Applying Filters to Identify a Person Based on a Face Image

Abstract

Today, there are many different types of systems that require a user identification procedure. Most often, identification in such systems occurs by entering a login and password, presenting a magnetic card, token, electronic pass and other identifying documents, which, as you know, can be easily lost, forgotten or forged. To solve this problem, biometric identification systems were invented, which are based on the unique biological characteristics of a person, which are difficult to fake and which uniquely identify a specific person. This article is devoted to the study of the capabilities of the Wiener filter to improve the accuracy of biometric face identification. The object of research is the recognition algorithm HOG or “Histogram of oriented gradients” using the dlib, opencv libraries. The Wiener filter is applied to restore low-quality images. The process of development of two software modules is described - face recognition by photo and application of the Wiener filter for distorted images. Python and C ++ were chosen as programming languages. Images are both regular photos and stills from a video file. Two types of image distortion are considered - Gaussian blur and Motion blur. For these distortions, experiments were carried out to measure the Euclidean distance in order to find out the effectiveness of the filter.