Adaptive active appearance models

摘要: The active appearance model (AAM) is a powerful tool for modeling images of deformable objects and has been successfully used in variety alignment, tracking, recognition applications. AAM uses subspace-based models to represent the certain object class. In general, fitting such complicated previously unseen using standard optimization techniques computationally complex task because gradient matrix be numerically computed at every iteration. critical feature fast convergence scheme which assumes that fixed around optimal coefficients all images. Our work this paper starts with observation inevitably specializes region texture space, not good estimate actual as target moves away from region. Hence, we propose an adaptive algorithm linearly adapts according composition image's obtain better gradient. We show significantly outperforms basic AAM, especially are particularly challenging algorithm. terms speed accuracy, idea presented provides interesting compromise between technique recomputes iteration approach AAM.