A Hybrid Approach to Human Skin Region Detection

Face recognition is important in research areas like machine vision and complex security systems. Skin region detection is a vital factor for processing in such systems. Hence the proposed paper focuses on isolating the regions of an image corresponding to human skin region through the hybrid method. This paper intends to combine the skin region detected from RGB and YCbCr color spaces image by the explicit skin color conditions and the skin label cluster identified from CIELab color space image, which is clustered by Hillclimbing segmentation with K-Means clustering algorithm. Then the resultant image is dilated by arbitrary shape and filtered by the median filter, in order to enhance the skin region and to avoid the noise respectively. The proposed method has been tested on various real images, which contain one or more human beings and the performance of skin region detection is found to be quite satisfactory.