Rusty Image Utils

Port of some slow functions from Python to a Rust library with Python bindings, mainly to experiment with rust-cpython

Published as rusty-img-utils on pypi for Python 3.5, 3.6 and 3.7 on linux and macos

Functions

darken_pixels

Darkens all pixels in the image by percentage, specified by amount. Any pixel that doesn't have a subpixel below than the cutoff will be ignored.

import img_utils
img_utils.darken_pixels(
    src_path="in_file.jpg",
    dst_path="out_file.jpg",
    amount=80,
    cutoff=200,
)

will take the in_file.jpg and lower each subpixel of the image by 80%, unless all the subpixels are above 200.

The RGB pixel 100, 220, 220 will be turned into 20, 44, 44 while 210, 220, 230 will be left alone.

extract_blues

Selects all pixels that have the blue pixel above a certain limit and either green or red the minimum difference blow the blue pixel. The selected pixels will be output as completely black while any other pixel is output as white.

import img_utils
img_utils.extract_blues(
    src_path="in_file.jpg",
    dst_path="out_file.jpg",
    min_diff=30,
    min_blue=100,
)

will take the in_file.jpg and any pixel that has the blue pixel 100 or higher and either red or green pixel at 70 or lower (100 - 30) will be black, all others will be white.

Performance comparison

Testing with 100x100 and 1000x1000 random noise images on a 2018 MBP

darken_pixels

*** Testing small size
Python_Darken: 12.14 ms avg (412 runs)
Rust_Darken: 1.15 ms avg (4357 runs)

Rust is 10.58x faster for small size

*** Testing normal size
Python_Darken: 1103.74 ms avg (5 runs)
Rust_Darken: 46.41 ms avg (108 runs)

Rust is 23.78x faster for normal size

extract_blues

*** Testing small size
Python_Extract: 4.94 ms avg (1012 runs)
Rust_Extract: 1.17 ms avg (4282 runs)

Rust is 4.23x faster for small size

*** Testing normal size
Python_Extract: 392.25 ms avg (13 runs)
Rust_Extract: 47.71 ms avg (105 runs)

Rust is 8.22x faster for normal size