Core objects#


A pipeline, strictly speaking, is a series of connected tubes running a fluid. In the scientific literature, the word refers to processing pipelines in which data are flowing, going through processing units as in tubes.

prose contains the structure to build modular image processing pipelines with three key objects: Image going through a list of Block assembled into a Sequence.

Main objects#

  • An Image contains the image data and metadata

  • A Block is a single unit of processing acting on the Image object, which can |read| and |write| its attributes, or |modify|.

  • A Sequence is a succession of Block.

With this architecture prose can deal with any type of image (especially FITS images)

Example: Hello World#

Let’s create a random set of images

from prose import Image, Block, Sequence
import numpy as np

images = [Image(data=np.random.rand(10, 10)) for i in range(5)]

Here is a block printing hello world and the image mean

class HelloWorld(Block):
    def run(self, image):
        image.mean = np.mean(
        print(f"Hello world (mean: {image.mean:.2f})")

and running a sequence with it

sequence = Sequence(
Hello world (mean: 0.47)
Hello world (mean: 0.50)
Hello world (mean: 0.52)
Hello world (mean: 0.49)
Hello world (mean: 0.52)