An Novel Dynamical Multiscale ASM for Facial Feature Location

Aiming at the problem in facial feature location, the Active Shape Models algorithm has no evaluable criterion to judge its convergence, and tend to run into local minimum when using the fixed search scale, and how to choose the scale is always an issue. This paper presents a method based on the whole shape gray grads to evaluate the search effects, and an improved search policy of multiscale. The experiment results show that, this method can enhance the robustness of ASM feature location effectively and reduce the computational complexity.