Improve array extraction speed using OpenCV and NumPy

Question

I am trying to extract the alpha channel of a RGBA matrix in the following format:

[                      [
 [                      [
  [1, 2, 3, 4],   to     [4],
  [5, 6, 7, 8]   ====>   [8]
 ]                      ]
]                      ]

I wanted to know if the code I currently have can be improved in terms of speed:

import numpy
import cv2

image = np.full((10, 10, 4), 0, numpy.uint8)

r, g, b, a = cv2.split(image)

rgb = cv2.merge((r, g, b))
alpha = np.array([np.vstack(e).tolist() for e in a])  # Can this be faster?

With big matrices, the last line can take up to more than 0.1s to execute. I need it to be less than that.

Thank you!

@Divakar Wow this actually works perfectly! Can you explain to me briefly why this works? — Marc VILLAIN, Apr 30 '18 at 07:27
https://stackoverflow.com/a/40383002/ and the comments underneath. — Divakar, Apr 30 '18 at 07:33

score 0 · Answer 1 · answered Apr 30 '18 at 07:31

0

Just reshape it!

a.reshape((sizeX, sizeY, 1))

That should give you the same effect with a simple metadata change.

answered Apr 30 '18 at 07:31

shash

965
8
16

Improve array extraction speed using OpenCV and NumPy

1 Answers1