Machine vision-based high-precision and robust focus detection for femtosecond laser machining.

We propose a machine vision-based focus detection method (MVFD) for femtosecond laser machining. By analyzing the laser focus pattern, the defocus direction and distance are obtained simultaneously. The proposed technique presents high precision with an average error of 0.047 µm and a root mean square error (RMSE) of 0.055 µm. Moreover, the method is robust and is less affected by the tilted sample. For the curved surface sample, the average error and RMSE are 0.093 and 0.145 µm, respectively. Thus, the proposed focus detection method can be easily combined with laser processing equipment, which is widely used in large-range and high-precision femtosecond laser processing.