Article Per Year (5 Year)
p-Index From 2021 - 2026
0.408
P-Index
This Author published in this journals
All Journal International Journal of Robotics and Control Systems
Mahsa Madadi Masouleh
University of Victoria
Control & Systems Engineering Electrical & Electronics Engineering

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search
Journal : international journal of robotics and control systems

 1 
Improving the Recognition Percentage of the Identity Check System by Applying the SVM Method on the Face Image Using Special Faces Azita Mousavi; Amir Hossein Sadeghi; Ali Mojarrad Ghahfarokhi; Fatemehalsadat Beheshtinejad; Mahsa Madadi Masouleh
International Journal of Robotics and Control Systems Vol 3, No 2 (2023)
Publisher : Association for Scientific Computing Electronics and Engineering (ASCEE)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31763/ijrcs.v3i2.939

Abstract

Face recognition has attracted tremendous attention during the last three decades because it is considered a simple pattern recognition and image analysis method. Also, many facial recognition patterns have been introduced and used over the years. The SVM algorithm has been one of the successful models in this field. In this article, we have introduced the special faces first. In the following, we have fully explained the SVM method and its subsets, including linear and non-linear support vector machines. Suggestions for improving the recognition percentage of a person's identity check system by applying the SVM method on the face image using special faces are presented. For this test, 10 face images of 40 people (400 face images in total) have been selected from the ORL database. In this way, by choosing the optimal parameter C, determining the most suitable training samples, comparing more accurately with training images and using the distance with the closest training sample instead of the average distance, the proposed method has been implemented and tested on the famous ORL database. The obtained results are FAR=0.23% and FRR=0.48%, which shows the very high accuracy of the operation following the application of the above suggestions.
Comparison of Feature Extraction with PCA and LTP Methods and Investigating the Effect of Dimensionality Reduction in the Bat Algorithm for Face Recognition Azita Mousavi; Hadis Arefanjazi; Mona Sadeghi; Ali Mojarrad Ghahfarokhi; Fatemehalsadat Beheshtinejad; Mahsa Madadi Masouleh
International Journal of Robotics and Control Systems Vol 3, No 3 (2023)
Publisher : Association for Scientific Computing Electronics and Engineering (ASCEE)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31763/ijrcs.v3i3.1057

Abstract

Face recognition is one of the challenging subjects of image processing. Facial recognition is often a biometric method that basically uses faces to recognize people. The face recognition system consists of three main steps: finding the face in the image, feature extraction and classification. The face recognition system faces challenges such as changes in lighting, changes in age, changes in facial expressions, etc. One of the important issues in this system is the algorithm execution speed. For this purpose, the dimensions of the feature vectors should be small enough, especially when the database is large. Since the face recognition system must be performed on a wide range of databases, dimensionality reduction techniques are required to reduce time and increase accuracy. Dimension reduction methods are used for this purpose. Two methods of dimensionality reduction, including LTP and PCA, are given in this research. In this research, first, the LTP feature vectors are extracted from the face image, and then the effective features are selected using the Bat algorithm. Therefore, this algorithm has three main phases of feature extraction, feature selection and classification. This algorithm is implemented on the ORL database, which contains 400 images of 40 different people with a size of 112×92 pixels. In addition to reducing the time required for testing, the proposed method has provided a very good accuracy of 99%.
Co-Authors Ali Mojarrad Ghahfarokhi Amir Hossein Sadeghi Azita Mousavi Fatemehalsadat Beheshtinejad Hadis Arefanjazi Mona Sadeghi