Encode rows of boolean numpy array to bytes

Question

I have a numpy array of dimensions Nx8, with dtyp=boolean I want to convert it into a numpy 1-d array where each row is turned into a byte, by bin2dec

x = array([[ True,  True, False, False,  True,  True, False, False],
       [ False,  False, False, False,  True,  True, False, False],
       [ True,  False, False, False,  False,  False, False, False]], dtype=bool)

I'd like the output to be:

y = array([204 ,12, 128], dtype=uint8)

but whats up with the dimensions? (and how big is N going to be?) — usethedeathstar
– usethedeathstar, Commented Oct 22, 2013 at 12:30
what is the question exactly? X is a 3x8 array, i want to turn each line of 8 booleans into a single int, so the output would be a 1-d array of length 3 and dtype int8. N is potentially quite large, can be 10,000-20,000 — eran
– eran, Commented Oct 22, 2013 at 12:34
This is actually an excellent question. I'm not sure why you're getting downvoted (other than mistyping int8 instead of uint8). — Joe Kington
– Joe Kington, Commented Oct 22, 2013 at 12:59

Steve Barnes · Accepted Answer · 2013-10-22 13:15:47Z

9

>>> np.packbits(np.uint8(x))
array([204,  12, 128], dtype=uint8)

How that?

answered Oct 22, 2013 at 13:15

Steve Barnes

28.5k6 gold badges68 silver badges80 bronze badges

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

1 Comment

Jaime Over a year ago

I'd go with x.view(np.uint8), because I know for sure it doesn't copy any data. Other than that, +1.

usethedeathstar · Accepted Answer · 2013-10-22 13:08:31Z

I guess this will do:

import numpy
x = numpy.array([[ True,  True, False, False,  True,  True, False, False],
       [ False,  False, False, False,  True,  True, False, False],
       [ True,  False, False, False,  False,  False, False, False]], dtype=bool)
x2 = 1*x # makes True become 1 and False become 0
x3 = numpy.zeros((3), dtype = numpy.uint8) # change 3 to 20000 or whatever the length of your array is
for j in range(x2.shape[1]):
    x3 += x2[:,j]*(2**(7-j))
print x3
[204  12 128]

Tell me how long it takes for your long array, if it goes too slow, i will try to push the for-loop down to numpy to speed it up. (needs to be uint8 instead of int8, otherwise the result is [ -52 12 -128])

edit: should actually not be that slow, since the for-loop only runs 8 times (once per float)

Collectives™ on Stack Overflow

Encode rows of boolean numpy array to bytes

2 Answers 2

1 Comment

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Related