A novel complete face detection method in color images
Minghua Zhao, Peng Li, Li Wang · 2010 3rd International Congress on Image and Signal Processing · 2010
A novel complete face detection method in color images based on facial characteristics is proposed in this paper. Firstly, the image to be detected is illumination-compensated and transferred from RGB to YCgCr color space. Secondly, Gaussian skin color model is established with 179221 skin pixels and skin color similarity image is obtained. Thirdly, a fast 2D Otsu method is used into skin color segmentation. Fourthly, the skin color segmentation results are processed by erosion operation, dilation operation, 8 neighborhoods connected component analysis, some limited conditions and Euler number. Finally, the eye information based on direction of the facial symmetry axis is considered and the final face region is determined.