Driver face recognition by using 3D discrete cosine transform


Gunlu G., BİLGE H. Ş.

IEEE Intelligent Vehicles Symposium, İstanbul, Turkey, 13 - 15 June 2007, pp.705-706 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Volume:
  • Doi Number: 10.1109/ivs.2007.4290195
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.705-706
  • Gazi University Affiliated: Yes

Abstract

Next generation vehicles are foreseen to use biometric recognition systems for person authentication. One of them is face recognition. In this study, a different approach of feature vector generation is presented for a face recognition system. To generate the feature vectors, DCTfaces, which are very similar to Eigenfaces, are used. DCTfaces are obtained by using 3D-DCT of face images after an image synthesis and 2D-Inverse DCT of the 3D DCT coefficient layers. Taking the advantage of DCT image compression capability, very few terms are used for these DCTfaces. Experimental results show that, DCTfaces can be successfully used for feature vector generation in face recognition.