hi all,
we are doing a project titled automatic currency identification system
. in that while taking photos using web camera we were not able to
illuminate all the spot with equal brightness.we are using one of RGB
as background. while converting that image in to matrix form we are
getting wide range of values for background due to brightness
variations . but we like to use the technique of eliminating
background for which we need constant value in matrix for background.
so suggest some technique to eliminate background and to overcome the
brightness problem.
thank you
do reply,
venkatesh
how to eliminate brightness variations?
Started by ●January 10, 2006
Reply by ●January 11, 20062006-01-11
Hi Venkatesh,
Using Luminance information instead of one of RGB and finding a
histogram(equalization) of the luminance values might help you in
overcoming the problem.
Venugopala Krishna M On 1/10/06, venkatesh_aim@venk... <venkatesh_aim@venk...> wrote:
>
> hi all,
> we are doing a project titled automatic currency identification system
> . in that while taking photos using web camera we were not able to
> illuminate all the spot with equal brightness.we are using one of RGB
> as background. while converting that image in to matrix form we are
> getting wide range of values for background due to brightness
> variations . but we like to use the technique of eliminating
> background for which we need constant value in matrix for background.
> so suggest some technique to eliminate background and to overcome the
> brightness problem.
> thank you
> do reply,
> venkatesh
Using Luminance information instead of one of RGB and finding a
histogram(equalization) of the luminance values might help you in
overcoming the problem.
Venugopala Krishna M On 1/10/06, venkatesh_aim@venk... <venkatesh_aim@venk...> wrote:
>
> hi all,
> we are doing a project titled automatic currency identification system
> . in that while taking photos using web camera we were not able to
> illuminate all the spot with equal brightness.we are using one of RGB
> as background. while converting that image in to matrix form we are
> getting wide range of values for background due to brightness
> variations . but we like to use the technique of eliminating
> background for which we need constant value in matrix for background.
> so suggest some technique to eliminate background and to overcome the
> brightness problem.
> thank you
> do reply,
> venkatesh
Reply by ●January 12, 20062006-01-12
Venkatesh,
Convert RGB into YCBCR or even better HSI colour
components. The brightness component is usually
contained in the Y and the I component of the YCbCr
and HSI colour spaces respectively.
You can try using colour instead of brightness to
segment your image. Here use some background whose
colour will be very different from the possible
coin-colours that you will encounter(say green or blue
or something).
Look at the histogram of the Hue(if HSI) of the
background and look at the possible values it takes.
It should be somewhat insensitive to
brightness(although not completely). Use this
information to remnove background.
Keep us updated if this helps you.
Regards
Ranjith
--- venkatesh_aim@venk... wrote:
> hi all,
> we are doing a project titled automatic currency
> identification system
> . in that while taking photos using web camera we
> were not able to
> illuminate all the spot with equal brightness.we are
> using one of RGB
> as background. while converting that image in to
> matrix form we are
> getting wide range of values for background due to
> brightness
> variations . but we like to use the technique of
> eliminating
> background for which we need constant value in
> matrix for background.
> so suggest some technique to eliminate background
> and to overcome the
> brightness problem.
> thank you
> do reply,
> venkatesh
>
Convert RGB into YCBCR or even better HSI colour
components. The brightness component is usually
contained in the Y and the I component of the YCbCr
and HSI colour spaces respectively.
You can try using colour instead of brightness to
segment your image. Here use some background whose
colour will be very different from the possible
coin-colours that you will encounter(say green or blue
or something).
Look at the histogram of the Hue(if HSI) of the
background and look at the possible values it takes.
It should be somewhat insensitive to
brightness(although not completely). Use this
information to remnove background.
Keep us updated if this helps you.
Regards
Ranjith
--- venkatesh_aim@venk... wrote:
> hi all,
> we are doing a project titled automatic currency
> identification system
> . in that while taking photos using web camera we
> were not able to
> illuminate all the spot with equal brightness.we are
> using one of RGB
> as background. while converting that image in to
> matrix form we are
> getting wide range of values for background due to
> brightness
> variations . but we like to use the technique of
> eliminating
> background for which we need constant value in
> matrix for background.
> so suggest some technique to eliminate background
> and to overcome the
> brightness problem.
> thank you
> do reply,
> venkatesh
>
