Human Face Detection and Recognition Protection System Based on Machine Learning Algorithms with Proposed AR Technology
Md. Imran Hossain Chowdhury,Newaz Mahmud Sakib,S. M. Masum Ahmed,Mohammad Zeyad,Md. Abul Ala Walid,Golam Kawcher
DOI: https://doi.org/10.1007/978-981-16-7220-0_11
2022-01-01
Abstract:This paper demonstrates the difference between two algorithms, the first one is “Template Matching” and another one is “Local Binary Pattern Histogram (LBPH).” The face identification and recognition security system prototype had implemented using the LBPH algorithm, Python, Raspberry Pi 3 Model B+ , and OpenCV technology. This concept introduces a system for identifying random faces with Haar classifier. This method does not search for individual matches like a biometric device, it compares people to all the same, instead provide matches based on first, second, and third findings in a static collection without having to deal with the databases. Similar to a CCTV, it can identify the persons; however, instead of storing a large amount of material, it contains just a small amount. However, the face detection rate of 90% had achieved when the LBPH method had used in bright lighting conditions. On the other hand, when the template matching method had used in this recognition method in bright light condition then the detection rate was 40%. A method has been proposed that would enable the successful recognition, visualization, and detection of the convicted person utilizing virtual platforms and 3D modeling of stored pictures, which might be innovative and used in the development of augmented reality (AR). Also, the detection performance of this device had analyzed that could detect the user’s face up to 15 m in proper lighting conditions without any hassle. This may be used for home and commercial surveillance, for identification, or in the event of an act of bank robbery, or for counterterrorism. Finally, the device had implemented with the LBPH algorithm and quite economical compare to another biometric security system.