Facial Recognition Technologies in Security Access Systems

Biometric identity is a type of permanent and unique user identifier used to keep user authentication secure (Kumar, Yadav, Saini, 2023: 2772–6711). The process commonly known as facial recognition actually takes place in two stages.

The first stage is face detection, which involves locating faces within a scene. Face detection is, in fact, the challenging part of the task — it is the process of removing the background (everything that is not a face) and isolating a clean face image.

The second stage is face recognition or face verification, where the system determines whose face it is. In short, the question that needs to be answered in the face detection phase is:

“Is there a face, and if so, where is it?” whereas in face recognition or verification, the question becomes: “Whose face is this?” (Konak, 2006:11)

Facial recognition methods have been developed primarily to meet personal and institutional security needs, although they are now used in many areas.

This method, frequently used in secure access control systems, offers highly practical usage.
In traditional access systems, carrying physical items such as keys or cards, and the risk of them being lost or stolen, make security management more difficult.

Likewise, systems such as fingerprint, retina scan, or palm recognition require physical contact, which makes face recognition a more advantageous alternative.

Facial recognition consists of detecting a face in an image and comparing it with the previously registered data in the system to verify whether it matches. If the verification succeeds, access is granted and the secure passage process is completed (Kaplan, 2018:1–3).

Common Application Areas of Facial Recognition Systems

• Military and Law Enforcement
• Border Crossings and Airports
• Banks and Financial Institutions
• Public Buildings and Hospitals
• Corporate Offices and Smart Building Systems

Facial recognition techniques have evolved significantly over the years.
Traditional methods relied on hand-crafted features such as edges and texture descriptors, combined with machine learning techniques like Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), or Support Vector Machines (SVM) (Shokouh, 2013:8).

Recently, however, these traditional methods have been replaced by deep learning models based on Convolutional Neural Networks (CNNs). The main advantage of deep learning methods is their ability to be trained on large datasets to learn the best feature representations for the data.

System Architecture of Access Control with Facial Recognition

Access control systems generally consist of:

• A camera that captures the image,
• A database storing the enrolled facial data,
• A computer that processes both reference and test images, and
• The user, whose image is captured.

The process includes face detection, feature extraction, and face recognition stages.
For face detection, methods like Viola–Jones (Viola, Jones, 2004:137–154) are widely used.
For recognition, both traditional methods such as PCA (Mamak, Konyar, Solak, 2020:497–504) and modern models such as CNNs can be utilized. A dataset of facial images is then employed to test and validate these models.

Why Are Facial Recognition Systems Preferred in Access Control?

Biometric applications are now used in many fields, and access control systems are among the most common. Although these systems serve various needs, their fundamental purpose is to ensure security by identifying individuals accurately. Access control systems are widely used in banking, defense, and communications, among other sectors.

Unique and immutable biometric traits such as face, fingerprint, palm, retina, and voice are considered essential sources for ensuring security. Among these, facial recognition stands out due to its contactless nature and easy applicability in many systems.

However, facial recognition systems still face several challenges, including:

• Variations in lighting conditions,
• Different facial poses,
• Emotional expressions and facial gestures,
• Aging-related changes,
• Obstructions such as beards, mustaches, glasses, or hats.

Phases of Facial Recognition

Facial recognition generally consists of three main phases: face detection, feature extraction, and face recognition.

Face Detection

This is the process of identifying human faces in an image. At this stage, the presence of a face is determined, and its location is framed within certain boundaries. Variations in lighting and facial expressions may hinder accurate detection.

To make facial recognition more robust, preprocessing steps are performed — methods such as Viola–Jones detector, Histogram of Oriented Gradients (HOG), and Principal Component Analysis (PCA) are commonly used. This step can also be adapted for detecting other types of objects.

Feature Extraction

The main task in this step is to geometrically analyze the distinctive facial features (such as the eyes, nose, mouth, and eyebrows) and their relationships. Every face is characterized by its unique structure, size, and shape.

Various techniques extract these details using dimensions and spatial relationships: HOG, Eigenfaces, Independent Component Analysis (ICA), Linear Discriminant Analysis (LDA), Scale-Invariant Feature Transform (SIFT), Gabor filters, Local Phase Quantization (LPQ), Haar wavelets, Fourier transforms, and Local Binary Patterns (LBP) are among the main algorithms used for feature extraction.

Face Recognition

This phase considers the extracted features and compares them to the stored data.
Facial recognition systems have two main functions:

• Identification — determining whose face it is,
• Verification — confirming that the face matches the claimed identity.

These processes aim to find the most probable match and decide whether to accept or reject the input face.

Main Framework of a Facial Recognition System

Data Acquisition

The system begins with acquiring data — capturing a facial image that will be used in subsequent stages for comparison.

Data Preprocessing

Before the feature extraction phase, face images are normalized to improve recognition accuracy.
This stage includes resizing, contrast adjustment, noise filtering, background removal, and rotation or translation normalization.

Feature Extraction

After preprocessing, the essential features of the normalized image are extracted.
The optimal feature vector obtained here is then prepared for the classification phase.

Classification

The extracted features of the face image are compared with those in the database, and the result is classified as known or unknown.

Personal Image Data

Facial images or feature vectors are stored in a personal database for recognition purposes.
However, this method may raise privacy concerns, as biometric data can be sensitive.

Therefore, solutions that do not store biometric data on devices or connected servers — but allow users to retain their biometric data under their own control — are increasingly in demand.

Conclusion

Facial recognition systems play a key role in enabling individual and organizational security to adapt to the digital era. With the high-quality, technology-driven solutions offered by Ones Technology, these processes become more reliable, faster, and user-friendly.

The core of the Zero Trust philosophy — “never trust, always verify” — is built upon advanced biometric authentication technologies, which elevate access control infrastructures to a new level of security.

As identity, password, and data breaches continue to rise globally, this principle has become the new mantra of security.

Within this context, 3D facial recognition and biometric identity verification technologies are becoming indispensable market leaders.

Furthermore, KVKK and GDPR-compliant solutions place technology on a more human-centric foundation, enhancing privacy protection and user trust.oturtmakta ve kullanıcı açısından tercih nedeni yapmaktadır.

Sources

Kumar, A. Yadav, R. K. Saini, D. J. B. (2023). Create and Implement a New Method for Robust Video Face Recognition using Convolutional Neural Network Algorithm, Advances in Electrical Engineering, Electronics and Energy 5: 2772-6711.

Konak, E. S. (2006). Bilgisayar Destekli Yüz Tanıma Sistemi Tasarımı: 11.

Kaplan, A. (2018). Gerçek ve Yarı Gerçek Zamanlı Yüz Tespit Etme: 1-3.

Shokouh, G. S. (2013). Gerçek Zamanlı Sayısal Görüntü İşleme ve Örüntü Tanıma Tekniklerinin Araştırılması ve Uygulanması: 8.

Çavdarcı, F. (2024). Gerçek Zamanlı Yüz Tanıma Sistemi Tabanlı Güvenlik Geçiş Sistemi Tasarımı http://www.softwebsolutions.com/resources/AI-based-face-recognition.html