Article · Advances in Knowledge-Based Systems Data Science and Cybersecurity · January 1, 2026
Face Image Recognition and Normalization Using Artificial Intelligence
Islambek Saymanov, Muhammadiyev Abbos, Maxmatqulov Tuychi
Abstract
In this paper, we look at a common problem in face recognition systems — when part of the face is covered by things like masks, glasses or scarves. The primary contribution is the development of a novel Reference Sample Adaptation (RSA)-based normalization technique designed to improve recognition performance when facial images contain significant occlusions. Our main goal is to build a system that can recognize people in real time, even when a big part of their face is covered. We also look at the problems of both old face recognition methods and newer AI based ones, specially when they have to deal with partly visible faces. To overcome these limitations and fill the existing research gap in occlusion-robust normalization, we propose an RSA-based normalization method. Our method first sort the face images into groups, then create reference images from clean photos of each person. After that, it normalizes the covered face image before sending it to the recognition stage. The proposed pipeline integrates triangle-based feature detection, pixel-level projections, and Viola–Jones features for robust landmark localization and obstruction removal, followed by deep learning models to extract highly discriminative features. This hybrid methodology advances beyond many state-of-the-art occlusion-handling techniques by combining classical preprocessing with deep feature extraction, while explicitly addressing the shortcomings of end-to-end deep detectors in highly occluded scenarios. Our tests showed that RSA normalization made the accuracy better and reduced FAR and FRR compared to other methods we tested against. Further validation against contemporary benchmarks, including margin-based losses such as ArcFace and recent masked/occlusionrobust approaches published after 2020, confirms the effectiveness of the method. The system we built can work well in real time and can be used in places like door access, security cameras and other situations where faces are often covered.
More by Islambek Saymanov
- Compressed-Domain Temporal Cues and Degradation-Aware Distillation for Lightweight Dynamic Facial Expression Recognition on Edge DevicesTechnologies · 2026
- Action-Unit-Constrained Graph Learning with Counterfactual Recourse for Auditable Facial Expression RecognitionComputers · 2026
- Three-Class Correspondence Reliability for Deformation-Preserving ICP: Controlled Synthetic Evaluation and a Cross-Domain Stress TestJournal of Imaging · 2026
- An EMC-Aware Multi-Objective Framework for Joint Radio-Resource and Transmit-Power Coordination in Ultra-Dense 5G/Beyond-5G NetworksJournal of Sensor and Actuator Networks · 2026
