Power Signature (ENF)
We study forensic techniques of exploiting an environmental signature, the electric network frequency (ENF) signal, which can serve as a natural time stamp for multimedia recordings and can thus be used for data authentication and tampering detection.
Analysis of ENF signal extraction from videos acquired by rolling shutters
​Jisoo Choi, Chau-Wai Wong, Hui Su, and Min Wu, "Analysis of ENF signal extraction from videos acquired by rolling shutters," submitted to IEEE Transactions on Information Forensics and Security (T-IFS), under review.
Electric network frequency (ENF) analysis is a promising forensic technique for authenticating multimedia recordings and detecting tampering. The validity of the ENF analysis heavily relies on the capability of extracting high-quality ENF signals from multimedia recordings.
​​
The paper above analyzes and compares two representative methods, the direct concatenation and periodic zeroing-out methods, for extracting ENF signals from visual signals acquired by cameras using the rolling-shutter mechanism, which facilitates the fundamental understanding of extracting ENF signals from videos.
​​​
Invisible geolocation signature extraction from a single image
​J. Choi, C.-W. Wong, A. Hajj-Ahmad, M. Wu, and Y. Ren, "Invisible geolocation signature extraction from a single image," IEEE Transactions on Information Forensics and Security, pp. 2598–2613, Jun. 2022.
This paper introduces a method for extracting ENF traces from a single image captured by cameras with rolling shutters for region-of-capturing localization. Compared to the recent art of extracting ENF traces from audio and video recordings, it is very challenging to extract an ENF trace from a single image.
We address this challenge by first mathematically examining the impact of the ENF embedding steps such as electricity to light conversion, scene geometry dilution of radiation, and image sensing. We then incorporate the verified parametric models of the physical embedding process into our proposed entropy minimization method.
​
Fig. 1 shows an illustration of the ENF embedding process. The input signal, the sinusoidal voltage signal supplied to indoor light sources, goes through the following steps: the power to illumination conversion, scene geometry dilution of radiation, and photons accumulation. The final output signal of the embedding procedure, the pixel intensity signal, is not an ideal sinusoid, but a sinusoidal-like signal with some trend. Based on this observation, we propose a more precise parametric model for image-embedded ENF traces, use it in our entropy minimization for the ENF estimation, and boost the ENF estimation accuracy.
