Re: [gsoc2020] Answer to histogram design questions from Slack

From: Mateusz Loskot (mateusz_at_[hidden])
Date: 2020-05-13 12:42:30

On Wed, 13 May 2020 at 14:01, Pranam Lashkari <plashkari628_at_[hidden]> wrote:
> On Tue, May 12, 2020 at 11:16 PM Mateusz Loskot <mateusz_at_[hidden]> wrote:
> > On Tue, 12 May 2020 at 15:05, Mateusz Loskot <mateusz_at_[hidden]> wrote:
> >
> > If I want a histogram of grey levels, I should not be forced to deal with
> > any multi-dimensional structure, any matrix.
> > I should be able to get a histogram in as simple form as
> > std::array<int, 256> or std::vector<int> or similar,
> > that is objects with minimal necessary interfaces
> > and optimal memory management
> > without unrelated 2-dimensional bloat.
>
>
> Currently, looking at the other image processing algorithms in the GIL,
> most of them have an implementation for a single channel(i.e: grayscale)
> and then this is extended to work with multi-channel by iterating over the
> channels. Right now, we have so many questions, so why not to do the same
> with histogram too? will make things simpler and also as we'd have some
> basic implementation we'll know where to take things further.

Yes, indeed.

By asking some questions I did not suggest they should be answered now.
I just hoped to direct the thinking/discussion towards quite obvious conclusion:
- do not focus on unnecessary stuff (e.g. controlling details of of histogram)
- focus on what we know to start prototyping/implementing early
- start from easy to complex

Obviously, solving a problem for a single channel is easier than
solving it for multiple channels and dimensions.

Best regards,
-- 
Mateusz Loskot, http://mateusz.loskot.net