deconvolve

aiapy.psf.deconvolve(smap, psf=None, iterations=25, clip_negative=True, use_gpu=True)[source]

Deconvolve an AIA image with the point spread function.

Perform image deconvolution on an AIA image with the instrument point spread function using the Richardson-Lucy deconvolution algorithm 1.

Note

If the cupy package is installed and your machine has an NVIDIA GPU, the deconvolution will automatically be accelerated with CUDA. This can lead to more than an order of magnitude in performance increase compared to pure numpy on a CPU. For more information on PSF deconvolution on a GPU, see 2.

Parameters
  • smap (Map) – An AIA image

  • psf (ndarray, optional) – The point spread function. If None, it will be calculated

  • iterations (int, optional) – Number of iterations in the Richardson-Lucy algorithm

  • clip_negative (bool, optional) – If the image has negative intensity values, set them to zero.

  • use_gpu (bool, optional) – If True and cupy is installed, do PSF deconvolution on the GPU with cupy.

Returns

Deconvolved AIA image

Return type

Map

See also

psf

References

1

https://en.wikipedia.org/wiki/Richardson%E2%80%93Lucy_deconvolution

2

Cheung, M., 2015, GPU Technology Conference Silicon Valley, GPU-Accelerated Image Processing for NASA’s Solar Dynamics Observatory