'measurement frame' should be double matrix

[tashrifbillah]

Guys, looks like we have a problem here. The measurement frame returned should be a double matrix, not integer. https://github.com/mhe/pynrrd/blob/master/nrrd/reader.py#L103

We need the exact measurement frame defined in the NRRD header. See the following returned by unu head file.nrrd:

measurement frame: (-1,0,-0.0023) (-0.00029999999999999997,-0.98850000000000005,0.1512) (-0.0022000000000000001,0.1512,0.98850000000000005)

However, due to typecasting to int, nrrd.read( ) is returning just

hdr['measurement frame']
array([[-1,  0,  0],
       [ 0,  0,  0],
       [ 0,  0,  0]])

Try out the following, you will notice the difference:

abc= np.array([[-1,0,-0.0023], [-0.00029999999999999997,-0.98850000000000005,0.1512], [-0.0022000000000000001,0.1512,0.988500000000
    ...: 00005]], dtype= 'double')

print(abc)
array([[-1.000e+00,  0.000e+00, -2.300e-03],
       [-3.000e-04, -9.885e-01,  1.512e-01],
       [-2.200e-03,  1.512e-01,  9.885e-01]])

ghi= np.array([[-1,0,-0.0023], [-0.00029999999999999997,-0.98850000000000005,0.1512], [-0.0022000000000000001,0.1512,0.988500000000
    ...: 00005]], dtype= 'int')

print(ghi)
array([[-1,  0,  0],
       [ 0,  0,  0],
       [ 0,  0,  0]])

I think we must make it a double matrix like other filed types set there.
cc: @ihnorton

[ihnorton]

Grr. This was fixed in #47 and broken in
#57 -- the test didn't catch it because the float (identity) array still compares as equal to the int one.

[ihnorton]

Fixing

[tashrifbillah]

Thanks!

[addisonElliott]

Sorry, that's on me! I don't use the measurement frame much in my day to day, thus I didn't bother checking it. Thanks 👍