Analysis of facial recognition techniques. (2022)

• Record Type: Journal Article
• Authors: Sivapriyan, R.; Pavan Kumar, N.; Suresh, H.L.
• Is Part Of: Materials today. Volume 57:Part 5(2022)
• Page Start: 2350
• Page End: 2354
• Publication Date: 2022
• Subjects: Facial recognition -- Fiducial mapping -- Database matching -- Statistical approach -- Contour mapping -- Feature mapping; Materials science -- Congresses -- Periodicals; 620.1
• DOI: 10.1016/j.matpr.2022.01.296
• Languages: English
• ISSNs: 2214-7853

Abstract

Abstract: Facial recognition has become a hot spot for researchers this past few decades due to its relevance in privacy and security application. The various facial recognition techniques can be catalogued into Database Matching, Statistical Approach, Contour Mapping, Fiducial Mapping, and Feature Mapping. This paper aims to compile the various techniques proposed into the five subcategories mentioned above for facile discrimination. Each technique involves training a classifier with a large number of facial images, applying statistical algorithms to the pixel matrix of the facial images, capturing the distinct variations in the contour of the facial image, tracing distance between points that stand out and differentiate the person and the use of a neural network to extract feature that best represents a particular face. The final object of the paper is to compile features and the drawback of each of the five categories of facial recognition, as there is does not exists a singular facial recognition technology that can perform facial recognition under various external factors such as poor lighting, different facial condition, and distortion in facial images. The compiled set of features for each technique will help the reader decide the best techniques for a given condition in which facial recognition has to be performed.