PCE (Peak-to-Correlation Energy)
Definition
A refined statistic that divides the peak of the cross-correlation surface by the energy of the non-peak region. PCE is more sensitive than raw NCC for detecting weak PRNU signals, and it is the preferred decision statistic in most contemporary PRNU systems.
- Formula basis
- Peak of cross-correlation surface divided by energy of the non-peak region
- Advantage
- More sensitive than raw NCC for weak PRNU signals
- Status
- Preferred decision statistic in most contemporary PRNU systems
Common questions
Why is PCE preferred over raw normalized cross-correlation for PRNU matching?+
Raw NCC scores are sensitive to image content and can produce a misleadingly high peak from strong scene structure rather than sensor noise. Dividing by the surrounding correlation energy normalises out much of that content-driven variation, making the peak a more reliable indicator of a true sensor match.
What makes a PCE score unreliable in practice?+
Heavy compression, strong image resizing, or very low-texture scenes can weaken or distort the PRNU signal itself, which lowers the correlation peak regardless of the statistic used, so PCE improves discrimination but cannot recover a signal that has been degraded below detectability.
Related terms
- BM3D (Block-Matching 3D)
- A state-of-the-art denoising algorithm that groups similar image patches across the image, transforms them jointly, thresholds in the transform domain, and inverts....
- Hypothesis Testing in Camera Attribution
- The framework in which H0 is 'the query image was not taken by this camera' and H1 is 'it was'. The PCE...
- NCC (Normalised Cross-Correlation)
- The standard measure of similarity between a query noise residual and a reference PRNU pattern. It produces a value between -1 and...
- Noise Residual (W_hat)
- The per-image estimate of PRNU, computed by subtracting a denoised version of the image from the original. It contains PRNU plus random...
- Wavelet Denoiser
- A denoising filter that works in the wavelet transform domain, thresholding coefficients at each scale and direction to remove random noise while...