The peak detection for the autocorrelation of the ruler signal is a little fragile. The signal looks clear enough, but the right peaks aren't consistently detected. Needs more robustness
As a check, can look at the distribution of the line separations. They should be almost the same, any real variation probably indicates a problem
Can probably make use of the fact that the line separations are expected to be the same, and adjust the peak detection parameter to get the best result
The peak detection for the autocorrelation of the ruler signal is a little fragile. The signal looks clear enough, but the right peaks aren't consistently detected. Needs more robustness
As a check, can look at the distribution of the line separations. They should be almost the same, any real variation probably indicates a problem