10
$\begingroup$

i've started experimenting with c++ and opencv because of i want to learn image processing.

Now, my first exercise is to create a skin detector with calcHist and calcBackProject.

But i don't understand few things:

  • the statistic interpretation of back project, and why it is named "back project"
  • i have quite good comprehension of what ranges parameter is in calcBackProject. But i'm really stucked with ranges parameter in calcHist function.
  • For more detection precision i think could be a good thing use backgrojection in more levels: in each channel of r-g-b ans in each channel of h-s-v. But i don't know how i can combine the different results of calcBackProject of separate channels of rgb and hsv.

And i think that my not-well comprehension is caused by lack in theory of what i'm doing with those 2 methods (see the first point). So please explain me in plain english.

Improve this question
$\endgroup$
1
  • $\begingroup$ I am also very interested in this. Shoot me an email and we can talk about it some more. $\endgroup$ – Spacey Mar 11 '12 at 19:28
3
$\begingroup$

See What is Back Projection in the openCV tutorials

An image histogram measures the distribution of colour (and brightness) of the pixels in an image.

If you take an image and identify a region of interest eg. a hand, and calculate the histogram of the pixels in that object.

Then take that histogram and a second image and essentially reverse the process - you pick the pixels in the second image that match the histogram from the first. It's this reverse process that gives it the name back-projection.

You then make the assumption that areas of the image in the second image that have the same colour distribution as an object in the first image are an image of the same (or similar) object.

Improve this answer
$\endgroup$
3
  • $\begingroup$ Answers should contain an answer, not just a link. $\endgroup$ – endolith Mar 12 '12 at 21:10
  • $\begingroup$ @endolith - true, but is there much point in my writing a long explanation on how to use back projection in opencv when they have a tutorial about it? $\endgroup$ – Martin Beckett Mar 12 '12 at 21:12
  • $\begingroup$ "By all means link to another website, but you should at least summarize the contents of the page you link to." meta.stackexchange.com/a/13370/130885 $\endgroup$ – endolith Mar 12 '12 at 21:17

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.