Enhancement of alignment-technologies for lenses used in high-end optics

We present a way to enhance state of the art alignment-technologies using digital image processing. This work is motivated by the continuously increasing demand for high-end optics. Today the image quality is determined by the manufacturing tolerances which has to be below 1 micron. Some of the aberrations are directly linked to the decenter of the lenses, Coma for example. The current system uses a half covered photo-detector, generating a sinusoidal signal, which is evaluated by a lock-in-amplifier to drive two hammers to adjust the alignment chuck. Some of the typical lenses produce internal reflections, which cause problems for the current system. Digital image processing allows us to fix these with image recognition. When possible our algorithm use a threshold method to convert the pictures into a binary format, which is evaluated to determine the position and size of the spot. If a simple conversion fails, a gradient technique is used for edge-detection to eliminate interfering reflexes, stray light, etc. The resulting images are fitted to a circle, to determine the radius as a measure for the decentricity. With this new technique lenses can be centered within 0.25 microns.