4
$\begingroup$

I'm currently trying to develop an algorithm to control the camera characteristic of a camera (the camera characteristic can be controlled to adopt it to the different lightning situation which can occur)

In this case I need a good measurement of the exposure/lightning on the current image taken (i.e., if it's under/overexposed, well-exposed, maybe low/high contrast etc). And I'm therefore looking for a measurement on how to decide that. So far I couldn't find much about it. I found only ONE paper:

It's called "An optimal measure for camera focus and exposure" (DOI) from 2004 which describes some measurements like mean-values, entropy, standard deviation etc. based on the histogram to somehow measure the image quality (concerning focus and exposure). I also found many other papers which try to measure the image quality concerning things like blur, distortion etc etc but I don't need that obviously as I only want to control the camera characteristic to get a good contrast and exposure of my image -> thus, a well-balanced histogram.

Do you know any other measures which could help me with that or do you maybe have another idea on how solve that issue and how to build a good controller of my camera characteristic?

Improve this question
$\endgroup$
2
  • 2
    $\begingroup$ Can you provide more info an what you can measure (I assume capturing an image is your measurement) and what you're able to control (focus, aperture, perhaps some physical characteristics of the sensor)? $\endgroup$
    – Phonon
    Commented Nov 2, 2013 at 22:50
  • $\begingroup$ I want to control the camera characteristics curve (in this case it's a combination of linear and logarithmic response, called linlog camera: photonfocus.com/html/eng/cmos/linlog.php? so that it has the best contrast in the picture. And therefore I need some sort of measurement to judge how well the contrast is and how it's developing when I'm changing the camera characteristic. And that's why I'm looking for such a measurement, which could be for example the entropy, mean-value of the image histogram etc etc but I don't know which one would be the best. Any ideas? Thank you a lot! $\endgroup$
    – tim
    Commented Nov 4, 2013 at 9:16

1 Answer 1

Reset to default
1
$\begingroup$

I understand that what you are looking for is a contrast measure. Since, I am not sure as to which different research papers you have looked at, I am giving here links to some important and good papers in this area. Refer to them. If you are unable to make sense of them, reply back, and I would try my best to help

Measuring pereptual contrast in color images

Measuring the perceived contrast of natural images

Hope this helps

Improve this answer
$\endgroup$
2
  • $\begingroup$ eri.harvard.edu/faculty/peli/papers/ContrastJOSA.pdf $\endgroup$
    – Ujjwal Aryan
    Commented Nov 5, 2013 at 12:52
  • $\begingroup$ Yeah something like that, sounds nice, thank you so far! I will take a look at them and see if they are relativly easy to implement and running quick as a measure in a realtime application. $\endgroup$
    – tim
    Commented Nov 5, 2013 at 17:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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