Wavefront coding finds increasing use
Engineers at CDM Optics (Boulder, CO) have demonstrated that their combined optical and electronic processing technique, known as wavefront coding, can dramatically cut an optical system's size...
Engineers at CDM Optics (Boulder, CO) have demonstrated that their combined optical and electronic processing technique, known as wavefront coding, can dramatically cut an optical system's size (see Laser Focus World, January 1999, p. 28; and December 1999, p. S26). Using the technique in the design of a conformal IR-imaging system resulted in a 50% reduction in physical components, a 45% reduction in weight, and a commensurately large reduction in cost. In addition, the company's founder and president, Thomas Cathey, has just been awarded a patent for a new high-speed confocal microscope. Because the system can acquire an entire slice of data at once without crosstalk, it may eventually allow time-resolved images of living organisms.
Wavefront coding attempts to match the optical design to the detector electronics and application. In it, unexploited resolution is traded for increased depth of field and a slight increase in noise. The resolution may be unused, for instance, because the pixel separation is significantly greater than the focal-spot size, and can be exchanged for either a deeper image, or for one with a more relaxed focusing tolerance (see figure). One application of this is in color imaging: a single lens can be designed that focuses light over a broad bandwidth—like white light—without chromatic dispersion. The technique works by designing optics that have a large—but uniformly distributed—point-spread function (PSF). Images produced by such optics appear blurry but, because the blur is uniform, it can easily be extracted by image processing: the PSF can be used as a kernel filter.
Wavefront coding can provide much greater positional tolerances in biometric imaging than traditional optical systems. A traditional system can provide a good image if the finger is placed correctly at best focus (upper left). When moved 9 cm beyond this position, however, the print becomes invisible (upper right). With the wavefront-coded system, the correctly positioned finger looks just as good (lower left). Crucially, the important information is still retained at 9 cm away (lower right).
Lenses designed to produce the wavefront to be decoded generally look very different from conventional optics. In the case of the conformal IR-imaging system, the imaging side is conventional but the detector side has a cosine-form surface—one with three teardrop-shaped peaks circled around, and pointing toward, the center.1 This design not only takes account of the imaging characteristics that are required, but also of the image-processing overhead. Because increasing the kernel-filter size rapidly increases the processing time, the size had to be limited to 10 × 10 pixels. The ability of wavefront coding to accommodate such constraints is one of the things that makes it so powerful.
Joseph Mait of the U.S. Army Research Laboratory (Adelphi, MD) and National Defense University (Washington, D.C.) has followed the technique. "Wavefront coding considers the physics of the optics together with the processing power of electronics," he says. It is optimizing the utility of the different components in this way that allows for smaller, less expensive solutions.
Another wavefront-coding project is under way at the National Cancer Institute (Bethesda, MD). Jim McNally, imaging director in the Lab of Receptor Biology and Gene Expression, oversees a light-microscopy facility. He is interested in the approach because it allows images to be collected from living cells faster than by any other means, doing it in a single exposure rather than having to scan through a number of focal planes. This, he says, is an advantage when trying to track rapid changes in cell structure.
The patent may eventually speed up other applications as well. Cathey's confocal microscope involves an array of light sources, each of which is focused on the specimen and is matched with its own pixel on the detector. To prevent crosstalk, each of the sources is modulated at a different frequency. At the detector, this light is combined with that of a reference signal at almost the same frequency, so that—together—they produce beats of a particular frequency. A filter is then used to reject all others. In this way, an entire slice can be acquired at once.
1. Kenneth Kubala et al., Optics Express 11(18), 2102 (Sept. 8, 2003).