Facial Authentication System by Combining of Feature Extraction using Raspberry Pi

Abstract

This work provides the mixture of two main component analysis (PCA) and linear discrimination analysis (LDA) feature extraction using Euclidean distance classifications for upgrade the efficiency on face-based real-time authentication method. Combining the extraction of functions that measure LDA and PCA values so that the matching scores become acquired. The matching scores represent sums extraction values from both functions. Distinctive values from face images could be improved also by combined extractor. Measure the distance to find a shortest distance between some of the template images by using Euclidean distance for test images. The output of a system authenticates for personal identity dependent from each user's face image. Results are obtained from the 11-user image indicate that perhaps the combined extractor offers a greatest result compared to the performance of the single extraction feature. The current technique provides greater average performance than just a single extractor, however on the facial authentication device may overcome low performance. The real-time system will be more compatible and flexible with existing service conditions, so improved results will be needed.