Enhanced Local Face Feature Encoding via the Petersen Graph Framework

This study proposes the Weber Petersen Graphic Local Face Feature Pattern (WPLFP), a novel local feature extraction method. WPLFP integrates Weber's Law, utilizing the Weber-Petersen number to encode structural relationships in a 5x5 window. It captures complex inter-pixel patterns through Petersen graph permutations and enriches features with LBP and CS-LBP. A multi-scale model combines vertical (WPLFPv) and horizontal(WPLFPh) histograms. Experiments show that WPLFP outperforms state-of-the-art methods, validating its effectiveness in facial feature extraction and recognition. This underscores the benefits of interdisciplinary research and innovative thinking in addressing complex pattern recognition challenges.