An approach to enhance performance of age invariant face recognition

Abstract

Age invariant face recognition (AIFR) is currently a study topic with several potential uses. It offers a variety of real-world applications, including passport renewal, driver’s license renewal, locating missing children, locating criminals, providing security to VIPs etc. In the field of AIFR, scientific efforts have increased. Matching faces of big age differences is thus a challenge, owing to the significant variation in appearance between young and elderly age. The appearance and form of the face deteriorate with age, making facial recognition the most difficult task. AIFR has become a highly common and difficult chore in recent years. In this discipline, the set of feature extraction and classification algorithms is crucial. This paper addresses the issues raised above by proposing an enhanced ASM approach for extracting features from 2D search regions using handcrafted and deep image features in conjunction with a 7-layer CNN architecture and a smaller image size of 32x32 pixels to reduce delay time and space complexity. Using the standard dataset LAG, the study approach entails running many tests to evaluate the proposed system’s performance. The results show that the suggested method beats state-of-the-art algorithms in face recognition and achieves good accuracy throughout the age spectrum. The presented methodology achieves a maximum accuracy of 91.76 percent for the LAG database, outperforming all existing state-of-the-art methodologies.