Creating a new numpy array with modified element data type

Question

I had expected the following to create a new numpy array from the shape of an existing array but with modified element data type. My original array is an image with 8bit RGB pixels. I want to create a new array using same shape but with uint16 data type. The purpose is then to convert the image into 16bit pixels and perform some math. To my surprise the following didn't work.

>>> import scipy.misc        as msc
>>> import numpy             as np
>>> img_rgb = msc.imread('Jupiter_20160417_53.png')
>>> img_rgb.dtype
dtype('uint8')
>>> img_rgb.shape
(480, 640, 3)
>>> new= np.zeros(img_rgb.shape,dtype=uint16)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'uint16' is not defined

What did I miss?

Thanks, Gert

>>> a = np.zeros((3,3),dtype=np.uint16) works for me (import numpy as np of course) — user1121588
– user1121588, Commented Jun 12, 2016 at 6:28

hpaulj · Accepted Answer · 2016-06-12 06:29:04Z

1

The np.:

In [2]: np.zeros((3,4),dtype=np.uint16)
Out[2]: 
array([[0, 0, 0, 0],
       [0, 0, 0, 0],
       [0, 0, 0, 0]], dtype=uint16)

'uint16' (the string) would have worked as well.

int and float are Python names; almost all the other dtypes are numpy specific, and require either the np. namespace or a string name (which numpy understands).

The error NameError: name 'uint16' is not defined means that uint16 is not a variable in the main namespace. In other words it's not a builtin variable (or function), and it hasn't been imported as such. It is part of the numpy namespace which you imported a np.

e.g.

In [8]: z
...
NameError: name 'z' is not defined

answered Jun 12, 2016 at 6:29

hpaulj

233k14 gold badges260 silver badges392 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

Gert Gottschalk Over a year ago

Thank you hpaulj. The 'np.' prefix worked. What is the better code style? np.zeros((3,4),dtype='uint16') or np.zeros((3,4),dtype=np.uint16) ?

hpaulj Over a year ago

I don't think it matters. More details here: docs.scipy.org/doc/numpy-1.10.0/reference/arrays.scalars.html

Flash Thunder Over a year ago

I wonder why it doesn't it work for me, np.uint8 does, but on np.uint16 CV throws an exception that Assertion failed in cvShowImage

Collectives™ on Stack Overflow

Creating a new numpy array with modified element data type

1 Answer 1

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related