FACIAL RECOGNITION FOR PRECISE PERSON IDENTIFICATION

Siraj Shaik, Anusha Sanapathi, Sesha Venkata Amar Pulavarthi, Rishikesh Singam,
Mohammad Afrid Shaik
Vishnu Institute of Technology, Bhimavaram, Andhra Pradesh, India
([email protected] , [email protected], [email protected] ,
[email protected] , [email protected] )

Abstract expression variations. It transforms facial

images using Local Binary Patterns (LBP),
Face recognition is crucial for enhancing computes LBP histograms for texture
security by providing accurate and details, and concatenates them into a
efficient identity verification, and they global feature vector for recognition.
play a vital role in streamlining processes
like access control, authentication, and LBPH's resilience to lighting changes and
surveillance in various sectors, efficiency in capturing facial expressions
contributing to overall safety and without extensive computational
efficiency. resources make it widely applicable, from
security systems to user authentication.
The web application features three main Its implementation reflects the synergy of
buttons: Home page, Admin page, and texture analysis and pattern recognition,
Face Recognition. Admins can log in with advancing automated face recognition
credentials and access Add Person, View capabilities.
Person, and Person Logins functions. In
the Add Person page, details are keywords: Local Binary Patterns
submitted, and the webcam captures 200 Histograms (LBPH), Face recognition,
frames, storing them in the database. The Texture details, Lighting variations, Facial
View Person page trains collected data expressions, Distance metrics, Global
using the LBPH algorithm, and in the feature vector, Pattern recognition
Person Login page, recognized person techniques(OpenCV).
details can be downloaded in CSV format.
1. Introduction
Clicking on the Face Recognition button
utilizes OpenCV to identify individuals by The human face is a vital component for
comparing facial textures with trained identifying a person. Thus, face
data. The LBPH algorithm, a key recognition and identification are needed
component, excels in recognizing to identify and recognize the person.
individuals based on facial features,
demonstrating simplicity and Face recognition is a form of
effectiveness in handling lighting and identification system that is designed to
detect distinctions in face characteristics, accuracy and to improve the efficiency of
biometrics on parts of the body, or even the model.
human behavior. Systems determine
people based on their faces, which have 2. Related Work
varied forms and textures. The identified
picture may be used to find the The literature review is the most critical
information about a person's facial step in the software development process.
features. Before building the tool, it is vital to
assess the time factor, economy, and
This is used mainly at criminal capture, company strength. Once these
attendance monitoring, Employee and requirements are met, the next 10 phases
visitor door entry for business facilities, involve determining which operating
making an entry at events by advancing system and language can be utilized to
these to the next levels. construct the tool. Once the programmers
begin developing the tool, they require a
The current methods for face recognition lot of external assistance. This help can be
mostly rely on deep learning techniques, gained from senior programmers, books,
particularly Convolutional Neural or online. The aforesaid considerations
Networks (CNNs), which have been are taken into account when constructing
utilized to improve face identification the suggested system.
accuracy by learning complicated
hierarchical characteristics. These O. S. Kulkarni, S. M. Deokar, A. K.
methods require considerable training on Chaudhari, S. S. Patankar and J. V.
huge datasets, which allows the model to Kulkarni, "Real Time Face Recognition
distinguish between intricate facial Using LBP Features", 2017 International
features under a variety of settings. To Conference on Computing
ensure that the input data is uniform, Communication Control and Automation
preprocessing techniques such as face (ICCUBEA), pp. 1-5.
detection, alignment, and normalization
are performed first. LBPH algorithm for Face Recognition,
[online] Available:
The Local Binary Patterns Histograms https://iq.opengenus.orgllbph-algorithm-
(LBPH) algorithm emerges as a solution, for-face-recognitionl.
offering a unique blend of simplicity,
efficiency, and adaptability. Its ability to V. Aza, Indrabayu and I. S. Areni, "Face
accurately capture and analyze facial Recognition Using Local Binary Pattern
features under diverse conditions Histogram for Visually Impaired People",
motivates its exploration and 2019 International Seminar on
implementation. This does not require Application for Technology of
lots of training data to improve the
Information and Communication dataset comprises our facial data (i.e, real
(iSemantic), pp. 241-245, 2019. time data), serving as the foundation for
the overall work training and recognition
A. M. Jagtap, V. Kangale, K. Unune and P. capabilities.
Gosavi, "A Study of LBPH Eigenface
Fisherface and Haar-like features for Face
recognition using OpenCV", 2019
International Conference on Intelligent
Sustainable Systems (ICISS), pp. 219-224,
2019

