A Method for Head—shoulder Segmentation and Human Facial Feature Positioning
HuTianjian, CaiDejun · 通信学报 · 1998
This paper proposes a method of head-shoulder segmentation and human facial feature allocation for videotelephone application.Utilizing the characteristic of multi-resolution processing of human eyes,analyzing the edge information of only a single frame in different frequency bands,this method can automatically perform head-shoulder segmentation and locate the facial feature regiongs(eyes,mouth,etc.)with rather high precision,simple and fast computation.Therefore,this method makes the 3-D model automatic adaptation and 3-D motion estimation possible.However,this method may fail while processing practical images with a complex background.Then it is preferable to use some pre-known information and multi-frame joint processing.