Invariant Feature Extraction for Component-based Facial Recognition

This paper proposes an alternative invariant feature extraction technique for facial recognition using facial compo-nents. Can facial recognition over age progression be improved by analyzing individual facial components? The individual facial components: eyes, mouth, nose, are extracted using face landmark points. The Histogram of Gradient (HOG) and Local Binary Pattern (LBP) features are extracted from the individually de-tected facial components, followed by random subspace principal component analysis and cosine distance. One of the preprocessing steps implemented is the facial image alignment using angle of inclination. The experimental results show that facial recognition over age progression can be improved by analyzing individual facial components. The entire facial image can change over time, but appearance of some individual facial components is invariant.

@article{Hassan2020,
  title     = {Invariant Feature Extraction for Component-based Facial Recognition},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {11},
  number    = {3},
  year      = {2020},
  publisher = {The Science and Information Organization},
  author    = {Adam Hassan and Serestina Viriri},
  doi       = {10.14569/IJACSA.2020.0110386},
  url       = {https://doi.org/10.14569/IJACSA.2020.0110386}
}