A comparative assessment of three approaches to pixel-level human skin-detection

Abstract

This paper assesses the merits of three diflerent approaches t o pixel-level h u m a n skin detection. T h e basisfor the 3 approaches has been reported recently in the literature. T h e first two approaches [1, 21 use simple ratios and colour space transforms respectively, whereas the third is a numerically eficient approach based o n a 3-D RGB probability map, first implemented by Rehg [3]. T h e Bayesian probabilities are made possible t o compute only with the availability of a large appropriately labeled database. Over 12,000 images f r o m the Compaq skin and nonskin databases [4] are used t o quantitatively assess the three approaches. Thresholds are determined empiricall y t o detect 95% of all skin-associated pixels and assessment is then made in terms of the percentage of non-skin pixels incorrectly accepted. T h e lowest of these false acceptance rates is found t o be about 20% given by the 3-D probability map.

Color based skin classification

“…It has been shown that RGB, YCbCr and HSV have similar performance while CbCr is different since being a 'non-invertible' color transformation. The skin probability Map (SPM) with RGB color space is reported as the best combination in [3], compared to Red-Green (rg) ratio and linear color transformation of RGB color space into YIQ color space.…”

Measuring facial grimacing for quantifying patient agitation in critical care

“…Specifically, the R/G ratio of the intensity value on the red (R) channel to the intensity on the green (G) channel can be used to represent skin [32,33]. Using the position of the dot to get a sample of the skin hue on the forehead, it is thus possible to detect skin on the face of a specific patient.…”

“…Using the position of the dot to get a sample of the skin hue on the forehead, it is thus possible to detect skin on the face of a specific patient. Note that the studies of [32,33] successfully detect skin on a very large data set containing a wide number of skin types. In critical care patients there is the possibility of paler skin due to their severe illness.…”

Perceptive three dimensional interface via stereo observation

“…The key idea of this method is to estimate skin colour distribution from the training data without having to derive an explicit model of the skin colour. This method is also referred as a SPM (Skin Probability Map) (Brand & Mason 2000, Gomez 2000. By using this method, a skin colour database was established as a training data.…”