通过PyFITS裁剪FITS数据立方体图像

Question

I have looked to find a solution for my issue, but couldn't. I have a FITS data cube, and I need to crop it by PyFITS. When I do it by my script, finally I'll have a 2-D FITS image! The first dimension is energy, and the second and thirds are longitude and latitude, respectively.

My script is as below:

#!/usr/bin/env python
import pyfits
import os
import sys


def CropFitsFile( src, dst, xs, xe, ys, ye):
    fh = pyfits.open(src)
    for eng in range(0,2):
        img = fh[0].data[eng,ys:ye,xs:xe]
        header = fh[0].header
        newfh=pyfits.PrimaryHDU(data=img,header=header)
        if os.path.exists(dst):
            os.remove(dst)
        newfh.writeto(dst)


if __name__ == "__main__":
    CropFitsFile(
        src=sys.argv[1],
        dst=sys.argv[2],
        xs=int(sys.argv[3]),
        xe=int(sys.argv[4]),
        ys=int(sys.argv[5]),
        ye=int(sys.argv[6])
        )

Answer 1

If I understand correctly you want to slice a 3D array but keep the third dimensions (even if it's just size 1).

This is a question about Numpy arrays. When you have an N-dimensional numpy array, passing a scalar index for one dimension returns an array of dimension N-1, sliced along the axis you indexed. For example:

>>> arr = np.arange(27).reshape(3, 3, 3)
>>> arr
array([[[ 0,  1,  2],
        [ 3,  4,  5],
        [ 6,  7,  8]],

       [[ 9, 10, 11],
        [12, 13, 14],
        [15, 16, 17]],

       [[18, 19, 20],
        [21, 22, 23],
        [24, 25, 26]]])
>>> arr[0]
array([[0, 1, 2],
       [3, 4, 5],
       [6, 7, 8]])
>>> arr[1]
array([[ 9, 10, 11],
       [12, 13, 14],
       [15, 16, 17]])

You can also slice along a different axis like:

>>> arr[:,0,:]
array([[ 0,  1,  2],
       [ 9, 10, 11],
       [18, 19, 20]])

If you want, for whatever reason, to return an N-dimensional array instead of an N-1 dimensional array, the easiest way is to explicitly request a slice of size 1, instead of using scalar index. For example:

>>> arr[0:1]
array([[[0, 1, 2],
        [3, 4, 5],
        [6, 7, 8]]])

I'll give the same advice I've given on other questions like this: this isn't really a question about PyFITS beyond the fact that your data came out of a FITS file. PyFITS, like most scientific Python libraries, returns data as numpy arrays. These are the main data structure used for numeric data in most scientific Python applications, so learning some basics of numpy are sort of a prerequisite, for better or worse, to doing data analysis in Python. If you've ever used MATLAB, NumPy arrays are similar to arrays in MATLAB. You can start with my short tutorial, but there are others (and probably better ones too :) github.com/embray/notebooks/blob/master/numpy.ipynb

Answer 2

from astropy.io import fits

Ccube = fits.open('Cha_binned_ccube.fits', mode='update')
Ccube.info()
Ccube[0].shape
Ccube[0].data = Ccube[0].data[0:3,0:181,0:402]
Ccube[0].writeto('Cha_binned_ccube_resize.fits')