基于二维照片进行人脸识别的算法研究

摘要

This paper put forward a method of constructing 3-d samples for face recognition method using two-dimensional pictures. Through extracting the hierarchical structure characteristics of 2-d face image and using the parameters vector, the 3-d parameters curved surface was built up and the the face automatic recognition was carried out. The experiments show that the improved method can effectively avoid the limits in 2-d image information and accurately complete face recognition.%研究图像人脸识别的问题,针对直接对照照片进行人脸识别时,受二维照片信息量的限制,以及识别时受检测人员的主观经验的影响,极易造成误识别,不能满足人脸识别准确率的要求.为了解决上述问题,提出了一种利用二维照片构建三维样本进行人脸识别的方法,通过提取二维图像中人脸的层次化特征,并利用特征向量重构人脸的参数曲面,然后在三维曲面中完成人脸的自动识别.实验表明,改进方法能够有效避免二维图像信息量的限制,以及检测人员主观经验的影响,准确完成人脸的识别,具有一定的使用价值.