Illumination normalization techniques for makeup-invariant face recognition. (January 2021)

• Record Type: Journal Article
• Authors: Saeed, Usman; Masood, Khalid; Dawood, Hussain
• Is Part Of: Computers & electrical engineering. Volume 89(2021)
• Publication Date: 2021-01
• Subjects: Face recognition -- Illumination normalization -- Facial makeup -- WeberFaces; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2020.106921
• Languages: English
• ISSNs: 0045-7906

Abstract

Highlights: Illumination normalization can be used to reduce variation in facial makeup images. WeberFaces is the most performant normalization method for facial makeup images. LGBP is a robust method for representing images with illumination/makeup variation. Abstract: Face recognition research in unconstrained environments has focused mainly on the three classical causes of variation, i.e., Pose, Illumination, and Expression (PIE). Another recently identified issue is facial makeup that poses a significant challenge to face recognition systems. We believe that applying makeup can cause changes similar to the ones observed due to illumination variation. Therefore, in this paper, we have investigated the effectiveness of illumination normalization techniques for decreasing the variations caused by makeup in a face recognition system. First, we apply photometric illumination normalization techniques with their parameters adjusted for face recognition. Next, we extract facial features using texture-based feature extraction methods and perform face recognition using Support Vector Machines. Experiments carried out on both constrained and unconstrained databases clearly show that illumination normalization techniques improve face recognition results. Graphical abstract: Image, graphical abstract