janelia_core.dataprocessing.utils
Utilities for working with time series data.
Module Contents
Functions
|
Gets image data for a single image. |
|
Gets registered image data for a single image. |
|
Gets processed image data for multiple images. |
|
Extracts one or more planes from image files, writing planes to separate files. |
|
Writes specified planes from a 3d image file to separate .h5 files. |
- janelia_core.dataprocessing.utils.get_image_data(image, img_slice: slice = slice(None, None, None), h5_data_group: str = 'default') numpy.ndarray
Gets image data for a single image.
This is a wrapper that allows us to get image data for a single image from a file or from a numpy array seamlessly in our code. If image is already a numpy array, image is simply returned as is. Otherwise, image is assumed to be a path to a image which is opened and the data is loaded and returned as a numpy array.
- Args:
image: Either a numpy array or path to the image.
img_slice: The slice of the image that should be returned
h5_data_group: The hdfs group holding image data in h5 files.
- Returns:
The image data.
- janelia_core.dataprocessing.utils.get_reg_image_data(image, image_slice: slice, image_shape: numpy.ndarray = None, t: dipy.align.imaffine.AffineMap = None, h5_data_group: str = 'default')
Gets registered image data for a single image.
This is a wrapper around get_image_data that allows the user to register an image before getting data from it.
- Args:
image: Either a numpy array or path to the image.
image_shape: The shape of the original image. This is used to check that the requested window contains valid data after registration. If t is none, this argument does not need to be supplied.
t: The registration transform. If this is none, no registration will be performed.
image_slice: The slice of the image that should be returned. Coordinates are after image registration.
h5_data_group: The hdfs group holding image data in h5 files.
- Returns:
The image data
- Raises:
ValueError: If the requested slice for the registered image includes voxels for which there was no data in the original image to calculate the registered image for.
- janelia_core.dataprocessing.utils.get_processed_image_data(images: list, func: types.FunctionType = None, img_slice=slice(None, None, None), t_dict: dict = None, func_args: list = None, h5_data_group='default', sc: pyspark.SparkContext = None) list
Gets processed image data for multiple images.
This is a wrapper that allows retrieving images from files or from numpy arrays, applying light processing independently to each image and returning the result.
Images can be optionally registered before apply preprocessing.
- Args:
images: A list of images. Each entry is either a numpy array or a path to an image file.
func: A function to apply to each image. If none, images will be returned unaltered. Should accept input of the form func(image: np.ndarray, **keyword_args)
img_slice: The slice of each image that should be returned before any processing is applied. If registration is applied (see t_dict parameter), then the slice coordinates are for images after registration.
t_dict: A dictionary with information for performing image registration as images are loaded. If set to None, no image registration will be performed. t_dict should have two fields:
transforms: A list of registration transforms to apply to the images as they are being read in. If none, no registration will be applied.
image_shape: This is the shape of the original images being read in.
func_args: A list of extra keyword arguments to pass to the function for each image. If None, no arguments will be passed.
h5_data_group: The hdfs group holding image data in h5 files.
sc: An optional pySpark.SparkContext object to use in speeding up reading of images.
- Returns:
The processed image data as a list. Each processed image is an entry in the list.
- janelia_core.dataprocessing.utils.write_planes_to_files(planes: numpy.ndarray, files: list, base_planes_dir: pathlib.Path, plane_suffix: str = 'plane', skip_existing_files=False, sc: pyspark.SparkContext = None, h5_data_group='data') list
Extracts one or more planes from image files, writing planes to separate files.
- Args:
planes: An array of indices of planes to extract.
files: A list of original image files to pull planes from.
base_planes_dir: The base directory to save plane files into. Under this folder, separate subfolders will be saved for each plane.
plane_suffix: The suffix to append to the file name to indicate the file contains just one plane.
skip_existing_files: If true, if a file for an extracted plane is found, it will not be overwritten. If false, then errors will be thrown if files for extracted planes are found to already exist. Setting this to true can be helpful if there is a need to run this function a second time to recover from an error.
sc: An optional spark context to use to write files in parallel.
h5_data_group: The h5_data_group that original images are stored under if reading in .h5 files.
- Returns:
A list of the directories that images for each plane are saved into.
- janelia_core.dataprocessing.utils.write_planes_for_one_file(file: pathlib.Path, planes: numpy.ndarray, plane_dirs: list, plane_suffix: str = 'plane', skip_existing_files=False, h5_data_group='default')
Writes specified planes from a 3d image file to separate .h5 files.
The new files will have the same name as the original with an added suffix to indicate they contain just one plane.
- Args:
planes: An array of indices of planes to extract.
file: The original image file to pull planes from.
plane_dirs: List of directories to save the file for each file into.
plane_suffix: The suffix to append to the file name to indicate the file contains just one plane.
skip_existing_files: If true, if a file for an extracted plane is found, it will not be overwritten. If false, then errors will be thrown if files for extracted planes are found to already exist. Setting this to true can be helpful if there is a need to run this function a second time to recover from an error.
h5_data_group: The h5_data_group that original images are stored under if reading in .h5 files.