Abstract

Human Computer Interaction (HCL) demands the study, design, drafting evaluation and needs of the interaction between people and computers. HCI must understand the people in a group to decide the end effect respective to their nature and characteristics. To know the details from the group image, face is a major cue for analysis. Hence, a foremost step is to identify the faces of human in a picture. Detection of face strongly depends on visual properties such as color; complex background and face pose variation. To sort out these issues, a face detection algorithm using color and appearance based edge distribution is proposed which shows good tolerance to illumination, scale and complex back ground. Techniques for detection of faces in mug shots or group photo under wild conditions are proposed in this paper. The proposed method uses a Gaussian skin color model using HSV color space for skin region detection which helps for face localization. Detected skin regions are verified as whether face and non-face using edge features. Edge based feature vectors are adopted for reducing the false positive rates and false negative rates.

Keywords: Human Computer Interaction (HCI), Feature Vectors, Face detection.