Improving Human Face Recognition using Deep Learning based Image Registration and Multi-Classifier Approaches

Abstract

Face detection, registration, and recognition have become a fascinating field for researchers. The motivation behind the enormous interest in the topic is the need to improve the accuracy of many real-time applications. Countless methodologies have been acknowledged and presented in the past years. The complexity of the human face visual and the significant changes based on different effects make it more challenging to design as well as implementing a powerful computational system for object recognition in addition to human face recognition. Using supervised learning often requires extensive training for the computer which results in high execution times. It is an essential step in the face recognition to apply strong preprocessing approaches such as face registration to achieve a high recognition accuracy rate. Although there are exist approaches do both detection and recognition, we believe the absence of a complete end-to-end system capable of performing recognition from an arbitrary scene is in large part due to the difficulty in alignment. Often, the face registration is ignored, with the assumption that the detector will perform a rough alignment, leading to suboptimal recognition v performance. In this research, we presented an enhanced approach to improve human face recognition using a back-propagation neural network (BPNN) and features extraction based on the correlation between the training images. A key contribution of this paper is the generation of a new set called the T-Dataset from the original training data set, which is used to train the BPNN. We generated the T-Dataset using the correlation between the training images without using a common technique of image density. The correlated T-Dataset provides a high distinction layer between the training images, which helps the BPNN to converge faster and achieve better accuracy. Data and features reduction is essential in the face recognition process, and researchers have recently focused on the modern neural network. Therefore, we used using a classical conventional Principal Component Analysis (PCA) and Local Binary Patterns (LBP) to prove that there is a potential improvement even using traditional methods. We applied five distance measurement algorithms and then combined them to obtain the T-Dataset, which we fed into the BPNN. We achieved higher face recognition accuracy with less computational cost compared with the current approach by using reduced image features. We test the proposed framework on two small data sets, the YALE and AT&T data sets, as the ground truth. We achieved tremendous accuracy. Furthermore, we evaluate our method on one of the state-of-the-art benchmark data sets, Labeled Faces in the Wild (LFW), where we produce a competitive face recognition performance. In addition, we presented an enhanced framework to improve the face registration using deep learning model. We used deep architectures such as VGG16 and VGG19 to train our method. We trained our model to learn the transformation parameters (Rotation, vi scaling, and shift...

An improved subpixel-level registration method for image-based fault diagnosis of train bodies using SURF features

“…Image registration technology has been widely used in multimode registration [37], unmanned aerial vehicle (UAV) image registration [38], face image registration [39], medical image registration [40], satellite image registration [41], [42] and so on. In recent years, with the rapid development of transportation technologies [43][44][45], supervised learning and unsupervised learning fault detection schemes based on signal processing have continuously been reported [46][47][48][49].…”

A Case Study of Facial Emotion Classification Using Affdex

“…In terms of computational power is the reduction in data and function in the face recognition process essential and the researchers have recently been focused on the use of modern neural networks for automated analysis [67].…”

Deep Learning Architecture Based on Convolutional Neural Network (Cnn) in Image Classification

“…However, the use of this method has several disadvantages, including limitations in extracting complex features and non-linear relationships between these features. This is what drives the emergence of CNN as a more effective solution [1] [2].…”