combine two uint8 into one uint16

Question

I have an array in bytes, content is camera frame 480x640 x 2 bytes per pixel

raw = np.asarray (d.GetFrameData ( ), dtype=np.uint8)
print (raw, raw.shape, type(raw))`

print output is below, 614400=480x640x2, pixel value is 0~255

[0 0 0 ... 0 0 0] (614400,) <class 'numpy.ndarray'>

what's correct way to convert it into an array with 480x640 with pixel value from 0~65535 ?

Thx in advance

i tried to use dtype=np.uint16, but its shape is still (614400,）

reshape (640, 480, 2) is workable, but that's not what i need

i need a result in reshape (640, 480) with uint16 size

Answer 1

Thx @mechanic-pig If I replace the first line of your code to my frame data, it works

raw = np.asarray (d.StreamFrameData ( ), dtype=np.uint8)
#raw = np.random.default_rng().integers(0, 256, 480 * 640 * 2, np.uint8)
print (raw)
print (raw.view(np.uint16).reshape(640, 480))
print (raw.view('<u2').reshape(640, 480))
print (raw.view('>u2').reshape(640, 480))

Answer 2

What you are looking for is numpy.ndarray.view():

>>> raw = np.random.default_rng().integers(0, 256, 480 * 640 * 2, np.uint8)
>>> raw
array([205,  10,  58, ...,  26, 204,  55], dtype=uint8)
>>> # dtype can also be 'u2', '=u2', 'H' or '=H',
>>> # whether it is big or little endian depends on the hardware
>>> raw.view(np.uint16).reshape(640, 480)  
array([[ 2765, 50234, 24681, ..., 44897, 64548, 55999],
       [20626, 58518, 46835, ..., 28429, 29253, 16188],
       [65218, 63270, 55857, ..., 57002, 32086, 50488],
       ...,
       [ 4825, 17130, 49826, ..., 33953,  7028, 59005],
       [64006, 14762, 55052, ..., 37743,  6392, 31599],
       [33882,  6208, 34247, ..., 28039,  6749, 14284]], dtype=uint16)
>>> raw.view('<u2').reshape(640, 480)  # or '<H', it is little endian
array([[ 2765, 50234, 24681, ..., 44897, 64548, 55999],
       [20626, 58518, 46835, ..., 28429, 29253, 16188],
       [65218, 63270, 55857, ..., 57002, 32086, 50488],
       ...,
       [ 4825, 17130, 49826, ..., 33953,  7028, 59005],
       [64006, 14762, 55052, ..., 37743,  6392, 31599],
       [33882,  6208, 34247, ..., 28039,  6749, 14284]], dtype=uint16)
>>> raw.view('>u2').reshape(640, 480)  # or '>H', it is big endian
array([[52490, 15044, 26976, ..., 25007,  9468, 49114],
       [37456, 38628, 62390, ...,  3439, 17778, 15423],
       [49918,  9975, 12762, ..., 43742, 22141, 14533],
       ...,
       [55570, 59970, 41666, ..., 41348, 29723, 32230],
       [ 1786, 43577,  3287, ..., 28563, 63512, 28539],
       [23172, 16408, 51077, ..., 34669, 23834, 52279]], dtype='>u2')