A novel approach for detecting facial image spoofing using local ternary pattern

Biometrie spoofing is widely understood as the ability to fool a biometric system in recognizing an illegitimate user as a genuine one. To keep a goodlevel of security, reliable spoofing detection tools are necessary. Immense growth of technology permitted the use of biometrics in diverse fields such as forensics, access control, surveillance or online commerce. The new biometric paradigm has transformed passwords and cards into human as the best key. The proposed approach considers facial texture as a feature. Further the microtextural features of the facial images are extracted using Local Ternary Pattern approach and the extracted Local Ternary patterns are converted into upper pattern and lower pattern. Further the patterns are used to generate histograms. The feature histograms are fed into SVM classifier to classify the features from which the outlier is detected i.e whether any spoofing is done with the image or it is a real image.