Machine Learning Approach for Face Identification using Dimensionality Reduction Algorithm

Abstract

Introduction: Identifying faces mainly for surveillance purpose is a field of trending. Identifying faces from the image and performing recognition based on existing dataset is a challenging task. Different pose, age-based images, Blurriness, illumination are the factors which creates a challenge to perform face recognition and identification with utmost accuracy. Facial features are used to identify individual faces. Such applications if built can be used in various domains mainly including surveillance system. In past 10 years regressive research is being done in the field of computer vision to attain greater accuracy along with efficiency with less computation time. Principal Component Analysis is a statistical approach which transforms the dataset from high dimensionality to lower dimensionality keeping intact the necessary information. Eigen face is one of the most classical and hence one of the best approaches for feature extraction. Integrating PCA with CNN and SVM can lead to great results and less computation time. The study captures the details of the dimensionality reduction algorithms. A systematic comparison in presented in the paper along with the proposed model.