镜像图与粗细层次结合的稀疏人脸识别方法研究

人脸在实际环境中,伴随着各种不可预知的情况,会呈现出复杂多变的特性.为了提高人脸识别率及更好的显示人脸特征,本文提出一种镜像图与粗细层次结合的稀疏识别新方法.该方法首先利用人脸的镜面性生成新的人脸图像,将原来的人脸训练样本和新生成的镜像图样本结合起来,使用粗细层次结合的分类方法来进行识别.新方法一方面增加了训练样本的数目,克服由于光照和姿态等外部因素带来的影响,另一方面选取合适的训练样本,丢掉不合适样本对于人脸识别所造成的不利影响.实验结果表明,新方法在人脸识别率上有了明显的提高.

Study on sparse face recognition method based on the combination of coarse-to-fine and mirror image

Along with a variety of unpredictable conditions,face in the actual environment will show a complex and changeable characteristics.In order to improve the accuracy of face recognition and better display facial feature,we propose s a new method which based on the combination of mirror image and coarse-to-fine face recognition.The method firstly used the mirror image of the face image to generate new samples,and then devised representation which based method simultaneously uses the original and new training samples to perform a sparse coarse-to-fine representation.The new method increases the number of training sam-ples,overcomes the problem of the variation of the pose and illumination of the original face image,it also uses a small number of classes what are near to the test sample to represent and classify it,and"far"from inappropriate samples that caused adverse effects in face recognition.Experimental results show that the new method has been significant improvement in the accuracy of recognition rate.