针对三维人脸识别算法中的高精度分类器设计问题,采用人脸全局特征和局部特征共四个相互独立的多特征信息分类后进行D-S数据融合技术来实现。通过SVM分类器对三维人脸图像中相互独立的全局特征(面廓)和局部特征(眼睛、鼻子和嘴)共四个特征进行一对一的单特征识别,并将其结果进行数据归一化处理后,作为D-S证据理论的BPA,按照D-S理论融合全局特征和局部特征数据,计算出更加准确的识别结果。经过融合数据结果分析,发现该算法可靠有效,大大提高了三维人脸的识别效率。%In view of the high accuracy classification design problem of 3D face recognition algorithm,use face global features and local features,a total of four independent feature information classification before D-S data fusion technology to realize. The global features (surface profile) and local features (eyes,nose and mouth),a total of four characteristics in the 3D face images for one-on-one single feature recognition is conducted by SVM classier,and the results are normalized data fusion as BPA of D-S evidence theory,calculating the more accurate identification results. The fusion result analysis shows that this algorithm is reliable and effective,greatly improving the recognition efficiency of 3D face.
展开▼
