Customized iris feature extraction for intruder detection system to maximize detection rate using PCA algorithm in comparison with ICA algorithm

• Main Authors: Bhargav, T., Balachander, Bhuvaneswari
• Format: Conference Proceeding
• Language: English
• Subjects: Accuracy; Algorithms; Customization; Feature extraction; Feature recognition; Independent component analysis; Intrusion; Principal components analysis; Recall; Security systems
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0197600

The objective of this project is to acquire features of iris texture using Independent component analysis (ICA) for the purpose of novel intruder detection and comparison to Principal Component analysis (PCA). Methods: Customized extraction of iris features is employed to identify novel intruders in security systems. The sample size is determined with the G-Power tool being 0.8 for customizing the extraction of iris features in order to create novel intruder detection systems. Sample eye images from 50 individuals with 5 images are partitioned into two groups (Group 1=125 and Group 2 =125) and are split into a training set and a testing set. Conclusion: The accuracy rate of Principal Component Analysis (PCA) is 95.61 %, whereas the accuracy rate of Independent component analysis (ICA) is 90.57 %. The accuracy rate is 94.72 % for Principal Component Analysis (PCA), whereas the results of Independent component analysis (ICA) are 89.83%. The frequency of recall is 94.38% for Principal Component Analysis (PCA) and for Independent component analysis, the frequency of recall is 90.08 %. The critical value is p