多姿态、光照不均人脸图像的子空间人脸识别方法研究

针对人脸识别中存在的光照不均匀问题，提出了一种预处理链技术，能达到很好的光照补偿效果。为了提高多姿态、多表情、多细节人脸图像的人脸识别率，设计了一种将最近邻分类器与支持向量机相结合的分类算法（NN-SVM），基于该分类算法提出了一种基于Gabor变换和NN-SVM的子空间人脸识别方法。在FERET和ORL两大人脸数据库中对所提方法进行性能评估，实验结果表明所提出方法能有效的解决人脸识别中光照不均匀问题，大大的提高人脸识别率，而且相比其他现存的人脸识别方法，所设计的方法具有更好、更稳定的识别效果。

Study of Subspace Face Recognition Methods for Various Posture and Uneven Illumination Facial Image

Aiming at the problems of uneven illumination in face recognition, a pro-processing chain technology is proposed which can achieve an excellent illumination compensation effect. In order to improve the face recognition rate of facial images with variations in pose, expressions and details, a classification method combination of nearest neighbor classifier and support vector machine (SVM) is designed and a robust subspace face recognition method is proposed, that is BDPCA LDA algorithm based on Gabor transform and NN-SVM classification algorithm. The performance of proposed strategy is evaluated on FERET and ORL face databases. Experimental results show that the proposed technology can solve the problems of uneven illumination in face recognition effectively and improve the face recognition rate sharply. Compared with other existing face recognition methods, the designed method has a better and more stable recognition effect.