Image Quality Measures Based Face Spoofing Detection Algorithm for Online Social Media

Recent years, selfie becomes an important part of photography and considered as the powerful and trustful communication media. Nowaday's technology has been advanced much more so any one can manipulate image with available image editing software. So it became difficult to identify whether the image is spoofed or not. Many spoof detection algorithm are being developed to find fake face in biometrics system. But face spoofing recognition still persists as a challenge in online social media. In this paper, a fast and robust algorithm is proposed to detect fake faces from photographs which have shared through the online social networks. This algorithm extracts face segments in different color spaces and estimates the illuminate map of the region. In addition, this algorithm measures the image quality parameters of segment and compares with the back ground region. Reference based and Non reference based image quality parameters are used to generate feature vectors. Quadratic Discriminant Analyzer (QDA) classifier is used to recognize the fake face. The proposed algorithm is successfully tested with DSO-1, DSI-1 dataset and the result shows 96.5% accuracy in detecting fake faces.