How can I use Python to convert RGB565 byte array to RGB888 byte array?

Question

According to my question for RGB888 to RGB565, I would like to do RGB565 to RGB888, here is my testing code, however I got stuck on convert to RGB888 byte array.

import numpy as np
np.random.seed(42)
im = np.random.randint(0,256,(1,4,2), dtype=np.uint8)

# >>> im.nbytes
# 8
# >>> im
# array([[[102, 220],
#        [225,  95],
#        [179,  61],
#        [234, 203]]], dtype=uint8)

# Make components of RGB888
R8 = (im[...,0] & 0xF8).astype(np.uint32) << 8
G8 = (im[...,0] & 0x07).astype(np.uint32) << 5 | (im[...,1] & 0xE0).astype(np.uint32)
B8 = (im[...,1] & 0x1F).astype(np.uint32)
RGB888 = R8 | G8 | B8

# >>> RGB888.nbytes
# 16 <= here I think it should be 12 (4x3 bytes)

# >>> RGB888.reshape(1, 4, 3)
# Traceback (most recent call last):
#   File "<input>", line 1, in <module>
# ValueError: cannot reshape array of size 4 into shape (1,4,3)

When I use astype(np.uint16), some values become 0 because it need larger data type to store, that's why I use unit32 on above code.

I know unit32 will make above code's RGB888 size to be 16, so I would like to ask if any other correct way to transfer RGB565 to RGB888?

You should start from a 2-dimensional np.uint16 array where each element corresponds to a single packed pixel. — AKX
– AKX, Commented May 15, 2020 at 9:55
As @AKX says, RGB565 is naturally a uint16 type, so it would make more sense to start with that. — Mark Setchell
– Mark Setchell, Commented May 15, 2020 at 9:59

AKX · Accepted Answer · 2020-05-15 12:10:13Z

5

Something like this should get you from RGB565 uint16 to three uint8 channel arrays, which you can then dstack into a single 3-dimensional RGB image:

import numpy as np
np.random.seed(42)
im = np.random.randint(0,65536,(4,4), dtype=np.uint16)

MASK5 = 0b011111
MASK6 = 0b111111

# TODO: BGR or RGB? Who knows!
b = (im & MASK5) << 3
g = ((im >> 5) & MASK6) << 2
r = ((im >> (5 + 6)) & MASK5) << 3

# Compose into one 3-dimensional matrix of 8-bit integers
rgb = np.dstack((r,g,b)).astype(np.uint8)

EDIT: to convert an W x H x 2 array of uint8s into a W x H array of uint16s,

import numpy as np
np.random.seed(42)
im = np.random.randint(0,256,(4,4,2), dtype=np.uint8)

b1 = im[:,:,0].astype(np.uint16)
b2 = im[:,:,1].astype(np.uint16)
im = (b1 << 8 | b2)

You may need to swap b1 and b2 around depending on the endianness of your source array.

edited May 15, 2020 at 12:10

answered May 15, 2020 at 10:02

AKX

171k16 gold badges146 silver badges228 bronze badges

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

8 Comments

Mark Setchell Over a year ago

You can stack with RGB = np.dstack((r,g,b)) :-)

AKX Over a year ago

Cheers Mark :) Adding that in. (Funny that stack's documentation doesn't link to dstack...)

Mark Setchell Over a year ago

OP may also need .astype(np.uint8) tacked on the end, since we started with np.uint16.

AKX Over a year ago

@MarkSetchell Yup, fair point (though I'm sure OP'd notice sooner rather than later! :-D )

AKX Over a year ago

@MarkSetchell Seeing as split already had see-alsos for the variants, I figured why not make a PR to fix the docs :) github.com/numpy/numpy/pull/16249

|

Collectives™ on Stack Overflow

How can I use Python to convert RGB565 byte array to RGB888 byte array?

1 Answer 1

8 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

8 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related