IMAGING: Diamond turning extends Cassegrain design to portable cameras

In lens design, it helps to set aside preconceived ideas. “In starting with a blank slate, sometimes the impossible can become possible,” said Joseph Ford, an associate professor at the Jacobs School of Engineering at the University of California San Diego. This appears to be the case with a new approach to camera design emerging from Ford’s laboratory at UCSD, in which, for example, a digital camera having the size and dimensions of the lens cap of a 35 mm single-lens-reflex (SLR) camera is producing imagery comparable to that produced by a full-size SLR lens and camera (with both mini and full-size cameras using identical CMOS sensors). The key enabler is diamond-turning technology that extends the Cassegrain refractive-telescope design to a handheld camera (see figure).^{1, 2}

More than three centuries old, the Cassegrain design reduces telescope length by collecting incident light around a central obscuration point and focuses it through internal reflection. “This was normally done once,” Ford said. “No one ever considered doing it eight times, because the tolerances were not workable. But diamond turning of calcium fluoride made this possible.” The researchers chose calcium fluoride because it produces a nice surface when diamond-turned, has low dispersion, and results in lenses that work across the visible band, Ford said. But for manufacturing systems in volume, it will be much more practical to work with molded-glass aspheres, or first-surface-reflection lenses that could be made of plastic or any number of materials.

The inducement for Ford and colleagues to think unconventionally came from Dennis Healy and Ravi Athale at the Defense Advance Research Projects Agency (DARPA; Arlington, VA), who launched a program dubbed Montage that, according to the DARPA Web site, “seeks to develop and demonstrate truly revolutionary imaging systems obtained by intelligent integration of the advancing capabilities of the individual optical, detection, and processing subsystems.”3

“They were looking for really thin cameras, with light-collection and resolution capabilities of regular cameras but ten times thinner,” Ford said. “They want to place them in unmanned aerial vehicles (UAVs) or on someone who needs to wear a camera without displaying a bulky box. DARPA was looking to meet a packaging constraint.”

Within DARPA’s Montage program, Ford’s research at UCSD is part of the work of Multi-Domain Optimization Team, headed by Mark Neifeld at the University of Arizona (Tucson, AZ). “The idea is to do optical-system design, but not just with a lens-design program,” Ford said. “You can use such a program to make a good lens, but the real problem has to do with generating a high-resolution picture at the image plane. What we do is incorporate the lens design and the post-detection processing, which enables us to explore less obvious solutions. So we optimize the technology and develop software that allows us to do that.”

In addition to Ford’s group at UCSD focusing primarily on physical optics, the Multi-Domain Optimization Team includes CDM Optics (Boulder, CO) working on wavefront coding and point-spread function (PSF) engineering; Yeshaiahu Fainman at UCSD looking at materials such as photonic-bandgap structures; and researchers at MIT exploring holographic approaches. Ford’s group also partnered with Distant Focus (Champaign, IL) to develop electronics and software.

Two approaches

Upon first fabricating a camera the size of a lens cap using a counterintuitive design approach, Ford said that everyone was in shock when it actually worked. “Then we showed it to the head guy at DARPA who said, ‘That’s cute, but there’s no depth of field.’ ” So they went back to the drawing board to address the depth-of-field issue from two different directions. One approach was PSF engineering in collaboration with CDM Optics.

The second approach involved changing the shape of the camera itself from a lens cap to something about the size and shape of a domino. Because poor depth of field can be caused by blurring from the circumference, they took a diamond saw and cut off portions of the lens they were not using (about 40%). This provided great depth of field, but less resolution and poorer light collection, Ford said. So metal mirrors were replaced with dielectric stacks for higher reflectivity, and the original detector was replaced with a more sensitive copper CMOS sensor. In the end, the even smaller domino-size camera outperformed the lens-cap version. Ford’s team is now working on a focusable camera with a better DOF and wider FOV. It consists of four aspheres, two on the front and two on the back.

In spite of its promise, the technology behind the new imaging method is not expected or intended to replace refractive lenses, Ford said. “It’s excellent for extremely compact packages and produces better results than you would expect possible, but it still involves design tradeoffs. So, I expect it will be very useful in specific applications (like cellphones and micro-UAVs) where bulk is really the limiting factor.”

REFERENCES

1. E. J. Tremblay et al., Applied Optics 46(4) 463 (Feb. 1, 2007).

2. E.J. Tremblay et al., OSA Topical Meeting on Computational Optical Sensing and Imaging, Charlotte NC, Paper (June 2005).

3. www.darpa.mil/mto/montage