Facial feature localization based on an improved active shape model

The original active shape model (ASM) has already been applied to the areas such as image segmentation, feature points localization, and contour extraction. However, the original ASM suffers from the loss of accuracy and low speed in real time applications. Due to this, a new scheme of active shape model for facial feature extraction is proposed in this paper. In this scheme, the improvement of the performance of the original ASM concerns the following three aspects. Firstly, the profile of the original ASM is extended from 1D to 2D. Secondly, each profile related to different features are constructed separately. Thirdly, the length of the profile varies with different levels. The simulations are carried out using the SJTU dataset, which contains 2273 face images. Experimental results demonstrate that the proposed scheme exhibits better performance than the original ASM.