Spatial Domain-based Face Recognition using MSB, and MLTP with ANN Classifier

Abstract

The human face images are an essential biometric trait in the authentication of individuals as the face images can be taken without the knowledge and physical contact of the person leading to several recent applications. We propose Spatial Domain-based Face Recognition using MSB and MLTP with ANN Classifier in this paper. The benchmarked face datasets are considered for testing the proposed method by converting color images into greyscale images and uniform image sizes. The 8-bit binary of each pixel is segmented into two groups viz., Most Significant Bits (MSB) and Least Significant Bits (LSB). The important information is available only in MSBs than LSBs hence MSBs are considered for feature extraction leading to compression of binary bits. The MSB four bits are converted to decimal values, and Histogram Equalization (HE) is used to enhance the contrast level of images and resize them. The Modified Local Ternary Pattern (MLTP) is applied to an image to extract compelling features. The Artificial Neural Network (ANN) classifies face images to recognize human beings. It is noticed that the proposed method's recognition results are better than the existing techniques. Keywords: Artificial Neural Network (ANN), Biometrics, Face Recognition, Histogram Equalization, MSB, MLTP.