A Comparison of Facial Recognition Techniques

EasyChair Preprint 350

Abstract

In this paper, facial recognition has been widely studied due to its importance in many applications in the civilian and military domains. Although this computer vision problem was initially challenging due to the dynamic nature of the human face and the different poses it can take, however, the research conducted over the last two decades made huge advances with many algorithms reporting high accuracy in the published literature. However, this accuracy is usually reduced in real-life usage especially in the presence of different types of noise. In this paper, six different facial recognition algorithms are evaluated and compared, namely, principle component analysis (PCA), two-dimensional PCA (2D-PCA), linear discriminant analysis (LDA), discrete cosine transform (DCT), support Vector Machines (SVM) and independent component analysis (ICA). The effect of the presence of Gaussian and salt and Pepper noises are also considered during the evaluation of these algorithms. The results show that the best performance was obtained using the DCT algorithm with 92% dominant eigenvalues and 95.25 % accuracy which makes it the best choice under different noise conditions.

Keyphrases: DCT, ICA, LDA, PCA, SVM, face recognition

@booklet{EasyChair:350,
  author    = {Ansam Almatarneh and Mohamed Shehata},
  title     = {A Comparison of Facial Recognition Techniques},
  howpublished = {EasyChair Preprint 350},
  year      = {EasyChair, 2018}}