Wavefront coding keeps a focus on applications

03 Oct 2003

Wavefront coding can generate crisp images of scenes that would otherwise appear blurred. Rob van den Berg reports on how the technology is benefiting everything from microscopes and endoscopes to machine-vision systems and cameras in mobile phones.

From Opto & Laser Europe October 2003

Sometimes the best solutions to tricky problems can be counterintuitive, and that's definitely the case with an emerging imaging technology now being commercialized by CDM-Optics, a small company based in Boulder, Colorado, US.

Tradition dictates that the usual way to increase the depth-of-field (the distance over which objects remain in focus) of an imaging system is to "stop down" the aperture by making it smaller. However, the drawback of this approach is that it also dramatically reduces the amount of light that can be gathered.

In the early 1990s, radar engineer Edward Dowski and his thesis adviser Thomas Cathey at the University of Colorado, US, came up with an alternative technique for obtaining sharp images over a long distance. Their unorthodox approach was to deliberately "blur" the image - not by moving the focusing knob, but by passing the light through a waveplate - and then reconstruct the original image with the help of a computer. This method, dubbed wavefront coding, turned out to increase the depth-of-field by a factor of 10 compared with conventional lenses.

Commercialization When Dowski finished his PhD in 1996 the University of Colorado did not show any interest in marketing wavefront coding, so together with a former founder of Ball Aerospace Merc Mercure, Dowski and Cathey started their own company, CDM-Optics. Since then their technology has found its way into microscopes, endoscopes, biometrics devices and machine-vision systems, and is also being explored for use in the cameras found inside the latest mobile phones.

Sitting in his new office - the company has just moved into larger premises in Boulder - Cathey explained how the technique works: "With a simple camera lens only the light rays coming from objects within the depth-of-field will end up focused in the image plane. All others will be focused either in front of or behind [it], and depending on the object distance will give a more or less unsharp image. We came up with the idea to deliberately distort the light rays by passing them through a waveplate with a saddle-like shape: relatively flat in the middle, but with scalloped edges. This causes a specific optical aberration: the image looks blurry, but the de-focus is the same over a large range of distances."

In this way the light is "encoded" when it passes through the lens system: instead of converging to a point on the focal plane, the rays coming from each imaged object spread out and form a broad light-beam in which they are roughly parallel to one another. These rays are then used to create a blurred image which is "decoded" by a computer filter.

"Since the distortions in the image are mainly determined by the shape of the de-focusing optical element, which is accurately known, a computer is able to remove the blur point by point," explained Cathey. "The computer decodes the image using what is essentially a digital filter, and thus creates an image which is sharp over a large depth-of-field."

An increased depth-of-field has benefits for industrial imaging applications, especially machine vision. In August 1999, CDM-Optics licensed its wavefront-coding technology to Adaptive Optics Associates (AOA) of Cambridge, Massachusetts, for use in the firm's CCD cameras. With an extended depth-of-field of more than 36 inches and resolution of 250 dpi, these cameras are ideal for parcel-label imaging and 1D and 2D bar code reading.

"There are several great uses for this technology in the automatic data-collection market-place," commented Jeff Yorsz, assistant general manager of AOA. "Parcels come in all different shapes and sizes, and now we can read labels and bar codes on curved and slanted surfaces."

In the last three or four years the scope of wavefront-coding applications has widened extensively. In 2001, optical systems giant Carl Zeiss of Oberkochen, Germany, initiated plans to use wavefront-coded modules in its optical microscopes to extend the depth-of-field. A year later, Olympus Optical followed suit for its endoscopes.

Biometrics applications, such as iris or fingerprint recognition, are another area in which wavefront coded optics are starting to make a difference. The technology means that it is now possible to capture images of the eye's iris at greater distances than ever before.

Cathey admits, however, that the method has its trade-offs: "The digital filtering operation introduces noise that roughens the image. However, nonlinear signal processing can reduce that noise," he said.

Aberration killer The technique can also be used for much more than simply increasing the depth-of-field of an imaging system. According to CDM-Optics, wavefront coding can correct for all kinds of focus-related optical aberrations such as astigmatism, spherical aberration, curvature of field or chromatic aberration.

"In a normal lens the red, green and blue light rays all have a slightly different focus," Cathey explained. "You can correct for this by using combinations of lenses, but we can do it much more easily by wavefront coding because it becomes irrelevant whether the different colours end up at different distances from the lens."

The technique also corrects for the blur caused by the temperature-related aberrations associated with inexpensive plastic optics. Optical elements can be manufactured in plastic providing a lower-cost, high-volume solution.

The waveplate does not even have to be a separate surface, but can be made as an integral part of the lens: it is easy to adjust the mould for a plastic lens to include the shape of the wavefront-coded optics.

CDM-Optics is currently working with mobile phone manufacturers to make a prototype for a phone with an integrated camera using wavefront optics. Cathey says that while such devices usually contain two or three lenses to correct for field curvature and chromatic aberration, just one lens would be required using wavefront optics. "The signal processing capacity that is required for this is available, as many of these cameras have digital image processing capacity anyway," he explained.