A numpy extension for efficient and powerful image analysis workflow

impy is an all-in-one image analysis library, equipped with parallel processing, GPU support, GUI based tools and so on.

The core array, ImgArray, is a subclass of numpy.ndarray, tagged with information such as

• image axes
• scale of each axis
• directory of the original image
• and other image metadata

Documentation

Documentation is available here.

Installation

• use pip

pip install impy-array

• from source

git clone https://github.com/hanjinliu/impy

Code as fast as you speak

Almost all the functions, such as filtering, deconvolution, labeling, single molecule detection, and even those pure numpy functions, are aware of image metadata. They "know" which dimension corresponds to "z" axis, which axes they should iterate along or where to save the image. As a result, your code will be very concise:

import impy as ip
import numpy as np

img = ip.imread("path/to/image")       # Read images with metadata.
img["z=3;t=0"].imshow()                # Plot image slice at z=3 and t=0.
img_fil = img.gaussian_filter(sigma=2) # Paralell batch denoising. No more for loop!
img_prj = np.max(img_fil, axis="z")    # Z-projection (numpy is aware of image axes!).
img_prj.imsave(f"Max-{img.name}")      # Save in the same place. Don't spend time on searching for the directory!

Supported file formats

impy automatically chooses proper reader/writer according to the extension.

• Tiff file (".tif", ".tiff")
• MRC file (".mrc", ".rec", ".st", ".map", ".map.gz")
• Zarr file (".zarr")
• Other image file (".png", ".jpg")

Switch between CPU and GPU

impy can internally switches the functions between numpy and cupy. You can use GPU for calculation very easily.

img.gaussian_filter()  # <- CPU
with ip.use("cupy"):
    img.gaussian_filter()  # <- GPU
ip.Const["RESOURCE"] = "cupy"  # <- globally use GPU

Seamless interface between napari

napari is an interactive viewer for multi-dimensional images. impy has a simple and efficient interface with it, via the object ip.gui. Since ImgArray is tagged with image metadata, you don't have to care about axes or scales. Just run

ip.gui.add(img)

Extend your function for batch processing

Already have a function for numpy and scipy? Decorate it with @ip.bind

@ip.bind
def imfilter(img, param=None):
    # Your function here.
    # Do something on a 2D or 3D image and return image, scalar or labels
    return out

and it's ready for batch processing!

img.imfilter(param=1.0)

Commaind line usage

impy also supports command line based image analysis. All method of ImgArray is available from commad line, such as

impy path/to/image.tif ./output.tif --method gaussian_filter --sigma 2.0

which is equivalent to

import impy as ip
img = ip.imread("path/to/image.tif")
out = img.gaussian_filter(sigma=2.0)
out.imsave("./output.tif")

For more complex procedure, it is possible to send image directly to IPython

impy path/to/image.tif -i

thr = img.gaussian_filter().threshold()

or to napari

impy path/to/image.tif -n