摘要: 针对三维人脸特征提取中非特征区域干扰的问题,在分析人脸生理结构和人脸共性的基础上,提出了一种基于面分块的三维人脸特征提取方法,即先按照人脸面貌器官分布规则将人脸进行分解,然后各部分单独采取不同的方法进行特征提取,最后合并起来完整地描述人脸特征。试验结果证明该方法提高了特征提取的精确性,并具有强抗干扰能力。
关键词:
面分块,
向前差分,
区域生长,
特征提取
Abstract: Aiming at questions of interference of non-feature areas in facial feature extraction in 3D facial system, on the basis of analyzing the character of face physiological structure of human being and considering the commonness of faces, a facial feature extraction method based on surface subdivision of the human face is presented, in which, according to the organizing of apparatus, 3D face is decomposed to several parts. Each of which can be dealt with respectively, then united to describe the whole characteristic of it. Experimental results show that this method has better recognition accuracy and robust capability of anti-interference.
中图分类号:
魏佼佼,耿国华. 基于面分块的三维人脸特征提取及自动定位[J]. , 2004, 30(12): 139-140.
WEI Jiaojiao,GENG Guohua. Feature Extraction and Automatic Orientation of 3D Facial Model Based on Surface Subdivision[J]. , 2004, 30(12): 139-140.