Improved facial recognition based authentication approach to secure big data

Abstract

When deploying biometric identification techniques over the massive data available on web for user authentication purposes, maintaining quality, security and integrity of confidential data are imperative. It is required to make sure the data is captured and stored over a trusted server and is readily available for authentication/ user identification without any interference. In this paper, facial recognition is used as a measure of biometric authentication to address the security issues in Big data. Discrete Wavelet Transform (DWT) is applied to normalize and de-noise the input image, in order to eradicate the unwanted variations preserved while storing the biometric data using traditional methods such as Principle Component Analysis (PCA). Following this, Gabor Filter bank is used to extract the facial features. Further, Expansive Discrete wavelet Transform (EDWT) is used to linearize the dimensional sub-space, using its high expansiveness to curb the number of features extracted from the facial data. The approach uses the spatial orientation of the processed image’s high-frequency textural features to improve the accuracy of the trained data for overcoming the shortcomings which results in a 74% efficient algorithm which viably and feasibly achieves the objective of minimizing the expanse of features extracted.