Recognizing images across age progressions: A comprehensive review

With increase in age, there are changes in skeletal structure, muscles mass, and body fat. For recognizing faces with age variations, researches have generally focused on the skeletal structure and body mass. We incorporate weight information to improve the performance of face recognition with age variation which utilizes neural network and random decision forest to encode age variation. This age invariant recognition technique is classified into two categories-Discriminative based techniques and Generative based technique. Generative models are typically more flexible than discriminative models in expressing dependencies in complex learning tasks. We have analyzed various approaches available in Discriminative and Generative based techniques. Standard face public databases such as LFW, YTF, FGNET, MORPH, WhoIsIt, Gallagher, Pubfig83, Twin days dataset are considered for analysis. From our analysis we have found that the Neural Network and random decision forest based classification technique has 28.532±1.03% accuracy for WHOISIT database in Discriminative based technique and SVM method has 92.8±.28% accuracy for Pubfig83 database.