Implement Box blur fast filter that could approximate gaussian filter

[light-le]

solve #168. The algorithm was derived from this blog post

[johnnv1]

can you add benchmarks to it as well?

[light-le]

You mean in crates/kornia-imgproc/benches/bench_filters.rs ? Sure ok

[edgarriba]

I would some simple numbers test too similar to the other functions to verify that’s doing the right thing

[light-le]

So I added 2 tests here: test_box_blur_fast() and test_gaussian_blur(). Both has the same input (0..25) to show that the outputs are not that much different.

I did attempt to use the same input (all 0.0, 9.0 in the middle) as test_fast_horizontal_filter(), if that's how you mean by this comment. The result was a little disappointing as there's a big difference between the outputs of the 2 methods. I figured it's because the test input was odd. It might be fitting for test_fast_horizontal_filter() but not for these. Therefore I went with something more randomized.

[edgarriba]

Why should be different? The test you describe should give you a box of ones, right ?

[light-le]

Not really. The test that gives me a box of ones was because the fast_horizontal_filter() was applied twice. But an actual fast_box_blur() test has the filter applied 6 times (or more). The numbers are a lot more spread out.

[edgarriba]

@johnnv1 any idea why python tests are failing (I believe it’s unrelated to this PR). Shouldn’t we be using the new just commands in https://github.com/kornia/kornia-rs/blob/main/.github/workflows/python_test.yml#L40

[johnnv1]

@johnnv1 any idea why python tests are failing (I believe it’s unrelated to this PR). Shouldn’t we be using the new just commands in https://github.com/kornia/kornia-rs/blob/main/.github/workflows/python_test.yml#L40

yeah, seems unrelated, but should be working

[edgarriba]

Can you expand this benchmark and report the numbers so that we know wether this method is really making what’s expected ?

https://github.com/kornia/kornia-rs/blob/main/crates/kornia-imgproc/benches/bench_filters.rs

I highly suggest once you have the benchmark setup that you play around with it and try to do micro optimisations like reusing as much as possible pre-computed variables as I suggested in the review to see how affects in the benchmarks.

[edgarriba]

Why should be different? The test you describe should give you a box of ones, right ?

[light-le]

So regarding performance, box_blur_fast() filter is independent of the kernel size, but you have to apply the fast_horizontal_filter() 6 times over in 1 run. Therefore it would be slower than native Gaussian blur when the kernel size is small. Here's the performance with all of your suggested micro-optimization applied:

• 256x224 image size, box_blur_fast() takes 2.98 ms.
  • Kernel size 3, native blur takes 1.5 ms
  • Kernel size 5, 1.9 ms
  • Size 7, 2.3 ms
  • 9, 3 ms
  • 11, 3.68 ms
  • 17, 5.69 ms
• 512x448 image size, box_blur_fast() takes 12.25 ms
  • Kernel size 3, native blur takes 6 ms
  • Kernel size 5, 7.78 ms
  • Size 7, 9.5 ms
  • 9, 12.1 ms
  • 11, 14.84 ms
  • 17, 22.64 ms
• 1024x896 image size, box_blur_fast() takes 50.3 ms
  • Kernel size 3, native blur takes 24 ms
  • Kernel size 5, 31.4 ms
  • Size 7, 38.54 ms
  • 9, 48.3 ms
  • 11, 59.3 ms
  • 17, 91 ms

Maybe there're some other optimizations I can do? I'm kinda afraid to introduce unsafe Rust to my code but that's something I could try.