A. Thakral and A. Vohra, "Comparison

between local binary pattern histograms
and principal component analysis
algorithm in face recognition system", Fig.1: Dataset
2017 International Conference On Smart
Technologies For Smart Nation 4. Methodology
(SmartTechCon), pp. 973-978, 2017
The approach focuses on applying the
V. Aza, Indrabayu and I.S. Areni, "Face LBPH algorithm for face recognition by
Recognition Using Local Binary Pattern gathering and storing up to 200 photos
Histogram for Visually Impaired People", for training. When someone approaches
2019 International Seminar on the camera to identify themselves, the
Application for Technology of In LBPH algorithm uses the training of the
formation and Communication photos to identify the person's face.The
(iSemantic), pp. 241-245, 2019. approach focuses on applying the LBPH
algorithm for face recognition by
3. Dataset gathering and storing up to 200 photos
for training. When someone approaches
The dataset for our face recognition work the camera to identify themselves, the
involves capturing a video of a person LBPH algorithm uses the training of the
standing in front of our webcam, photos to identify the person's face.
generating 200 frames per individual
with the dimension of 201*201 as an This face recognition algorithm is a well-
image dataset. These frames are stored in liked and effective technique for face
our database for further analysis and recognition in computer vision. Local
identification. A specific image from this binary patterns (LBP), a texture
dataset is then utilized to display descriptor, serve as the foundation for the
recognized faces and showcase the straightforward but reliable method used
successful identification process. The
by the LBPH algorithm. The following are
the main steps that comprise the Creating a Feature Vector: This process
algorithm: involves concatenating the histograms
Local Binary Pattern (LBP) Calculation: that are computed for each cell to create a
The LBPH algorithm starts by single global histogram that encompasses
transforming the face images to an the entire image. The feature vector for
intermediate representation known as the facial picture, which captures the local
the LBP image. To accomplish this, each texture properties on the face, is this
pixel is compared to its immediate global histogram.
neighbors and designated as the core
pixel. A surrounding pixel is labeled with Face Recognition: The LBPH algorithm
a 1 if its intensity exceeds or equals that uses a comparison between the input face
of the central pixel, and 0 otherwise. This image's histogram and the face
procedure creates each pixel in the LBP histograms in a training database to
image and assigns a binary number to identify faces. A basic metric, like the Chi-
each one. This number is then converted square or Euclidean distance, is
into a decimal value. frequently used for the comparison. The
recognized face is the one whose
7
histogram in the database most closely
LBP = ∑ z ( f n−f c ) 2 n resembles the input image.
n=0

Here, f n = Neighbor Pixel Value 5. Face Recognition System

f c = Center Pixel Value Architecture
z ( x ) ={0 , x< 0 1, x ≥0

Dividing the Image into Cells: Once the

LBP image is obtained, the next step is to
divide this image into several small
regions, which are often referred to as
cells or blocks. The LBP values within
each cell are used to build a histogram,
which essentially counts the frequency of
each LBP value occurring within the cell.

Histogram Calculation: For each cell in Fig.2: Face Recognition System Architecture
the image, a histogram of LBP values is
computed. The histograms capture the Overview:
local texture information of the image. Admin Input: The Admin interacts with
Since LBP values are invariant to the system to input person face details.
monotonic. These details include images of faces, and
other relevant information. This associated with that individual from the
information is stored in the system. database.

