Hi Ramesh, You need some information about the spectrum of the LED output. Also you really need an integrating sphere to collect ALL the light from the LED, otherwise you will have to estimate the  how much of the LED light is actually hitting your CCD.

Correct the counts per second from your CCD for its efficiency.. ie it may only count 90% of photons landing on it. Ideally then distribute the total photon counts over your LED spectrum.. or assume it is a symmetrical function.  Multiply it up to cover the total light from the LED, and then use E = hf (Energy of 1 photon (eV) = planks constant times frequency of the light) to work out the energy from it. You can probably find more detailed maths of it all if you search for "Planks constant" on the web.

Dear Ramesh,

your LED is emitting divergent light,

so for getting the photon flux density [ph/(m²*s)] at least in the direction of the detector please

  a) divide your  cts/s value by the solid angle omega set up by the detector (or the detector pixels)

  (omega =  ad/(rd)2 ; ad = detector (pixel) area  and rd = distance between LED and detector (pixel) )

and

 b) divide by  the net  LED  output  area aL.

So your      photon flux density = y / (omega*aL)    *).

Now you have your photon flux density in photons per second, per solid  angle, and per area of your LED.  

Please handle the  units  m, mm, and µm properly.

Phil is right:

please pay attention to the detection efficiency , the wave length distribution and the angular dependence when you will evaluate integral values.

*) when operating very off axis you have to consider appropriate angular functions 

Thank you.  These answers were very helpful.