Facial Recognition Performance Based on the Lighting Set-Up Models Applied to Home Security Door Access using Principal Component Analysis and Raspberry Pi Controller

Security protects individuals, data, and properties together with its corresponding measures in line to the emerging application of technology like the face recognition. This study aims to test the effects of lighting models on the recognition performance along with different angles and distance applied for door access by providing signals to Raspberry PI Controller. The study built 240 training datasets and applied the best algorithms – Haar-Cascade for face detection, Principal Component Analysis for extraction, Support Vector Machine for classification and Euclidean Distance for recognition. The study tested the model with five subjects which then marked a score of 100% for fluorescent light, 74.65 % for candle light, and 88 % for flashlight with an average percentage accuracy of 89.5. The results implied an adverse recognition result influenced by the lighting conditions, the face angle positions and distance issues.