Face Recognition Using LBPH, Haar Cascade Classifier and Back Propagation Neural Network

Impactful research in the area of face recognition has been conducted in the field of image processing. Since this has an application potential in many areas, extensive research is going on this area to increase the face recognition accuracy. We have proposed a framework with two approaches to recognize the countenances inside the image. In the first approach, we have used the modified LBPH with Haar cascade classifier. The improvement in this method is done by taking mean of the distance measures Euclidian, Manhattan and Canberra distance rather than using the traditional method of single distance measure. In the second approach, backpropagation classifier is used to train the face image dataset. The proposed face recognition methods were implemented in python. The results produced through this experiment thus show that the BPNN classification achieved a good recognition rate compared to the modified LBPH and Haar cascade classifier approach.