Search

We describe existing concepts for CMOS3-based Hartmann-Shack wavefront sensors and propose the next step for a further miniaturization of adaptive optics by integrating the modal wavefront reconstruction on chip. In conventional Hartmann-Shack wavefront sensing a CCD4 camera is placed on the focal plane of a microlens array. The spot pattern in the focal plane is captured, analyzed by image processing and the wavefront is often decomposed into orthogonal modes, usually expressed in coe.cients of Zernike polynomials. The single-chip modal wavefront reconstructor would include all these processing steps in a single chip. We propose to use already existing CMOS-based wavefront sensor concepts and add a hardware arti.cial neural network for modal wavefront reconstruction.

Wavefront technology was originally developed for the improvement of star images in optical astronomy. In 1978, this technology was for the first time applied to ophthalmology. At the University of Heidelberg, Germany, a closed loop adaptive optical system was designed to compensate for optical aberrations of the human eye. In the meantime, superresolution in retinal imaging has been achieved, providing detailed information on photoreceptors, small blood vessels, and nerve fiber structure in the human eye in vivo. Recently, new wavefront technology has been developed for the assessment of the human eye and it is now possible to apply a custom treatment to the eye, using the excimer laser. An adaptive optical closed loop system has been devised for preoperative simulation of refractive outcomes of aberration-free refractive surgical procedures. Treatment of a patient’s entire wavefront error has been demonstrated to improve the patient’s visual acuity beyond best spectacle-corrected vision.

The optical errors caused by the elements of the optical apparatus of the eye, as the cornea or the lens, strongly degrade the image on the retina. These errors are lager for larger pupil sizes.

In Ophthalmology the application of ultrashort laserpulses is already well established for special intraocular surgeries like posterior capsulotomy or periphere iridotomy (see e. g. [SP85]). Also the application of picosecond pulses for intrastromal ablation in refractive surgery is under investigation. Recently the process of plasma-induced or photodisruptive laser-tissue interaction is also discussed with respect to other medical applications, e. g. lithotripsy, angioplasty and the resection of brain tumours in neurosurgery [Fis+94]. As the pulse duration is much shorter than the thermal diffusion time, tissue ablation without thermal side-effects (carbonisation and coagulation zones) is possible. On the other hand, the expansion of the microplasma leads to the generation of a shock wave, a high pressure transient moving radially outword with an initial velocity exceeding the sound velocity. Histological examinations do not reveal any dammage of adjacent tissue, but with these methods possible effects on a subcellular level cannot be examined.

Es wurde eine objektunabhangige Rekonstruktionsmethode fur dreidimensionale lichtmikroskopische Bilder entwickelt. Durch inverse Filterung des Bildes mit der dreidimensionalen optischen Ubertragungsfunktion des Lichtmikroskops kann das Original weitgehend wieder hergestellt werden. Die Rekonstruktion von Testobjekten mit bekannter dreidimensionaler Struktur gewahrleistet die Zuverlassigkeit des Verfahrens.

In spite of the dramatic changes in feature size, die size and defect density, wafer inspection continues to be done manually. However, fast automatic wafer inspection will be a key element for efficient submicron VLSI-process technologies. A new two-step laser scanning procedure is described, resulting in processing speeds of up to 40 million pixels per second and spatial resolution down to 0.1 μm.

Die Korrelation stellt ein allgemeingultiges Konzept fur die Klassifizierung von Mustern dar. Mit analogen oder digitalen koharenten optischen Filterungsmethoden konnen Objektmuster erkannt werden. Im Falle des Spatial Filtering nach Van-der-Lugt [1] wird fur jedes Muster ein angepastes Filter benotigt. Daraus resultiert bei grosen Mustervielfalten fur die optischen Filterungsmethoden eine grose Filterzahl. Um diesen Nachteil zu vermeiden, wurden Methoden zur Erzeugung analoger und digitaler Komponentenfilter entwickelt, die einer Reihe von Objektmustern angepast sind. Dadurch wird der Merkmalraum in Unterraume mit vorgegebenem semantischen Inhalt aufgeteilt.