Database: The database comprises two 6. Results

primary tables: Facial recognition is critical in
Person Face Details: Contains face data applications that require rapid
(images, feature vectors) for individuals. identification processes, such as security
Person Details: Stores additional details checks. An administrator grants
about individuals (name, ID, attributes). permission to access the system. After
logging in, the administrator's entered
User Interaction: When a person information is saved. The device then
approaches the system, such as for face starts face recognition by taking 200
recognition at an entrance, the system photos of the person. For training, these
takes the following actions: Captures the photos are processed with the Local
person's facial image using a camera. Binary Patterns Histogram (LBPH)
Analyzes the captured image to extract algorithm. This method streamlines the
distinctive features or create a set of process by removing the requirement for
these features. Compares the extracted significant image acquisition and training,
features to those stored in the database resulting in a more efficient identification
(Person Face Details). If a matching set of system. Facial recognition technology
features is identified, the system retrieves entails taking and storing photos for
personal information, such as name, training. When a trained individual is
contact number, address, and person observed, the algorithm compares the live
image from the Person Details table in the image to the stored photographs to
database. identify similar matches. If a match is
found, the individual can be uniquely
Face Recognition: This feature identifies identified. This unique information can
individuals by analyzing their facial then be utilized to present the
features. It employs the LBPH algorithm individual's information.
to match the input facial image with saved
face data. If a certain similarity threshold
7. Conclusion
is met, the system verifies the match as
successful.
Essentially, the face recognition system In the realm of face recognition, the Local
grants access to authorized users based Binary Patterns Histograms (LBPH)
on their facial recognition. The database method is utilized, showcasing an
securely stores face information provided impressive ability to reliably and
by the administrator. Upon successfully effectively identify distinct faces. It is
recognizing a user's face, the system extremely relevant and usable for real-
retrieves additional information world systems due to its ease of use and
flexibility in responding to changing 2. LBPH algorithm for Face
factors like illumination and facial Recognition,
reactions. By converting facial images [online] Available:
into a texture-specific representation https://iq.opengenus.org/lbph-
through Local Binary Patterns (LBP), the algorithm-for-face-recognition/
algorithm effectively works by taking off 3. Face Recognition: Understanding
the intricate texture details intrinsic to LBPH Algorithm,
individual faces. The segmentation of [online]Available:
these images into smaller regions, https://towardsdatascience.com/f
followed by the aggregation of LBP values ace-recognition-how-lbph-works-
into histograms, allows for the creation of 90ec258c3d6b
a detailed feature vector that uniquely 4. OpenCV Documentation,
identifies a facial image. Looking ahead, [online] Available:
the potential for the LBPH algorithm https://docs.opencv.org/index.ht
extends far beyond its current ml
applications, promising a future where 5. V. Aza, Indrabayu and I. S. Areni,
face recognition technology becomes "Face Recognition Using Local
even more integrated into our daily lives. Binary Pattern Histogram for
The adaptability and efficiency of the Visually Impaired People", 2019
LBPH framework open up possibilities for International Seminar on
its application in emerging technologies Application for Technology of
and new domains. One such area is in Information and Communication
augmented reality (AR) and virtual reality (iSemantic), pp. 241-245, 2019.
(VR), where LBPH can enhance user 6. [online] Available:
experiences by providing more https://en.wikipedia.org/wiki/Fac
personalized and interactive content e_detection
based on facial recognition. 7. [online] Available:
https://www.csie.ntu.edu.twl-
9. References fuh/personaliFaceDetectionandSm
ileDetection.pdf
1. O. S. Kulkarni, S. M. Deokar, A. K. 8. [online] Available:
Chaudhari, S. S. Patankar and J. V. https://link.springer.com/
Kulkarni, "Real Time Face chapter/10.1007/978-981-13-
Recognition Using LBP Features", 0311-1_39
2017 International Conference on 9. A. M. Jagtap, V. Kangale, K. Unune
Computing Communication and P. Gosavi, "A Study of LBPH
Control and Automation Eigenface Fisherface and Haar-like
(ICCUBEA), pp. 1-5. features for Face recognition using
OpenCV", 2019 International
Conference on Intelligent
 Sustainable Systems (ICISS), pp.
219-224, 2019.
10.A. Thakral and A. Vohra,
"Comparison between local binary
pattern histograms and principal
component analysis algorithm in
face recognition system", 2017
International Conference On Smart
Technologies For Smart Nation
(SmartTechCon), pp. 973-978,
2017.
11.V. Aza, Indrabayu and I.S. Areni,
"Face Recognition Using Local
Binary Pattern Histogram for
Visually Impaired People", 2019
International Seminar on
Application for Technology of In
formation and Communication
(iSemantic), pp. 241-245, 2019.