Face recognition system using Viola Jones, histograms of oriented gradients and multi-class support vector machine

Abstract

This study aims to determine the best distance to install a CCTV camera in identifying a person's face in the passenger inspection area at the airport, which can be developed for suspect detection systems. The training data used is in the form of an image with five different angles per person, while the testing data is in the form of video. The initial stage conducted is face detection based on local features (a pair of eyes, nose, and mouth) in the input data using the Viola-Jones method. The face detection results process with pre-processing, feature extraction, and classification. In the pre-processing stage, the Brightness Enhancement (BE), Grayscale and Contrast Limited Adaptive Histogram Equalization (CLAHE) methods are used to improve the quality of the detected face. Furthermore, the feature extraction and classification stages use the Histogram of Oriented Gradient (HOG) and the Multi-class Support Vector Machine (MSVM) methods, respectively. The result shows that the best accuracy obtained is 86.76% for a CCTV camera distance of 300 cm and a height of 250 cm.