Eye-Landmarks Based Face Alignment In Non-Cooperative Face Biometrics
Journal Title: American journal of Engineering Research - Year 2018, Vol 7, Issue 10
Abstract
Current research in biometrics is being carried out with expectation of making a robust and reliable person identification system to be deployed in unconstrained and non-cooperative environment. However, these kinds of biometric systems predominantly suffer from various covariates (degradation factors). Among several covariates, misalignment is one of the critical issue that brings significant degradation in face recognition system. This is due to improper registration in intra-class samples as well as inter-class samples. In this paper we have proposed and analysed a face normalization technique to overcome the negative impacts of misalignment occurred during face detection. We used the database for experimentation that has a wellstructured deviation in pose angles of subjects along with unintentional random non-uniform illuminations which makes the dataset to be more suitable for evaluating algorithms more robustly. The proposed method is based on locating the landmarks of face and then cropping the face region proportional to the parameter obtained from estimated landmarks. This simple way of face normalization has been tested and validated on original face samples used in UBIPosePr dataset. Thus, the objectives of this paper are three-fold. 1) We have proposed the method of normalizing the face overcoming the scale of face that appears in image sample. 2) We experimented with actual face detection algorithms and found our normalization algorithm mitigates the misalignment problem. 3) A very robust and yet well-structured dataset has been used in our experimentation in order to study the behaviours of non-cooperative environment for face recognition system.
Authors and Affiliations
Nilesh Pawar
Keywords
Related Articles
- EP ID EP398566
- DOI -
- Views 60
- Downloads 0
Demonstration of the Lorenz Force in the Study of Physics
the article considersexperiments that demonstrate the Lorenz force in the process of studying physics.Describes in details thedemonstration device which shows rotation of an electrolyte with a flowing current when the el...
Study of Various Rainfall Estimation & Prediction Techniques Using Data Mining
It is important to estimate accurately rainfall for effective use of water resources and optimal planning of water structures and availability. For this purpose, the various models and techniques are developed to estimat...
Web Based Genetic Algorithm Using Data Mining
This paper presents an approach for classifying students in order to predict their final grade based on features extracted from logged data in an education web-based system. A combination of multiple classifiers leads to...
Flexural Behaviour of High Performance Reinforced Concrete Beams Using Industrial Byproducts
In this study experimental investigations were carried out to explore the possibility of using Fly Ash, silica fume and metakaol in as a replacement of cement and bottom ash as a replacement of sand to make high performa...
Solution of Advection-Diffusion Equation for Concentration of Pollution and Dissolved Oxygen in the River Water by Elzaki Transform
In this article, a new integral transform called as Elzaki transform is used to solve the model consists of a couple of advection-diffusion equations. These advection-diffusion equations have much importance in chemistry...