Tezislər / Theses

particular field based on computer-human interaction is face detection. Face 
recognition is a biometrics pattern authentication technique which is used in 
a wide variety of computer vision applications. It plays a major role in video 
surveillance, security, digital video processing, content-based retrieval, etc. 
The objective of face detection is to detect any faces in an arbitrary image 
and return the image location and extent of each face. There are many 
factors that make the real-time face detection a challenging task. One of the 
main problems is the time and accuracy which determine the performance of 
face recognition system in real-time environments. Additionally, variations in 
Image 
preporcessin
g
Applying 
Principial 
Component 
Analysis
Adaptive 
Histoqram 
Equalization 
Threshold 
Level 
Determinatio
n
Converting 
the 
Subtracted 
Image to the 
Binary Image 
Colorizing 
the final 
result

---

THE 3
rd
 INTERNATIONAL SCIENTIFIC CONFERENCES OF STUDENTS AND YOUNG RESEARCHERS 
dedicated to the 99
th
anniversary of the National Leader of Azerbaijan Heydar Aliyev
180
scale, location, view point, illumination, occlusions can affect the overall 
accuracy of system.
There are many methods and algorithms available for face detection, 
such as Viola-Jones, SMQT features &SNOW Classifier, Neural Network-
Based Face Detection and Support Vector Machine-Based face detection. 
Although all these algorithms manage to detect faces from an image, their 
precision and recall values are different. This paper focuses on Viola and 
Jones face detection algorithm which has the high image processing speed 
and detection rates. 
Viola-Jones algorithm is implemented in OpenCV and combines four 
main concepts: 
1.
Haar like features 
Every human face has certain common properties. For example, eye 
region is darker than nose bridge region or upper cheek region is brighter 
than eye region, etc. Haar like features detect the difference in the black and 
light portion of the image and this computation forms a single rectangle 
around the face. There are some commonly used Haar features, like two 
rectangle feature, three rectangle feature, etc. 
2.
Integral image 
Integral image is an intermediate representation for the image and 
allows to compute the sum of values in a rectangle subset of a pixel grid in a 
very efficient and quick way. The integral image at location x, y contains the 
sum of the pixels above and to the left of x, y, inclusive: 
where i(x, y) is the pixel value of the original image and ii(x’,y’) is the 
corresponding image integral value. 
Figure 1.
Illustration of the integral image and 6 types of Haar-like rectangle features 

---