A high-discriminative facial recognition method based on shape and grey-level appearances using landmark-points

This paper proposes a high-discriminative facial recognition method that fuses shape and grey-level features of faces. Facial landmark-points are characteristic to individuals, hence can be exploited for recognition. Spatial relationships of these landmark-points, namely their Euclidean distances between each other, are included in the feature set. Besides, the mean grey-level values calculated at the vicinity of these landmark-points are also considered as a discriminative factor and incorporated into the feature set. The results of the comprehensive simulations show the remarkable and competitive performance of the proposed method regarding recognition accuracy, as well as robustness against partial occlusion, noise, expression changes and variances in illumination.

Automated summary

This study proposes a high-discriminative facial recognition method that incorporates shape and grey-level features of faces, utilizing spatial relationships of facial landmark-points and mean grey-level values for recognition. The comprehensive simulations demonstrate the remarkable performance of the method in terms of recognition accuracy and robustness.