A framework for image classification for an efficient dictionary based face recognition system

The digital multimedia image processing and its functionality of Face recognition have become more flexible. In a variety of digital forensic applications such as source identification, content analysis and authentication, it is necessary to recognize the source with training set. The proposed red length based image classification algorithm is to perform face recognition based on class specific dictionaries and to train the samples from / to each class. The class specific dictionaries are formed with some fixed number of color components. Then compares the test image with the threshold specific class and extract the minimal training set stored in the dictionary. The time taken for the image comparison leads to efficient dictionary based facial recognition which can be then used as proof of evidence towards image forensics.