Face Recognition Using Local Binary Pattern Histogram for Visually Impaired People

This study aims to build a real-time system that can help visually impaired to recognize people present in their surroundings. Visually impaired people experience many difficulties in daily activities, especially to recognize people. The proposed system applies the concept of a mobile personal assistant that runs in Android smartphone. The input data is a real-time video taken using 16-megapixel smartphone camera with a resolution of 1280×720. The system built uses Gamma Correction and Difference of Gaussian methods to improve the accuracy of the Local Binary Pattern Histogram (LBPH) algorithm. Based on the experiment with 148 cm and 172 cm user heights and 8 respondents, the system was able to increase the accuracy by 8.32% for 148 cm height, while the accuracy for 172 cm height increased by 6.22%. The output of this system is the sound of the identification result.