number of accepted pixels for a particle

Returns

cleaned up segmented image

Return type

imbw (DataFrame)

pysilcam.process.measure_particles(imbw, imc, settings, timestamp, nnmodel, class_labels)

Measures properties of particles

Parameters

  • imbw (full-frame binary image) –

  • imc (full-frame corrected raw image) –

  • image_index (some sort of tag for location matching) –

Returns

stats (list of particle statistics for every particle, according to Partstats class)

Classification functions

pysilcam.silcam_classify.check_model(model_path)

Raises errors if classification model is not found, or if it is not a valid file.

Parameters

model_path (str) – path to particle-classifier e.g. ‘/mnt/ARRAY/classifier/model/particle_classifier.h5’ usually obtained from settings.NNClassify.model_path

pysilcam.silcam_classify.get_class_labels(model_path)

Read the header file that defines the catagories of particles in the model

Parameters

model_path (str) – path to particle-classifier e.g. ‘/testdata/model_name/particle_classifier.h5’ usually obtained from settings.NNClassify.model_path

Returns

labelled catagories which can be predicted

Return type

class_labels (str)

pysilcam.silcam_classify.load_model(model_path)

Load the trained tensorflow keras model

Parameters

model_path (str) – path to particle-classifier e.g. ‘/testdata/model_name/particle_classifier.h5’

Returns

loaded tf.keras model from load_model()

Return type

model (tf model object)

pysilcam.silcam_classify.predict(img, model)

Use tensorflow model to classify particles

Parameters

  • img (uint8) – a particle ROI, corrected and treated with the silcam explode_contrast function

  • model (tf model object) – loaded tfl model from load_model()

Returns

the probability of the roi belonging to each class

Return type

prediction (array)