Age Estimation And Age Of The Face Of The Image-based Image Reconstruction

The changes in facial appearance of Aging faces is an active research area in face processing, such a research has both significant theoretic values and wide potential applications. It has made great progress especially in the past few years; however, evaluation results and practical experience have shown that such a research is currently far from mature. The paper has done a lot of research and experiment in the age prediction and aging face reconstruction, based on other's research in domestic and foreign, and the effect is good.The highlights and main contributions of the dissertation include:(1) A robust AAM fitting algorithm based on local texture constraints (LTC-RAAM) is presented, this algorithm has quicker convergence rate, and it mainly has strengthened the accuracy of the detail partial locating, and it has realized the robustness to the background. Comparing to the original algorithm, it's fast and accurate. The extracted face features is used for age estimating and reconstructing the aging face.(2) A new classifier that artificial immune recognition system(AIRS) based on fuzzy weighted pro-processing is presented, AIRS is a novel artificial immune supervised learning algorithm, the fuzzy weighted pro-processing and the AIRS is combined to form a new improved classifier, and the evaluation results have shown that its more accurate. It's used to estimate the age, through the classification of the face features according to the age.(3) The paper reconstructs the aging faces from face shape and face texture. A new function approximation that Boosting RBF neutral net is presented to calculate the face shape change between two different age sections, it can predict the face shape feature vector of the target age section. The paper used an improved novel texture transfer method, the improved method retained more face age information including the wrinkle, the spot and so on, so the simulation of aging face is more lifelike.The paper made a lot of experiences which have proven the accurate of the methods, we built a system to produces an age progressed or regressed image, In other words, given a single image of the subject and a target age, we construct a new image of the subject to reflect the new age, and the effect looked lifelike.