Image Retrieval System by Skin Colour and Edge Information

The efficient detection of human faces in images originates from a diversity of applications requiring intelligent human-computer interaction. Since the human face is highly dynamic, finding the location and extent of an unknown number of faces in images with possibly complex backgrounds is not an easy job. A successful algorithm must be robust to numerous variables such as facial expressions, artifacts, image orientation, illumination and camera distance. Current techniques of the feature-based approaches include colour analysis, feature searching and template matching. This research focuses on the idea of using connectivity of skin and edge information to segregate faces that are close and also to separate background from faces. The advantages of the proposed approach are demonstrated on real images. The algorithm is fast and can be used in real-time applications.