Authentication and Access Control by Face Recognition and Intrusion Detection System

The main worry with the rapid growth of technology has been cyber assaults. To counter these threats, sophisticated security systems have been-created, however none of them function completely error-free. This study uses face detection and recognition by Haar cascade classifier and LBPH for authentication initially, and then an intrusion detection system (IDS) using machine learning algorithm like FNT and KNN can identify fraudulent behavior. The typical accuracy for face detection is 90.2%. Whereas in recognition, it can be demonstrated that LBPH performs better in both still images and video than Eigen faces with respect to detection accuracy and execution speed. With a false positive rate of 1.6%, known and unknown intrusions accuracy detected by FNT is 97.2%. The detection rates for DOS, probe, U2R, and R2L in the known intrusion classifier by KNN are 98.7%, 97.4%, 97.8%, and 96.6%, respectively, whereas the false positive rates are 0.4%, 0.0.1.45%, 2.19%, and 1.97% respectively. The proposed known intrusion mechanism is demonstrated to outperform competing methods. The percentage of intrusion detection in the unknown intrusion detected by C-means clustering is 98.6%, and the rate of false positives is 1.32%.