Ordinary preserving manifold analysis for human age estimation

We propose in this paper a novel ordinary preserving manifold analysis approach for human age estimation using face and gait features. Motivated by the fact that high-dimensional human facial images and gait sequences may reside in low-dimensional aging manifolds and two samples of face images or gait sequences with distinct age difference can provide different discriminative information for devising the low-dimensional aging manifold, we project the high-dimensional face or gait samples into a low-dimensional submanifold such that the samples with similar age values (i.e., smaller age difference) are projected to be as close as possible while those with dissimilar age values (i.e., larger age difference), as far as possible. To uncover the relation of the projected features and the ground-truth age values, we learn a multiple linear regression function with a quadratic model for age estimation. Experimental results on the MORPH face database and the USF gait database are presented to demonstrate the efficacy of our proposed approach.