尺度不变特征变换的多表情人脸识别研究

Research on Multi-expression face recognition based on SIFT algorithm

为目前,人脸识别技术已广泛应用于移动支付、进站安检等领域,但传统的人脸识别技术对光照强度、拍摄角度等因素较为敏感,同时对于人脸的表情变化准确识别率较低。针对这些问题,论文引入尺度不变特征变换算法,并根据该算法的特点分析了人脸识别技术的原理,在此基础上对该算法进行优化处理以研究基于SIFT算法的多表情人脸识别技术,根据该技术可实现对同一人不同表情的识别,也可实现不同人相同表情的识别。经过在Jaffe表情库仿真测试,验证了多表情人脸识别技术对于表情识别准确率在95%以上,具有研发价值。

At present, face recognition technology has been widely used in mobile payment, inbound security inspection and other fields, but the traditional face recognition technology is sensitive to factors such as illumination intensity, shooting angle and so on. At the same time, the accurate recognition rate of facial expression change is low. In order to solve these problems, the scale invariant feature transform algorithm is introduced in this paper, and the principle of face recognition technology is analyzed according to the characteristics of this algorithm. On this basis, the algorithm is optimized to study the multi-expression face recognition technology based on SIFT algorithm. According to this technology, the different expression recognition of the same person can be realized, and the same expression recognition of different people can also be realized. Pass by in Jaffe The multi-expression face recognition technology is proved to be more than 95% accurate by the simulation test of expression database, which is valuable for research and development of multi-expression face recognition technology.