Multichannel Blind Separation and Deconvolution

     Hi i m implementing IEEE paper on the topic "Multichannel Blind
Separation and Deconvolution of Images for Document Analysis".IS there any
source code available for the implementation.Can anyone please explain the
concept of "Blind Separation" and "Deconvolution of Images" and
"Independent Component Analysis".Can anyone please explain also the source
code to implement the same.......Thank you...

On Jan 31, 7:27&#4294967295;am, "mari" <rmvenkat.16@n_o_s_p_a_m.gmail.com> wrote:
> &#4294967295; &#4294967295; &#4294967295;Hi i m implementing IEEE paper on the topic "Multichannel Blind
> Separation and Deconvolution of Images for Document Analysis".IS there any
> source code available for the implementation.Can anyone please explain the
> concept of "Blind Separation" and "Deconvolution of Images" and
> "Independent Component Analysis".Can anyone please explain also the source
> code to implement the same.......Thank you...

Works via the assumption that the original sources are independent
sources (normally the case)
Unlike LMS approaches (most of them that is) there is no reference as
such but the "reference" is a probablity distribution of the
statistical properties of the unknown sources. Thyere is a ginormous
literature on this including basic tutorials and the like. The
ambiguities are
1. Determine the right order that the separated sources are in
(doesn't matter for most cases)
2. The scaling of the sources. (again doesn't matter for most cases).

Mutual information is minimised between the sources. (there are a
handful of approaches that are commonly used).

The fundamental restriction in ICA is that the independent
componentsmust be nongaussian for ICA to be possible.
Source code...there is some about but don't be lazy write your own,
it's the only way to really learn.

http://www.cs.helsinki.fi/u/ahyvarin/papers/NN00new.pdf


Hardy

Thank you for your reply....