There were some shortcomings for traditional Gabor wavelet,such as high feature dimension and too much time consumption.A new way to use Gabor wavelet was presented.First,the magnitude of Gabor wavelet was directly multiplied by a face image so as to get Gabor images.Second,texture images were extracted by using LBP.Third, histogram information was obtained from the texture images adopted as feature of the face image.Last,SVM was used to classify face images.A result of 95.0 % recognition rate and 0.14 s time consumption per image was achieved in the unpreprocessed ORL database.This showed that the algorithm could satisfy practical use
参考文献/References:
-
相似文献/References:
[1]吴继鹏,蔡国榕,陈水利,等.基于FSLBP特征的人脸活体检测方法[J].集美大学学报(自然科学版),2017,22(5):65. WU Jipeng,CAI Guorong,CHEN Shuili,et al.Face AntiSpoofing Based on FSLBP Features[J].Journal of Jimei University,2017,22(4):65.