Face features comparison for real and pencil portrait using moments invariants

Abstract EN

In this paper, comparisons for the detected features of human face real photo and the human face pencil portrait is done to evaluate the features properties.

Moments based features are used in the comparisons so as to produce the seven invariant values for the features of the faces.

These moments are invariant to translation, scale, contrast and rotation in the faces presented as a real photo and a pencil portrait and both of these photos are represented as digital gray scale images.

Face's features in both images are detected by the user, cropped and sent to the implemented method so as to compute the seven moments for these parts of the faces' images, and then determining the range of differences between the moments resulted of the two face images features so as to determine that the two images are for the same person.

The results showed that the differences in moments values computed for the symmetrical features in the two photos (the real photo and the pencil portrait) lie in a limited range, and the moments values are invariant according to view, scale and contrast variations of the two photos.