Solving Optimization Problem in Face Recognition based on Filtering and Face Position

Authors

  • Amaal Kadum, Juliet Kadum

Abstract

For the sake of improving the system of face recognition, there are different conditions that should be taken into considerations like low-resolution face images, facial emotions, and varying illumination conditions. This paper proposed a new algorithm which is invariant to the quality of the face image and conditions of illumination in some Environments. The basic operations of preprocessing are getting face edges after decrease noise from the image, especially in the sensitive system.  It based on different edge operators pre-processing. Second-order derivative filters are used to obtain scatter of face edges. To investigate this issue clearly, a range of threshold values for each edge detection filter is applied. As a new method to find face position in the image, a double k-mean process used in this algorithm to find the position of the face in an image. Principle components analysis is used to reduce data and for feature extraction. Finally, a neural network has been applied as a classifier to make a decision. BIOID databased is used in training and testing phases. based on experimental results, the presented methods have accomplished A promised results with a high level of classification accuracy

Published

2020-02-29

Issue

Section

Articles