基于DCT的人脸特征提取

Face recognition system based on the discrete cosine transform

设计一个准确性和鲁棒性均很好的人脸识别系统.该系统利用DCT来提取人脸特征,并且采用归一化技术来提高人脸对几何特征与光照的鲁棒性.并在人脸图像库中对系统进行了测试.实验表明,该系统的识别率和速度要比采用其他识别方法的系统好很多.

An accurate and robust face recognition system was proposed. The system exploited the feature capabilities of the discrete cosine transform (DCT) and invoked normalization techniques which increase its robustness to the variation in facial geometry and illumination. The method was tested on a variety of available face database. The system was shown to perform very well on recognition rates as compared to other systems..