Hello Sharif,

you are correct there are many color spaces but from knowledge point of you i would like to suggest you better for working with CIELAB, because i have used this in color image processing and also for linear and non linear devices. here in this by using only 2d i.e., with x and y value we can find out z also like x+y+z=1 and z= 1-x-y.like this we can get the tristimulus values and it is very accurate also ok thank you.

Hello Sharif,

you are correct there are many color spaces but from knowledge point of you i would like to suggest you better for working with CIELAB, because i have used this in color image processing and also for linear and non linear devices. here in this by using only 2d i.e., with x and y value we can find out z also like x+y+z=1 and z= 1-x-y.like this we can get the tristimulus values and it is very accurate also ok thank you.

Dear Muhammad,

as I see it, the answer depends a great deal on the operating conditions.

The most important distinction is between device-dependent (e.g.: RGB, HSV, etc.) and device-independent (e.g.: CIE-Lab, CIE-Luv, etc.) colour spaces. If you have to compare images acquired with different cameras and illumination conditions, you will need a device-independent space. Obtaining device-independent colour data usually requires calibration (colour calibration). Otherwise, if you are planning to work with the same camera and under invariable illumination conditions, a device-dependent space will be ok. In this case RGB is usually good enough – in my experience.

You can find a discussion on this topic in the following paper (see Secs. 4.2 and 4.3):

F. Bianconi, A. Fernández, E. González and S.A. Saetta; “Performance analysis of colour descriptors for parquet sorting”; Expert Systems with Applications, 40(5), 1636-1644, 2013

Related Matlab code can is available at the following URL:

http://dismac.dii.unipg.it/parquet/

Hope it helps.

Best,

f.