0
$\begingroup$

I want to use the L*a*b color model for recognition of a particular colored object by calculating the euclidean distance between the desired color in current image and the same color in next image frame.

As the LAB color space was designed in keeping with human vision, I think it can be used for the purpose. Is this correct?

$\endgroup$
  • 1
    $\begingroup$ It sounds like you have a solution looking for a problem. What sort of images / objects are you trying to recognize? What sort of lighting is it? How are the images acquired? Can you calibrate anything about the image? Can you calibrate anything about the camera? $\endgroup$ – Peter K. Nov 26 '13 at 23:15
  • $\begingroup$ Will Lab do better recognition than RGB and CMYk or not? $\endgroup$ – Amit_DSP Nov 27 '13 at 5:47
  • $\begingroup$ In RGB, the Euclidean distance between colors does not follow human intuition. Lab is better in this regard. CMYK is more a printing industry norm, it is linked to how images are printed on paper (inks or pigments, subtractive color synthesis). $\endgroup$ – sansuiso Nov 27 '13 at 7:48
  • 1
    $\begingroup$ That really depends on your application! As @lennon310 says, which colours are best will depend on the objects you are dealing with, the background they are on and the sort of camera you are using. You haven't given anywhere near enough information to try to answer this question in any different way. $\endgroup$ – Peter K. Nov 27 '13 at 13:02
3
$\begingroup$

Yes Lab color is designed to approximate human vision. But there are also RGB and CMYK color models, and their transforms like HSL and HSV. Which color base can best make your object feature distinguishable needs your experiment to validate.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.