基于模型的人脸图像配准

Models-based Facial Image Alignment

为了提高活动形状模型算法在图像配准中的鲁棒性,设计了两种基于Fisher变换的局部外观模型,并基于改进的模型提出了一种人脸图像配准算法。新算法结合了改进的模型和传统外观模型,可以较好地抵抗噪声干扰。

In order to improve robustness of active shape models for image alignment, two local appearance models based on Fisher transform and a new face image alignment algorithm based on those are proposed. The new algorithm combines the improved models with traditional appearance model, thus it is robust to the interference of noise edges.