How to create a uint16 numpy array from a uint8 raw image data array

Question

When interfacing with a USB camera and grabbing image data, I am running into problems with saving the data as 16 bit. I have the camera pixel format set to Mono16, however whenever I save the raw image data to a numpy array, the array defaults to a uint8 datatype. I have noticed that the raw data array doubles in size when switching from Mono8 to Mono16. I have looked into numpy as_strided, but that still didn't fix the issues. Here is a snippet of code to help show a basic example of the situation.

>>>c.startCapture()
>>>fireSoftwareTrigger()
>>>im = c.retrieveBuffer()
>>>c.stopCapture()
>>>cols = im.getCols()
>>>cols
964
>>>rows = im.getRows()
>>>rows
724
>>>data = numpy.array(im.getData())
>>>len(data)
1395872
>>>rows*cols
697936
>>>data.dtype
dtype('uint8')

Is it defaulting when you save the array or when you create it? You can check the datatype with arr.dtype. — user2699
– user2699, Commented Mar 6, 2018 at 16:16
I edited the post to give a better explanation of the situation. As you can see, the image file size (rows,cols) is half the size of the actual raw data array. The datatype still defaults to uint8. — Jack Bates
– Jack Bates, Commented Mar 6, 2018 at 16:30
It seems that the raw data array is being returned as a ndarray with dtype of uint8. Even when the camera Format7 pixel setting is set to Mono16 — Jack Bates
– Jack Bates, Commented Mar 6, 2018 at 17:51
You could try view casting data = numpy.array(im.getData()).view(numpy.uint16) or if you have to control endianness '<u2' or '>u2' instead of numpy.uni16 — Paul Panzer
– Paul Panzer, Commented Mar 6, 2018 at 17:52

Paul Panzer · Accepted Answer · 2018-03-06 18:18:55Z

4

You can use view casting. Note that view casting reinterprets the data buffer without actually modifying it, so it is cheap.

data = numpy.array(im.getData()).view(numpy.uint16)

if the data's byte order matches your machine's native one.

Otherwise you'd need to use one of

data = numpy.array(im.getData()).view('<u2')
data = numpy.array(im.getData()).view('>u2')

the '>'/'<' in the dtype specification string stands for big/little endian. 'u' means unsigned int, '2' means two bytes.

answered Mar 6, 2018 at 18:18

Paul Panzer

53.3k3 gold badges59 silver badges103 bronze badges

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

Collectives™ on Stack Overflow

How to create a uint16 numpy array from a uint8 raw image data array

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related