基于体绘制思维的人脸识别算法优化研究.docVIP

基于体绘制思维的人脸识别算法优化研究 　　摘 要： 传统的人脸识别算法主要解决二维正面图像识别，如果人体姿态发生明显变化，或外界环境发生显著变化，则算法性能大大降低，无法获取准确的识别结果。体绘制算法针对三维数据场进行绘制，绘制的图像能够描述人脸的内部细节，可提高人脸识别精度。因此，提出一种基于体绘制思维的人脸识别算法，依据聚类思想对二维人脸库进行聚类，在各分类的基础上，构建人脸相似模型。通过错切变形体制算法，构建人脸体数据，实现人脸体数据坐标系统的转换以及人脸三维图像的合成，从人脸三维图像中采集人脸特征，利用相似性模型对人脸关键特征同数据库已有的特征进行匹配分析，完成人脸身份识别。实验结果说明，所提算法对于不同表情和不同光照条件下的人脸图像，都具有较高的识别率和鲁棒性。 　　关键词： 体绘制； 人脸识别； 三维图像； 相似性模型 　　中图分类号： TN911?34 文献标识码： A 文章编号： 1004?373X（2015）24?0019?04 　　Optimization research of face recognition algorithm based on volume rendering thinking 　　HUANG Xiaoping 　　（School of Electromechanical and Quality Technology Engineering， Nanning University， Nanning 530200， China） 　　Abstract： The traditional face recognition algorithm is used mainly for the two?dimensional positive image recognition. The algorithm performance will be greatly reduced when the human body posture and external environment have changed significantly， so it is unable to obtain the accurate identification results. The volume rendering algorithm conducts drawing for the three?dimensional data field， and the drawed image can describe the internal details of the face， so the accuracy of face recognition can be improved. Therefore， a face recognition algorithm based on volume rendering thinking is proposed， in which the two?dimensional face library is clustered according to the clustering thought， and the similar face models are constructed on the basis of each classification. The face volume data is constructed by the algorithm of shear deformation system to realize the coordinate system transformation of face volume data and synthesis of face three?dimensional images. The facial features are acquired in face three?dimensional image， and the similarity model is used to conduct matching analysis for the face key feature existing in the same database， then face recognition is completed. The experimental results indicate that the proposed algorithm has high recognition rate and robustness for the face images with different expr