Face verification across age progression using facial feature extraction

Face verification is a topic that is undergoing a lot of work in terms of research. As age progresses people undergo more or less the same changes across the same gender like for example growing of facial hair in men. Also these changes happen at more or less the same time for most people. In this paper an algorithm is proposed to verify the face across age progression by using features of the face such as nose, eyes etc. First a database is taken containing images of faces of people. Each person will be having atleast two images of different ages. Now the main features we use are the eyes, nose, chin, ears and lips. We now group sets of images having the same age difference i.e sets where first image is say 18 years old and the second say 30 years old. Once we form all these different sets, we take each feature extracted and find the difference. Now for each group that we have created we can find the average difference for each feature. Using this we will now have a predictor value for each feature of the face. We can then create a new face by first classifying the given image into a cluster or group using K-means algorithm and then using the average values determine the new expected face by changing the corresponding features as an estimation.