Abstract
The electric network frequency (ENF), which represents the supply frequency of an electric power grid, fluctuates around its nominal value (50/60 HZ) due to imbalances between power demand and supply. Variations in intensity and spectrum of artificial lighting, responding to grid alternating current, enable the estimation of ENF signal from videos recorded under such lighting. Over the decade, researchers have developed methods to successfully extract the ENF signal from videos recording, leveraging the sensor characteristics of recording devices. However, different lighting types including - incandescent, halogen, fluorescent and light emitting diodes exhibit distinct mechanisms of light generation and vary in response to voltage fluctuations, posing challenges to reliable ENF capture across all light sources. This study systematically examines the impact of various lighting types and recording conditions (e.g., still scenes, dynamic scenes, and moving objects) on the accuracy and feasibility of ENF extraction. Findings highlight the differential influence of lighting types in ENF signal reliability under varied recording conditions, providing insights into optimal setups for accurate ENF-based analyses in practical applications.