Implementation of Face Recognition by using of Decision Tree and PART Algorithms Based on LDA

  • Jassim Mohammed Sahan, Eyad I Abbas, Ziad M Abood
Keywords: Face detection, Face Classifier, J48, PART, Face recognition.

Abstract

Face recognition is one branch of pattern recognition that has become an important subject in modern life, as surveillance and security applications. So, face detection and recognition have gotten a lot of interest from computer vision and image processing technologies. This paper proposed a new face recognition method by use of the two classifier algorithms of J48 decision tree and PART (partial decision tree) algorithms for classifier human faces combined with linear discriminant analysis (LDA) algorithm to solve the problem of complexity of classifier of the colour images, that by use of LDA to the extraction of face feature and reduction of data dimension of the image from 2D to 1D feature vector before classifier. MUCT database consisted of 3755 colour images of person is pre-processing and extracted of features by LDA, these features are split into70% for training and 30% for testing of classifiers to the computation of performance of face recognition system. The results showed that the evaluation performance of the PART algorithm is better than the J48 decision tree for the classified face image in face recognition tasks. That the precision, recall, and F-measure of the PART algorithm are 96.99, 96.84, and 96.83 respectively, whereas J48 decision tree methods are 96.60, 96.445, and 96.44 respectively.

Published
2021-10-30
How to Cite
Ziad M Abood, J. M. S. E. I. A. (2021). Implementation of Face Recognition by using of Decision Tree and PART Algorithms Based on LDA. Design Engineering, 8754-8765. Retrieved from http://thedesignengineering.com/index.php/DE/article/view/5920
Section
Articles