Human Skin Range??

Since YCBCR method of skin recognition gives poor results,Can anybody suggest me a simple code to detect skin regions by using HSV ? Thanks in advance

[Jiro Doke]

I don't think this questions is related to MATLAB or any programming tools.

[Niranjan]

Why not? I am using MATLAB image processing toolbox for my work.

[Jiro Doke]

I see you changed your question after my comment. So yes, it's now more of a MATLAB question, but it seems very open-ended. What have you tried? Have you looked at the documentation for "Converting Color Data Between Color Spaces" ( http://www.mathworks.com/access/helpdesk/help/toolbox/images/f8-20792.html )? You may also want to look at sections on "Morphological Operations" ( http://www.mathworks.com/access/helpdesk/help/toolbox/images/f18-9530.html ) and "Analyzing and Enhancing Image" ( http://www.mathworks.com/access/helpdesk/help/toolbox/images/f11-9579.html ).

[John D'Errico]

Flesh tones live in a fairly nastily shaped part of color spaces. From what I recall in Lab, it is a curvy banana, especially if you sample all races adequately. Even at that, there are issues of what exactly is a flesh tone. At what point does badly sunburned skin no longer qualify as a traditional flesh tone?

Answer 1

[Walter Roberson]

Using YCrCb for skin detection is not recommended. Please examine this summary presentation, in which it is shown to be the poorest choice.

[DGM]

Link revival:

https://web.archive.org/web/20160418104914/http://www.svcl.ucsd.edu/~nikux/skin/Color%20Space%20for%20skin%20segmentation.ppt

Answer 2

[Walter Roberson]

Note that Niranjan did not ask about "flesh tones", but rather about skin recognition. Niranjan thus implicitly wishes to be able to recognize even skin that has been "body-painted"