When using a femtosecond laser to machine a single-crystal silicon wafer, it is accompanied with a diffraction spot of plasma. The existing literature reports that the brightness of the image of plasma can be used as an indicator to online measure the depth of the machined groove on a micrometer scale. Because the plasma spot is influenced by eruption and partial occlusion of ablated material, this method, which simply relies on the spot image brightness as a feedback parameter, is not reliable or accurate. The pixel area, perimeter, and brightness characteristics of the plasma spot image need to be comprehensively analyzed to provide a reliable and accurate feedback to establish close-loop micromachining technology. Therefore, we first analyze the chirped amplification principle of generating a femtosecond laser and the application of the diffraction spot of plasma during the micromachining processing using the femtosecond laser. Second, we experiment using femtosecond laser ablation with a piece of
and thickness of
single-crystal silicon wafer to obtain the corresponding relational data among parameters of laser processing power, processing speed, and laser spot image of plasma. Third, aiming at the characteristic of dim target of the laser spot image, the two-dimensional Otsu (maximum class square error method) is used to segment the laser spot image to improve the segmentation accuracy of the laser spot image. Finally, we analyze the relationship among area, perimeter of the laser spot image, and laser energy; the relationship among area, perimeter of the laser spot image, and the machined depth of groove; the relationship between brightness of the laser spot image and laser output power; and the relationship between brightness of laser spot image and machining speed.