وصف البحث
Abstract. This paper’s main contribution presents designing and implementing
a face recognition system for the android mobile phone platform from the live
mobile camera. The design methodology includes two main steps. The first step is
the extraction of the face features, and the second one is the recognition according
to the classification of patterns. The face detection is carried out using a face detector available in the Android called feature-based method machine learning (ML)
Kit Software Development Kit (SDK). The face features include nose detection,
mouth detection, eyes detection, and cheek detection. These features are detected
based on their geometrical dimensions. Twenty-five geometrical raw distances
are reduced to 23 normalized distances by referring them to their face width and
height. The recognition step is done by computing the correlation coefficient ratio
between the test image’s normalized distance and all normalized distances for
authorized persons stored in the training database. The system will allow access
to the mobile applications if the correlation coefficient is greater than a chosen
threshold; otherwise, it rejects the person. The proposed approach’s recognition
rate has achieved an accuracy more than 95% for appropriate chosen threshold.
The time taken to recognize a face is approximately 13 s of other approaches.
