Hello, is there any article/literature mention about the percentage of skin area in the detected face? In this case, i'm using Viola-Jones face detector
Hi Osman,
I think you want to get the pixel percentage of skin from an image. If you want this, just segment out the face portion along the eye and other portion. Then subtract the pixel data and calculate the percentage.
This paper may be helpful for this purpose.
Ghimire, Deepak, and Joonwhoan Lee. "A robust face detection method based on skin color and edges." Journal of Information Processing Systems 9.1 (2013): 141-156.
And if you want the skin ration for other reason like lesion ratio detection then you can read the following papers.
Roberts, Mark E., and Ela Claridge. "An artificially evolved vision system for segmenting skin lesion images." Medical Image Computing and Computer-Assisted Intervention-MICCAI 2003. Springer Berlin Heidelberg, 2003. 655-662.
Regards
Ehtesham
http://www.koreascience.or.kr/article/ArticleFullRecord.jsp?cn=E1JBB0_2013_v9n1_141
Please, read these :
http://graphicon.ru/oldgr/en/publications/text/gc2003vsa.pdf
http://ftp.aua.am/Skhachat/Public/Papers%20on%20Face%20Detection/RGB-H-CbCr%20Skin%20Colour%20Model%20for%20Human%20Face%20Detection.pdf
I think they will be helpful
Convert image into ycbcr using MATLAB function ycbcrmap = rgb2ycbcr(map) .
Now the skin pixel values would be Cb >= 77 & Cb =133 & Cr
VJ face detector is an AdaBoost object detector trained to respond to face-like patterns in gray-scale images, which outputs square shaped candidates for what it perceives to be faces. Each candidate square would typically (but not always) be centered on the face candidate so that the upper side crosses the forehead in the middle, the lower side crosses the chin approximately in the middle, and the left and right sides including the vertical face boundaries. All faces are different, of course, but one can expect that (if detected correctly) VJ face candidates would be expected to contain about 90% of skin, provided there are no occlusions, e.g. facial hair or sunglasses. It is possible to train a skin detector to estimate the %age of the skin in a particular data-set.
Thanks for the replies.
Since there is no articles mention about the skin percentages, I might do it manual by obtain the skin pixels (of face region), and get the average in all images.
I think you want to get the pixel percentage of skin from an image. If you want this, just segment out the face portion along the eye and other portion. Then subtract the pixel data and calculate the percentage.
This paper may be helpful for this purpose.
Ghimire, Deepak, and Joonwhoan Lee. "A robust face detection method based on skin color and edges." Journal of Information Processing Systems 9.1 (2013): 141-156.
Percentage of skin pixel in detected face - ResearchGate. Available from: https://www.researchgate.net/post/Percentage_of_skin_pixel_in_detected_face [accessed Apr 19, 2016].
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing