ABSTRACT Exemplary embodiments of the present disclosure are directed towards a face recognition system using Doubly Truncated multivariate Gaussian Mixture Model with DCT coefficients under logarithm domain. The system comprises a training phase unit configured to provide at least one of: face images with a database; DCT coefficients under logarithm domain; initialization of parameters through K-means/Hierarchical clustering algorithm; and refine the model parameters by EM algorithm. A testing phase unit configured to provide DCT coefficients under logarithm domain in feature vector with classification, a model building unit configured to build each face image and background for creating an image or face model, a comparing unit configured to find log likelihood difference between face model and generic model, a log likelihood inspection unit configured to check the log likelihood with the threshold level and a matching unit configured to match the found image with the maximum probability for recognizing face.
展开▼