Face Recognition Under Age Progression Using Hybrid Features of Textures and Geometric

Facial recognition technology faces challenges in accurately identifying individuals across different age stages due to changes in facial texture and geometry over time. This study addresses these challenges by implementing a hybrid approach integrating texture and geometric features to improve recognition accuracy. Texture features are extracted using Local Binary Pattern Histogram (LBPH) and Histogram of Oriented Gradients (HOG), while facial landmarks capture geometric characteristics. Principal Component Analysis (PCA) is applied to enhance computational efficiency and reduce feature dimensionality. Unlike age-group classification, which simplifies the task by grouping individuals into fixed age categories with reduced intra-class variations, this research focuses on identifying the same person across different age stages, a more complex problem requiring the model to handle gradual facial changes over time. Experiments conducted on the FGNET dataset, which includes 82 subjects, demonstrate that combining HOG and facial landmarks with PCA yields the highest recognition performance, achieving an accuracy of 64%, precision of 69%, recall of 64%, and an F1-score of 63%. This hybrid approach demonstrates the potential of combining texture and geometric features to improve age-invariant facial recognition.