Face detection and extraction predominant hue ranges

1- Detect Face using HaarClassifiers (keep the biggest hit in the image)
2- Create an ellipse surrounding the rectangle with the detected face
3- Create a binary mask: value 1 inside ellipse, value 0 outside ellipse
4- Create and calculate an Histogram using the binary mask and the image with the region of interest (ROI) established in the rectangle area (result of face detection).
5- Threshold the Histogram in order to reduce the noise
6- Normalize the Histogram


The Histogram shows the predominant HUE ranges. The same process can be done with SATURATION and HUE,so we obtain the predominant HSV ranges and we can use them , for example, to apply color segmentation algorithms.

This process can be also done using RGB color representation.

Despite of the diferent light conditions and skin colors, the predominant hue ranges are always near reds, so we could use a color segmentation algorithm in order to make a faster face detection.