Constrained Diffusion Decomposition

A new approach to image decomposition

View the Project on GitHub

An example is avaliable.

Input:

numpy nd array, of shape e.g. (nx, ny, nz).

Output:

result: numpy nd array, of shape (m, nx, ny, nz). The mth commponent contain structures of sizes 2$^(m-1)$ to 2$^m$ pixels.

residual: numpy nd array, of shape (nx, ny, nz) the input data will be recovered as input = sum_i result[i] + residual.

For example

More examples more examples

Usage:

(a) under the shell

python constrained_diffusion_decomposition.py input.fits

the output file will be named as input.fits_scale.fits.

(b) inside python

import constrained_diffusion_decomposition as cdd

result, residual = cdd.constrained_diffusion_decomposition(data)

How it is done:

Assuuming an input of I(x, y),t he decomposition is achieved by solving the equation

    \frac{\partial I_t }{\partial t} ={\rm sgn}(I_t) \mathcal{H}({- \rm sgn}(I_t) \nabla^2 I_t) \nabla^2 I_t

where t is related to the scale l by t = l**2.

References:

Li 2022, Multi-Scale Decomposition of Astronomical Maps – Constrained Diffusion Method