Use feature detection and extraction from your reference frame to create kernels for 2D cross correlation on the remaining frame. Then decide on threshold, scoring logic to define a match. Start with a noise free, simple object frames to set thresholds, scoring numbers.

@robertbristow-johnson perhaps if the white noise were 'riding' on top of a fundamental that is aliased to near zero frequency in the sample band.

@robertbristow-johnson fair enough. I've edited and replaced harmonic with fundamental. I think in general sampled white noise (or rather wide band noise) just winds up as aliased noise across the sample bandwidth, but I'm curious if it might ever be possible where the aliased noise is more severly limited to a lower bandwidth over the sample band.

@robertbristow-johnson No. Please read more carefully - I mention the sampling of harmonics to set up the question of sampling white noise. As for white noise it has infinitely many harmonics, all of the same power.

So then in my example when I re-sample the 3 kHz sampled signal at 1 kHz the folding frequency (Nyquist) is pushed down to 500 Hz? If I do an FFT on the final signal would I expect to see two folding frequencies or just the final one (assuming there is wide band noise in the background that exposes folding)?