Skin detection using HSV color component subtraction and texture information / Rizal Mat Jusoh and Norhazimi Hamzah

This thesis presents skin detection algorithm for detecting human skin regions in color images. The input color image in RGB format is converted into HSV format for color components subtraction. The value component minus hue component is applied for first stage of skin detection. The result of the subtraction is considered as skin region candidates. From the skin region candidates rectangular box that enclosed that regions are estimated where the mean and standard deviation of that region is calculated for thresholding in RGB color space. The texture features also calculated for these skin candidates regions, where contrast, entropy and correlation are used. The texture features that are used also give good separation between skin and non skin regions. These texture features are used to verify that the segmented blobs are skin regions. The algorithm is tested on color images that concentrated on palm and face skin regions. The results from the testing show that the detection rate of skin is more than 90 percents which is achieved the set target.