[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: conv2 performance
From: |
Søren Hauberg |
Subject: |
Re: conv2 performance |
Date: |
Mon, 01 Mar 2010 09:07:05 -0800 |
man, 01 03 2010 kl. 09:17 -0500, skrev John Swensen:
> On Feb 28, 2010, at 10:53 AM, John W. Eaton wrote:
> >
> >
> > Maybe there is still room for improvement here. I would happily use a
> > free software library with a GPL-compatible license to implement this
> > function, but I don't know whether one is available.
> >
> > jwe
> >
>
> I have recently been doing a lot of 2D convolutions. I think the
> fastest method should not involve loops of any kind. As convolution
> in the time domain (or spatial domain when considering images) is
> multiplication in the frequency domain, the fastest method is to take
> the FFT of both image and kernel, dot-multiply them, then take the
> inverse FFT.
>From what I understand, the downside is that this approach can loose
quite a bit of precision. Here are some of the thoughts by a Matlab
engineer:
http://blogs.mathworks.com/steve/2009/11/03/the-conv-function-and-implementation-tradeoffs/
Soren
- Re: conv2 performance, (continued)
- Re: conv2 performance, Jaroslav Hajek, 2010/03/03
- Re: conv2 performance, John W. Eaton, 2010/03/03
- Re: conv2 performance, Michael D. Godfrey, 2010/03/03
- Re: conv2 performance, Jaroslav Hajek, 2010/03/04
- Re: conv2 performance, Michael Goffioul, 2010/03/04
- Re: conv2 performance, John W. Eaton, 2010/03/04
- Re: conv2 performance, Jaroslav Hajek, 2010/03/04
- Re: conv2 performance, John W. Eaton, 2010/03/04
Re: conv2 performance, Søren Hauberg, 2010/03/01
Re: conv2 performance, John Swensen, 2010/03/01
- Re: conv2 performance,
Søren Hauberg <=
- Message not available
- Re: conv2 performance, Michael D. Godfrey, 2010/03/01
conv2 performance, Lukas Reichlin, 2010/03/03