Enhancing Single-Sample-Face-Recognition with Multi-Resolution Analysis Techniques

Abstract

Single Sample Face Recognition (SSFR) is a biometric technique that identifies individuals based on only one image of their face. Unlike traditional facial recognition systems that can use multiple images to enhance accuracy by capturing various angles, lighting, and expressions, SSFR relies solely on a single photo. This approach is particularly beneficial in scenarios where capturing multiple images is difficult or impractical, such as in immigration control, fugitive tracking, and video surveillance. The accuracy of SSFR is often impacted by factors like lighting conditions, pose variations, and facial expressions, making reliable identification more challenging. The paper proposes a solution to enhance the accuracy of SSFR systems through a multi resolution analysis method. This approach employs the discrete wavelet transform (DWT) to extract color-based binarized statistical image features (C BSIF) across different resolutions. By capturing more detailed information, the method aims to improve the system's capability to recognize faces under varying conditions. Experimental analyses conducted on the AR dataset demonstrate that this technique outperforms several state-of-the-art SSFR methods