This is probably a whole other repository, but I had an idea for depth finding.
Because of the slope of the Piezo 's movement a phase correlation of the entire image against the structural wouldn't work well, maybe instead of finding the entire plane, we chunk it up into multiple horizontal lines (e.g. 10-20 rows at a time) and then run each row through a phase correlation of the Z-stack. That way we include enough information to hopefully find a strong correlation peak and then we can smooth/fit the location across the Y-axis (since it's a linear movement) to improve the accuracy if there are areas with low signal.
This is probably a whole other repository, but I had an idea for depth finding.
Because of the slope of the Piezo 's movement a phase correlation of the entire image against the structural wouldn't work well, maybe instead of finding the entire plane, we chunk it up into multiple horizontal lines (e.g. 10-20 rows at a time) and then run each row through a phase correlation of the Z-stack. That way we include enough information to hopefully find a strong correlation peak and then we can smooth/fit the location across the Y-axis (since it's a linear movement) to improve the accuracy if there are areas with low signal.